AI Exposure Leaderboard
Every tracked stock ranked by its AI Exposure Score: our editorial 0-100 rating of how much the business is tied to the AI buildout. Not a financial metric.
Market data as of the close on . Prices are end of day, not live.
511 stocks ยท showing 1 to 50
Cerebras sells wafer-scale AI processors, CS-3 systems, and inference cloud services for training and running large models. The company exists solely to serve AI compute demand, making it a pure play on the buildout.
Sells the GPUs, networking, and CUDA software stack that nearly every AI training and inference cluster is built on, mostly to hyperscalers and AI labs. Data center AI demand is now the overwhelming majority of revenue and the entire thesis.
Nebius operates an NVIDIA-backed GPU cloud selling AI compute and infrastructure services, with Microsoft and Meta on its customer book. The company exists to monetize AI training and inference demand, making it a pure play on the buildout.
CoreWeave rents Nvidia GPU clusters to AI labs and enterprises for training and inference, with contracted capacity from customers like Microsoft and OpenAI. The company exists solely because of AI compute demand, making it a pure play on the buildout.
Aurora sells autonomous trucking capability, commercializing the Aurora Driver on freight routes with partners like PACCAR and Volvo. The company has essentially no non-AI business, so its value rides entirely on driverless trucking reaching scale.
Kodiak sells autonomous driving technology for long-haul trucking and unmanned defense vehicles, licensing its Kodiak Driver system to fleets and military customers. The company has no business outside AI autonomy, making it a pure play on self-driving commercialization.
Applied Digital builds and leases high density data centers and GPU cloud capacity purpose built for AI training workloads, selling to hyperscalers and AI companies. The company exists because of AI compute demand, making it a pure play on the data center buildout.
Mobileye sells EyeQ vision chips and SuperVision ADAS systems to automakers, monetizing AI perception in nearly every car it touches. The company is effectively a pure play on autonomous driving adoption, with value scaling on ADAS penetration.
Pony AI's entire business is an AI virtual driver monetized through robotaxi rides and robotruck freight in China. The thesis depends on regulatory approvals, safety records, and fleet economics, with no non-AI revenue to fall back on.
Sells voice AI software to automakers, restaurants, and enterprises for in-car assistants, phone ordering, and customer service. It is a pure play whose entire business depends on conversational AI adoption.
Every WeRide product is an autonomous vehicle platform running on the same AI stack, spanning robotaxis, robobuses, robosweepers, and robovans. The thesis depends on multi market commercial rollout, with the Uber partnership as the key international distribution channel.
Serve Robotics builds autonomous sidewalk delivery robots, earning fees per delivery through partners like Uber Eats and 7-Eleven. The entire business is embodied AI navigation at scale, making it a pure play on autonomous last-mile delivery.
BrainChip licenses its Akida neuromorphic processor IP for ultra-low-power edge AI inference in vision, audio, and sensor applications. It is a pre-scale pure play whose entire value rests on neuromorphic edge AI achieving commercial adoption.
Revenue is entirely fees for AI generated coronary analyses; the business would not exist without its models. The thesis depends on reimbursement, guideline adoption, and clinician trust rather than AI infrastructure spending.
All revenue comes from licensing and deploying the Kodiak Driver, an AI virtual driver for long haul trucks. The thesis depends on scaling driverless freight operations from a very small revenue base, which is why it sits just below the top pure plays.
Astera Labs sells PCIe retimers, active cables, CXL controllers, and fabric switches that move data between GPUs, CPUs, and memory inside AI servers. Nearly all revenue comes from AI platform deployments at hyperscalers, making it effectively a pure play on cluster buildouts.
Fermi is developing the multi-gigawatt Project Matador campus in Texas, combining gas, solar, and nuclear generation with data center capacity for AI tenants. It is essentially a pre-revenue bet that powered land and campuses for AI will command premium value.
Sells prebuilt enterprise AI applications and a development platform to energy, defense, manufacturing, and government customers. It is a pure play whose entire revenue and thesis depend on enterprise AI adoption.
Richtech builds AI service robots including the ADAM bartender robot, selling and deploying them into hospitality and commercial venues. The company has no business outside embodied AI, so the thesis rests entirely on service robot adoption and unit economics.
Assembles and sells GPU server racks and liquid-cooled AI clusters, largely NVIDIA-based, to data centers and AI cloud providers. AI systems are the bulk of revenue, making the business a direct play on AI capital spending.
Essentially all of Hesai's revenue comes from lidar sold into AI-driven autonomy and robotics programs, from driver assistance to robotaxi fleets. It is a pure play on the perception hardware autonomous systems require, with auto pricing pressure and China concentration the main thesis risks.
Symbotic sells AI-powered warehouse automation systems, deploying fleets of autonomous robots for Walmart and other major retailers. The company is effectively a pure play on warehouse robotics adoption, with its value tied to deployment pace and backlog conversion.
Recursion runs an AI drug discovery platform, pairing foundation models with massive automated biology datasets to industrialize target and molecule discovery, monetized through pharma partnerships and its own pipeline. The company exists as a bet that AI can decode biology, making it a pure play.
Innodata's growth is driven almost entirely by generative AI data work: training data creation, annotation, and model evaluation for a handful of hyperscale technology customers. The thesis depends on frontier model training budgets, with customer concentration among a few large AI labs as the main risk.
Absci uses generative AI models to design novel therapeutic antibodies, monetizing through drug creation partnerships with large pharma and its own pipeline. The company exists because of generative biology, so the thesis rests entirely on AI-designed drugs reaching the clinic.
Ambiq designs ultra-low-power Apollo SoCs that run AI inference on wearables and IoT devices at the network edge. The company is a pure play on on-device AI adoption, so the thesis hinges on edge inference spreading across battery-powered hardware.
WhiteFiber exists to sell AI compute and AI data center capacity, and essentially all revenue comes from GPU cloud services and high-density colocation. The thesis depends on GPU pricing, utilization, and continued enterprise demand for AI training and inference.
The entire company is AI autonomy software for industrial robots and defense drones; there is no hardware or legacy business left. The thesis depends on converting early deployments and defense contracts into recurring software revenue at very small scale.
Blaize is a pure play whose entire business exists because of AI inference demand, selling edge AI silicon, systems, and software. The company is small and financially strained, which is a real risk flag, but the thesis rests entirely on edge AI adoption in defense and industrial markets.
Credo sells active electrical cables, retimers, and SerDes chips that connect GPUs and switches inside AI data centers at 800G speeds. Hyperscaler AI cluster deployments drive the vast majority of revenue, making the company close to a pure play on the buildout.
Tempus sells AI-driven genomic sequencing, clinical data licensing, and therapy-matching tools to oncologists and pharma. The entire company is built on monetizing its clinical data library with AI, making the buildout thesis essentially the whole story.
Ouster exists to sell perception hardware and AI software into autonomy, robotics, and smart infrastructure, so nearly all revenue is autonomy linked. A slightly broader industrial and infrastructure mix keeps it just below the pure play tier.
Sells AI-powered decision intelligence and predictive analytics, largely to defense and government customers for logistics and security missions. Essentially all revenue is AI services, so the thesis rests on government AI contract wins.
Aeva's entire product line is perception for autonomous systems, a business that only exists because of AI-driven autonomy demand. Revenue is still pre-scale, so the thesis depends on automotive programs like Daimler Truck reaching production volume.
AI rack integration is now the core of the business, with volumes tracking Dell's AI server shipments almost one for one. The score sits just under Pure Play because nearly all revenue flows through a single customer relationship, so the thesis depends on both AI server demand and Dell keeping TSS as its integration partner.
AITX sells autonomous security robots and AI monitoring devices through its RAD subsidiaries on a subscription model, replacing human guards. The business exists entirely as a bet on AI robotics displacing security labor, though it remains small and unprofitable.
Innoviz earns its revenue entirely from lidar and perception software for autonomous driving and defense sensing. Small scale and financing risk temper what is otherwise near pure play exposure.
Sells AIP plus its Foundry and Gotham platforms so governments and enterprises can run AI on their operational data. AIP demand is the entire growth story and the reason for the valuation, making enterprise AI adoption the thesis.
AppLovin's revenue comes from its AXON machine learning engine, which prices and places mobile and e-commerce ads in real time. The business is essentially a bet on its AI models outperforming rivals, making AI the thesis rather than a feature.
Celestica builds networking switches and AI compute hardware for hyperscalers, and that data center business is now the majority of revenue and nearly all of the growth. The thesis depends on hyperscaler capital spending staying elevated, with legacy aerospace and industrial work keeping it just short of pure play.
Vertiv sells power distribution, thermal management, and liquid cooling for data centers, directly into the AI capacity buildout. Order growth and the entire investment case now hinge on hyperscaler and colocation AI capex, making it one of the most direct picks-and-shovels plays.
Oklo is developing Aurora fast reactors aimed squarely at powering data centers, with a pipeline of nonbinding customer commitments. It is pre-revenue and priced almost entirely on AI power demand converting into signed contracts and licensed reactors.
Solaris pivoted from oilfield services to mobile gas turbine power, and the large majority of its contracted backlog now serves AI data center customers, anchored by a joint venture with an xAI affiliate powering the Colossus campus. The thesis depends on hyperscalers continuing to buy fast behind-the-meter power while grid interconnection stays slow.
Upstart earns fees by originating consumer loans for bank partners using its AI credit model in place of FICO based pricing. The entire thesis is that its models price risk better than incumbents, making the business an AI pure play exposed to credit cycles.
TeraWulf converts its low-cost zero-carbon power sites into GPU hosting and AI data center capacity sold to compute customers. The investment thesis has shifted almost entirely from bitcoin mining to AI hosting contracts, making buildout demand the core driver.
Pagaya's entire business is machine learning underwriting: partner lenders route applications through its models and its investor network funds the approved loans. The thesis depends on consumer credit conditions and partner lending volumes, not on AI infrastructure spending.
POET sells optical engines built on its optical interposer for 800G and faster transceivers that connect AI data center clusters. Nearly all of its opportunity comes from AI interconnect bandwidth demand, so the thesis is a direct bet on the networking layer of the buildout.
Evolv's screening systems only work because AI models classify sensor data to flag concealed weapons, making it a physical security AI pure play by product. The thesis depends on venue security demand rather than the compute buildout, and the company has had to remediate past SEC and accounting issues.
Nearly all revenue is AI analysis delivered through the SOPHiA DDM platform; the company exists to sell data driven medicine. The thesis depends on hospital adoption of decentralized genomic testing at still small commercial scale.
Every product is an AI designed personalized implant and surgical plan; there is no off the shelf line. The thesis depends on surgeon adoption and reimbursement, with monetization through implants and instruments keeping it just below the pure play tier.
Essentially all revenue is AI speech and assistant software for vehicles, now shifting from scripted voice to generative assistants like CaLLM and xUI. The thesis depends on per vehicle royalty growth and automaker adoption, with legacy embedded voice predating the current AI cycle.
EHang's aircraft fly with no onboard pilot, making the AI flight system the core of every unit sold, unlike piloted eVTOL peers. The thesis depends on aviation certification, Chinese low altitude economy policy, and monetization through hardware sales rather than software.
Veritone sells its aiWare platform and AI services for media analytics, content management, and government intelligence applications. The company exists to sell applied AI, so the thesis is essentially a bet on enterprise and public sector AI adoption and its own path to profitability.
Datacenter optical transceivers are the majority of revenue and the entire growth story, with Microsoft and Amazon holding warrant agreements tied to purchase commitments. The thesis depends on hyperscaler AI cluster buildouts and on scaling 800G production, while a legacy cable TV segment keeps it just below pure-play territory.
Essentially all of Airship AI's revenue comes from AI surveillance and analytics platforms sold to government customers like the DOJ and DHS. The score is discounted from pure play because federal contract timing is lumpy and the company remains very small, so the thesis depends on converting its agency pipeline into recurring deployments.
SK hynix is the world leading producer of HBM, the memory that sits directly inside every major AI accelerator, and also sells conventional server DRAM. Its growth, pricing power, and capacity roadmap are now set almost entirely by AI accelerator demand from Nvidia and other buyers.
Sells high-speed Ethernet switches and network software that connect GPUs inside AI data centers, mainly to hyperscalers like Microsoft and Meta. AI cluster networking is now the central growth driver on top of its cloud networking base.
Sells custom AI accelerators for hyperscalers plus the optical interconnect chips that link AI data centers together. AI silicon has become the majority of data center revenue and the entire growth thesis, offsetting softer legacy markets.
Core Scientific converts its power-rich bitcoin mining sites into GPU colocation and high-performance computing hosting for AI customers. The investment case now rests on filling that capacity with AI tenants, with mining a declining legacy business.
Autonomous drone and counter-drone systems drive Ondas' growth and most of its story, backed by major defense orders for the Optimus and Iron Drone platforms. A legacy industrial wireless segment remains, so the thesis depends on defense autonomy procurement translating into sustained production revenue.
Ambarella sells edge AI vision processors into security cameras, automotive ADAS, and robotics, having pivoted its whole roadmap from video chips to AI inference SoCs. Nearly the entire investment case rests on edge AI adoption in cameras and autonomy.
Arbe develops 4D imaging radar chipsets with AI perception processing for automakers building ADAS and autonomous driving systems. The company is pre-scale and its entire value depends on radar based machine perception winning vehicle design slots.
MicroVision sells lidar sensors and perception software for automotive ADAS and industrial autonomy. It is a small pre-scale company whose value rests almost entirely on autonomous sensing wins, making the AI and autonomy buildout the whole thesis.
Sells custom AI accelerators for hyperscalers plus the Ethernet switch silicon connecting GPU clusters, alongside VMware software. AI semiconductors are the dominant growth engine, with legacy chip and software businesses providing a stable base.
Micron sells HBM and data center DRAM and NAND directly into AI accelerators and servers, with HBM capacity sold out well in advance. AI memory demand is now the dominant driver of pricing, mix, and the entire investment thesis.
Manufactures nearly all leading-edge AI accelerators for NVIDIA, AMD, Apple, and hyperscaler custom chips, including CoWoS advanced packaging. AI chip demand is now its primary growth driver, though smartphones and other logic remain a large share of wafers.
Applied Materials sells deposition, etch, and inspection equipment that chipmakers need to fabricate leading-edge AI processors and memory. AI-driven fab capacity is now the dominant demand engine for semicap spending, making the cycle central to the thesis.
Advantest supplies the test systems that qualify AI accelerators and HBM memory, and management attributes the majority of current demand to AI compute. The thesis depends on sustained AI chip volumes and rising test times per device.
Fabrinet manufactures the high speed optical transceivers that link GPUs and switches across AI clusters, with datacom optics for NVIDIA and hyperscale suppliers the dominant growth driver. The thesis depends on AI cluster networking staying optical and on Fabrinet holding its precision manufacturing lead through each speed transition.
Machine learning is central to test interpretation, the Caris Detect early detection launch, and the biopharma data business. The thesis depends on diagnostic reimbursement and clinical adoption, since revenue is billed as testing with sequencing as the substrate.
The discovery platform's value proposition is AI over massive single cell screening data, comparable to other computational drug discovery platforms. The thesis depends on partner milestones and royalties, which arrive slowly and keep revenue lumpy.
XTEND sells AI-piloted autonomous drone and ground robot systems under its SKYLORD brand to defense, homeland security, and commercial security customers, with revenue growing quickly off a small base and management projecting a sharp ramp through 2028. The company remains unprofitable and just became a public reporting entity through its SPAC merger, so the score reflects execution risk on an otherwise pure AI robotics business.
Sells Azure AI infrastructure and Copilot subscriptions into the cloud and software layers, monetizing its OpenAI partnership across enterprise customers. Azure AI growth now drives the investment thesis, though Office, Windows, and gaming remain large non-AI businesses.
Sells Instinct GPUs into AI data centers as the main merchant alternative to NVIDIA, plus AI-enabled CPUs for servers and PCs. The investment thesis is dominated by AI accelerator share gains, with client and gaming chips a cyclical base.
Lumentum sells lasers and 800G optical transceivers into the networking layer of AI data centers, primarily to hyperscalers and cloud builders. Datacom optics for AI clusters is now the dominant growth engine, offsetting weaker telecom and consumer 3D sensing.
Lasertec holds an effective monopoly in EUV photomask inspection, tying it directly to the leading edge capacity being built for AI compute. The thesis depends on EUV adoption and mask volumes continuing to grow, with equipment cyclicality the main caveat.
Camtek sells optical inspection and metrology systems for advanced packaging, HBM, and chiplets to chipmakers and OSATs. Its growth is driven almost entirely by AI accelerator packaging complexity, making AI the dominant engine behind orders.
Schrodinger licenses physics based computational chemistry software to pharma and uses the same platform for its own drug programs and partnerships. The whole franchise rests on simulation and ML replacing wet lab screening, so AI driven discovery is the thesis.
Luminar sells lidar sensors and an AI perception stack to automakers for assisted and autonomous driving programs. The company only works if machine perception becomes standard in vehicles, so nearly all of its value rides on the autonomy buildout.
Red Cat sells small military drones with autonomous navigation and reconnaissance AI, targeting US Army programs like short-range reconnaissance. The thesis is almost entirely defense adoption of autonomous drone systems, making AI autonomy central to the story.
ASM International sells atomic layer deposition and epitaxy tools that are mandatory for gate-all-around transistors in advanced AI chips. Its growth is tied directly to leading-edge logic capacity additions, which AI demand is now driving.
Disco dominates the grinders and dicers that thin and cut wafers, steps that HBM stacking and advanced packaging for AI multiply. The thesis depends on HBM and advanced packaging volumes continuing to outgrow the broader chip cycle.
Jabil's Intelligent Infrastructure segment builds AI servers, racks, networking hardware, and photonics for hyperscalers, and it is the fastest growing part of the company. The thesis depends on that segment continuing to outgrow a diversified base that still spans healthcare, automotive, and consumer manufacturing.
Contracted AI data center leases at campuses like Beacon Point now define Hut 8's forward value, dwarfing the legacy mining business. Current revenue is still mostly bitcoin, so the thesis depends on delivering contracted capacity on time and on the credit of its hyperscale tenants.
Designing, deploying, and managing GPU clusters is Penguin's stated identity and main growth engine, reinforced by a strategic investment from SK Telecom. The thesis depends on enterprises continuing to outsource AI cluster buildout and operations, while legacy memory and LED lines dilute the exposure.
BlackSky monetizes AI analytics through Spectra AI on top of its own satellite constellation, selling automated detection and monitoring rather than raw imagery. The thesis depends on defense and intelligence subscription growth, with AI as the differentiator that wins and expands those contracts.
Monetizes AI through Google Cloud, Gemini models, and custom TPUs, while AI powers its core search and advertising products. The thesis hinges on AI expanding cloud share and defending search economics, with a huge ad business behind it.
Sells the EUV lithography machines that TSMC, Samsung, and Intel need to produce leading-edge AI chips, holding a monopoly position. Exposure hinges on AI sustaining advanced-node capacity expansion, though memory and mature-node demand also cycle its orders.
Rents GPU capacity and cloud infrastructure through OCI to AI labs and enterprises, layered on top of its database and application franchises. AI training deals now drive its cloud backlog and the thesis, though legacy software still produces most revenue.
Tokyo Electron is a top wafer fab equipment maker whose coater/developers, etch, and deposition tools are required on every leading-edge logic and HBM line. Growth now tracks AI-driven capex, though the thesis still depends on the broader chip equipment spending cycle.
Sells da Vinci surgical robots, instruments, and services to hospitals, the dominant franchise in robotic-assisted surgery. It is a robotics pure play whose thesis rests on procedure growth, with AI features enhancing rather than driving the platform.
Bloom sells solid-oxide fuel cells that give data centers fast onsite power without waiting for grid interconnection, with large commitments including Oracle. Data center orders have become the dominant growth engine, so the thesis now rests on AI power scarcity persisting.
Onto sells inspection and metrology systems used heavily in advanced packaging for AI accelerators and high-bandwidth memory. Its growth is tied to HBM and 2.5D packaging capacity expansion, which is currently driven almost entirely by AI demand.
Nova's metrology tools are consumed most heavily by the leading edge logic, gate all around, and HBM process steps being built for AI chips. The thesis depends on AI driven equipment spending and measurement intensity rising with each node transition.
The ZT Systems deal makes Sanmina a primary US manufacturer of AI rack-scale systems with AMD as an anchor customer, layered on existing cloud and networking hardware work. The thesis depends on that new business scaling faster than its diversified legacy EMS revenue, which keeps it at the lower edge of Core.
Relay develops oncology drugs discovered through its Dynamo platform, which combines physics-based protein motion simulation with machine learning. The company exists as a computational-first biotech, though value ultimately hinges on clinical trial results, not compute demand.
Lemonade sells renters, home, pet, and car insurance run by AI chatbots for sales and claims and ML models for pricing and fraud. The whole thesis is that an AI native cost structure eventually beats incumbent insurers, though results still hinge on insurance loss ratios.
NuScale develops small modular nuclear reactors, and hyperscaler demand for carbon-free data center power is the core commercial catalyst. It is pre-revenue at scale, so nearly all upside hinges on AI-era electricity demand turning designs into orders.
One Stop Systems' strategic focus is rugged edge AI hardware that puts GPU compute on military aircraft, vehicles, and autonomous platforms. Legacy media and industrial compute remain in the mix, so the thesis depends on defense edge AI programs reaching production at a company of very small scale.
Amphenol's IT datacom interconnects have become its largest growth driver as AI servers demand denser high-speed connections and more power delivery per rack. The thesis depends on continued hyperscaler buildouts, while the broad industrial, automotive, and military portfolio keeps overall AI share of revenue moderate.
Constellation sells electricity from the largest US nuclear fleet, signing long-term deals like the Microsoft Crane restart to power AI data centers. Its growth story now hinges almost entirely on data-center demand lifting power prices and contracted volumes.
Besi leads die attach and sets the standard in hybrid bonding, the packaging step next generation HBM and chiplet AI accelerators depend on. The thesis depends on hybrid bonding adoption ramping, since today's revenue still leans on mainstream assembly markets.
AXT's indium phosphide substrates underpin the optical transceivers and silicon photonics connecting AI data center clusters, now the company's dominant growth driver. The thesis depends on AI optical connectivity demand and capacity expansion, with legacy gallium arsenide and germanium lines diluting purity.
Navitas has pivoted squarely at AI data center power, headlined by its work with NVIDIA on 800 volt DC architecture for next generation racks. Current revenue still comes largely from mobile and consumer chargers, so the score reflects a Strategic reality carrying a Core level narrative.
A large share of Arteris' licensing pipeline is AI chip designs, and its network-on-chip IP earns royalties on the AI silicon design wave rather than direct hardware sales. The thesis depends on AI design starts converting into shipped chips, since royalties follow volumes years after licenses are signed.
Data center power systems are now Power Solutions International's strategic core and largest growth driver, reinforced by vertical integration into power enclosures. The legacy industrial and transportation engine business keeps it below pure play, so the thesis depends on data center generation orders continuing to scale.
SoftBank is a founding partner and co-owner of Stargate, holds a controlling stake in Arm, and is a major backer of OpenAI, with Masayoshi Son chairing the Stargate venture. Reported results still swing on the value of a much wider investment portfolio.
Coherent sells 800G optical transceivers and photonics that link GPUs across AI clusters, plus lasers and SiC materials. AI datacom optics is its dominant growth engine and the core of the thesis, though telecom, industrial laser, and materials businesses remain sizable.
Vistra sells wholesale and retail electricity from nuclear, gas, and solar plants into markets tightening from data-center load. The stock's rerating hinges on AI power demand lifting prices and producing long-term compute-related contracts.
Qnity sells the consumables chipmakers use on every wafer, CMP pads, slurries, photoresists, and packaging materials, with growth guided to advanced nodes, HBM, and AI packaging. It serves the whole chip market, so the thesis depends on total wafer starts, not AI alone.
Iris Energy sells GPU cloud services and AI data center capacity built on renewable-powered sites originally used for bitcoin mining. The investment case now rests mostly on converting its power portfolio into contracted AI compute, with mining as the legacy business.
XPeng sells smart EVs in China differentiated by its in-house XNGP autonomous driving stack, AI chips, and robotics ambitions. Autonomy leadership is the core of the investment case, though results still hinge on EV volumes and pricing in a brutal market.
GDS builds and leases data center capacity in China and internationally, with high-density AI colocation and GPU hosting driving new bookings. The thesis hinges on Chinese and Asian AI compute demand filling capacity despite regulatory and chip supply constraints.
Keel owns energized power sites and is converting them into data center capacity for AI and high performance computing tenants, with Panther Creek in eastern Pennsylvania as the flagship campus. The thesis depends on signing hyperscaler and neocloud leases and delivering the first megawatts on schedule.
Blackstone Digital Infrastructure Trust owns newly built data center properties leased on long-term contracts to investment-grade hyperscale tenants, so its rent and occupancy growth track hyperscaler AI capacity expansion directly. Exposure is capped by its REIT structure, which depends on continued capital raising and property acquisitions rather than on owning or operating the compute itself.
Ibiden is a primary supplier of the IC package substrates under AI accelerators and server CPUs, and it is expanding substrate plants specifically for AI GPU demand. The thesis depends on accelerator volumes staying strong, since ceramics and automotive filter businesses dilute the mix.
Data center electrical work through the Communications segment is the largest and fastest growing part of the business, with hyperscale and colocation customers driving record demand. The thesis depends on the data center construction cycle continuing, since residential and commercial electrical work still make up much of the mix.
X-energy is developing the Xe-100 reactor and TRISO-X fuel, with Amazon as both an investor and an anchor customer for data center power. Reactor deliveries are years out, so the thesis depends on licensing, fuel production, and construction rather than current sales.
Spends heavily on GPUs and data centers to power AI ad targeting, recommendations, and open Llama models, monetized almost entirely through advertising. The thesis hinges on AI improving ad yield and engagement rather than direct AI product sales.
Sells electric vehicles today, but the valuation rests heavily on Full Self-Driving software, robotaxis, and the Optimus humanoid robot. Exposure hinges on autonomy and robotics shipping at scale, since car sales alone do not support the multiple.
Sells chip design and verification software that NVIDIA, hyperscalers, and AI chip startups depend on, plus AI-driven tools like Cerebrus. The AI silicon design wave is a top growth driver, but revenue spans all semiconductor and systems design.
Sells the EDA software used to design nearly every advanced AI chip, plus AI tools like DSO.ai that automate design work itself. AI chip design starts are a major demand driver, though its tools serve the whole semiconductor industry.
Rents wholesale and colocation data center capacity to hyperscalers and enterprises training and serving AI models. AI leasing has become a primary driver of bookings and its development pipeline, with legacy colocation the base.
Talen sells power from its Susquehanna nuclear plant and gas fleet, including an expanded 1,920 MW agreement supplying Amazon data centers. The thesis is monetizing scarce firm power into AI demand, making the buildout the central driver despite merchant market swings.
AI server and networking boards have made data center computing TTM's largest and fastest growing end market, driving new US capacity. The thesis depends on sustained AI hardware build rates and on TTM holding advanced PCB share, with aerospace and defense work as the other half of the story.
Advanced Energy has double AI leverage: hyperscale data center power supplies surging with AI server demand, plus the precision power systems inside semiconductor manufacturing tools. Both end markets are cyclical, so the thesis depends on hyperscaler capex and chip equipment spending holding up together.
Rambus sells memory interface chips and silicon IP that move data between processors and DRAM in servers, plus HBM and CXL controller technology. AI servers demanding faster memory bandwidth are the core growth driver, with licensing providing a stable non-AI base.
Sells automation software combining RPA robots with AI for document understanding and agentic workflows in enterprises. The thesis now hinges on agentic AI reviving growth, though AI could also commoditize traditional RPA work.
Data center driven electricity demand is the core reason Argan's power plant construction backlog has expanded to record levels, even though it never touches the data hall itself. The thesis depends on utilities and developers continuing to order new gas fired and renewable generation to serve AI load growth.
VNET operates data centers in China, with wholesale AI-ready capacity and GPU hosting driving its order book. The thesis hinges on Chinese AI compute demand and continued access to accelerators filling its high-density capacity.
AIB Data Centers builds and leases AI colocation campuses where tenants bring their own GPUs, earning rent from long term power and shell leases rather than from hardware it owns. The business exists entirely because of AI and HPC demand for dense, liquid cooled space, but revenue is still small and the company is early in converting its identified capacity into signed leases.
Sells AI compute and services through AWS, including Bedrock and custom Trainium chips, to enterprises building AI applications, plus warehouse robotics. AWS AI growth is a top driver, but retail and advertising remain most of the business.
Baidu sells AI cloud services, the ERNIE model family, and Apollo Go robotaxi rides on top of its search advertising base. AI is the entire growth thesis as search ads decline, making execution on cloud and autonomy pivotal.
BASX gives AAON direct data center cooling revenue, spanning liquid cooling and custom airside systems, and it has become the company's central growth story. The thesis depends on that segment scaling against a base of conventional commercial rooftop HVAC, keeping AAON Strategic and trending toward Core.
Data centers are the leading demand driver behind Everus's expanding backlog, and its acquisition of Epsilon Industries targets modular data center infrastructure manufacturing. The thesis depends on the data center construction cycle and on Everus converting backlog to revenue, with utility and commercial work as the base load.
DigitalBridge raises and manages funds that own data center platforms such as Vantage, Switch, DataBank, and Yondr, so fee earnings and carried interest track capital flowing into AI infrastructure. The thesis depends on continued fundraising for digital infrastructure rather than on operating the assets directly.
ADAS sensing silicon for radar, vision, and ultrasound is the growth engine and directly enables vehicle autonomy stacks. The thesis depends on rising sensor content per car, while user experience and electrification chips keep part of the portfolio outside the AI story.
Edge AI data centers and GPU hosting through Duos Edge AI have become the company's primary growth engine and capital focus, layered on a small legacy rail inspection business. The thesis depends on a micro-cap team executing large hosting agreements without stumbling.
Dell sells AI-optimized PowerEdge servers to enterprises and neocloud operators, with a multibillion dollar AI server backlog. AI infrastructure is now the company's main growth engine, though PCs, storage, and traditional servers remain the revenue majority.
Licenses chip architectures and collects royalties on processors across AI data centers, smartphones, and edge devices, including Neoverse CPUs in AI servers. Rising royalty rates on AI-related silicon drive the growth story, though smartphone royalties remain the base.
Sandisk sells NAND flash and enterprise SSDs, where AI training data storage and inference workloads are driving high-capacity drive demand. Exposure hinges on AI data-center storage attach rates lifting a cyclical memory business through the cycle.
GE Vernova sells gas turbines, nuclear services, and grid equipment into a market where electricity has become the binding constraint on AI. Its multi-year turbine and grid backlog is being driven substantially by data center load growth, though utility and renewables demand remains broad.
Rents interconnected data center space to enterprises and cloud providers, with xScale campuses built for AI and cloud capacity. AI-driven demand for power and space is a rising growth driver on a large colocation base.
Quanta builds electric transmission, substation, and power infrastructure, the grid capacity that data centers require. Despite minimal direct AI revenue, its record backlog is increasingly driven by AI-related load growth and grid expansion, making electrification for AI a central growth driver.
Data center cooling, led by its Silent-Aire hyperscale business and York chillers, is the standout growth vertical in Johnson Controls' applied HVAC backlog. The thesis still rests mostly on broad commercial buildings demand, which keeps exposure Strategic rather than Core.
Monolithic Power sells power management chips, including voltage regulator modules that power Nvidia and other AI accelerators in data centers. Enterprise data revenue tied to AI has become its biggest growth driver, though auto, industrial, and consumer remain large.
Comfort Systems installs mechanical, electrical, and plumbing systems, with a record backlog dominated by data center and chip fab construction. AI-driven industrial construction is now the primary growth engine, though it remains a diversified contractor underneath.
Ciena sells coherent optical transport systems and pluggables that carry traffic between and inside data centers for cloud and AI customers. Cloud provider spending on AI-driven interconnect bandwidth is now its key growth engine, alongside a large telecom carrier base.
Sterling performs site development, earthwork, and e-infrastructure construction, with the vast majority of that segment's backlog in mission-critical data center projects. AI construction demand is the dominant growth engine on top of a transportation and building base.
MKS content in RF power, vacuum, and photonics rides inside most etch and deposition tools, and Atotech chemistry feeds AI class substrates and packaging. The thesis depends on AI driven equipment spending, diluted by industrial and electronics end markets.
Data center heat rejection is a fast growing, named end market for SPX's cooling segment, which sells dry, evaporative, and hybrid coolers under brands like Marley. Its data center tilt is clearer than the large diversified HVAC names, but detection and measurement and other HVAC markets still carry the rest of the thesis.
Cognex sells machine vision systems and deep learning software that guide robots and inspect products in factories and logistics hubs. AI-based vision is the product itself, with growth tied to factory automation cycles in electronics, autos, and e-commerce.
Modine sells data center cooling equipment including CDUs, chillers, and CRAHs through its Airedale business, alongside legacy vehicle and HVAC thermal products. Data center cooling has become the growth engine and the reason to own the stock.
FormFactor sells probe cards used to test advanced logic chips and HBM memory at the wafer level, selling to foundries and memory makers. HBM and AI processor test demand is the primary growth driver in an otherwise cyclical test market.
AeroVironment sells autonomous drones and loitering munitions like Switchblade to the US military and allies, with onboard autonomy and the Crysalis control system. Rising demand for attritable autonomous weapons is the core thesis, though products span conventional unmanned systems too.
Cipher builds large powered data center sites and is leasing capacity to AI compute tenants alongside its bitcoin mining. The thesis has shifted toward AI hosting contracts, with its 600MW plus power pipeline valued mainly as AI-ready capacity.
D-Wave sells quantum annealing systems and Leap cloud access for optimization and hybrid quantum-classical machine learning workloads. The stock trades as a frontier compute bet adjacent to AI, and exposure hinges on quantum proving useful for commercial AI-scale problems.
Allegro's current sensors and motor drivers ride the AI server power and cooling buildout, and data center has become its fastest-growing market. Automotive and industrial still dominate revenue, so the thesis depends on content gains per AI rack continuing to outgrow the legacy base.
Zeta sells an AI-driven marketing platform that uses its identity data and machine learning to target and retain customers for brands. AI-based targeting is the core product differentiation, but the business ultimately rides advertising and marketing budgets.
Butterfly sells handheld semiconductor-based ultrasound probes to clinicians, with deep learning for guided scanning and automated image interpretation. AI is the core differentiator that lets non-specialists scan, so the thesis hinges on AI-enabled point-of-care imaging displacing cart systems.
Ichor builds the fluid delivery subsystems inside the etch and deposition tools driving AI era fab expansion, so revenue tracks equipment OEM shipments. The thesis depends on wafer fab equipment spending staying AI led, since Ichor has no differentiated AI product of its own.
Ultra Clean supplies gas delivery systems, chamber subsystems, and parts cleaning services to semiconductor equipment makers and fabs. Its revenue tracks wafer fab equipment spending, which AI chip capacity is now the primary force driving, though it also carries broad semicap cyclicality.
Vicor sells high density power conversion modules used in vertical power delivery to AI GPUs and accelerators in data centers. AI server power is now the main growth engine and thesis driver, layered on legacy industrial, aerospace, and computing power business.
Verint sells AI bots and analytics that automate contact center work, selling into enterprise customer engagement budgets. The whole strategy is converting its installed workforce software base to AI-driven automation, so the thesis is now mostly an AI adoption bet.
PROCEPT sells the AquaBeam surgical robot plus per procedure handpieces for BPH treatment, with AI assisted planning and real time tissue mapping. It is a robotics adoption story at heart, hinging on surgeons converting procedures to its robot rather than on AI demand broadly.
Lightwave Logic exists almost entirely to serve AI-era interconnect demand, with electro-optic polymer modulators aimed at faster, lower-power optical links and design-win engagements in the transceiver supply chain. The company is still pre-commercial-scale, so the thesis depends on converting those engagements into volume adoption, not on today's sales.
Root prices auto insurance with telematics data and ML models that score actual driving behavior instead of demographics. AI underwriting is the entire differentiation, but value ultimately depends on loss ratios and growth in a brutal auto insurance market.
Grid Dynamics sells digital engineering and consulting services, increasingly building generative AI, MLOps, and personalization systems for enterprise clients. AI project demand is the main growth story, though the base business is broader software outsourcing.
Terrestrial Energy is developing a molten salt reactor aimed at industrial and data center power, and has signed early stage agreements with large power users. The company has no reactor revenue and those agreements are not binding, so the score reflects development stage rather than delivered capacity.
Alibaba sells AI cloud computing and its Qwen model family through Alibaba Cloud, alongside heavy capex on AI infrastructure in China. Cloud AI is now the group's main growth story, but the much larger e-commerce business still anchors the valuation.
Data center cooling demand is a repeatedly cited growth driver for Trane's applied HVAC business, with chillers and thermal management systems going into AI server halls. The revenue base is still broad commercial and residential HVAC plus transport refrigeration, so the thesis depends mostly on general construction, not the buildout.
AI data center interconnects are TE's fastest-growing product line, backed by explicit hyperscaler design wins for high-speed sockets, cable assemblies, and connectors. The thesis depends on that ramp continuing, since transportation and industrial connectors still dominate total revenue.
Advanced packaging for AI accelerators is Amkor's highest-growth business, and its planned Arizona campus ties it directly to the US AI supply chain alongside TSMC. The thesis depends on AI packaging volumes migrating to Amkor's 2.5D and fan-out lines, since mobile and automotive packaging still dominate revenue.
Energy Vault builds storage and dispatchable power projects and has moved into dedicated generation for AI data centers, including a gigawatt scale hyperscaler agreement and an owned campus in Texas. Its results still depend on the broader grid storage market and on turning that pipeline into operating assets.
Alpha and Omega has real content in AI accelerator and server power stages, which now drive its computing segment growth. PC, consumer, and industrial products still make up most of revenue, so the thesis depends on per-server power content rising as accelerators draw more current.
Sells deposition and etch equipment used to fabricate AI logic chips and the high-bandwidth memory they require. AI-driven capacity expansion at foundries and memory makers is a key demand driver, though orders cycle with all chip capex.
Sells inspection and process control equipment that improves yields on advanced-node AI chips and packaging. Leading-edge AI capacity buildouts drive incremental demand, but revenue depends on overall semiconductor capital spending cycles.
Eaton sells switchgear, UPS systems, and power distribution equipment installed in nearly every hyperscale data center. Data centers are its fastest-growing vertical and the core of the bull case, while a large industrial, aerospace, and vehicle business remains.
Sells Now Assist and AI agent add-ons on top of its enterprise workflow platform, priced as premium SKUs for IT, HR, and service teams. AI upsell is a top growth driver, but the base business is workflow subscriptions.
AI-related advanced packaging and test is ASE's fastest-growing revenue line, and it absorbs overflow from TSMC's CoWoS ecosystem for accelerators and networking chips. The thesis depends on that leading-edge work scaling, since mainstream consumer and communications packaging still makes up the bulk of volume.
HPE sells AI servers and Cray supercomputers to enterprises, governments, and model builders, plus Juniper and Aruba networking into data centers. AI systems are its fastest growing segment and a top-two thesis driver, but margins there are thin and legacy infrastructure remains the base.
Semtech sells analog chips including CopperEdge retimers and optical components used to link GPUs inside AI data centers. That data center line is the fastest-growing segment, but LoRa IoT and consumer connectivity remain a large non-AI base.
Forgent designs and manufactures the switchgear, transformers, and power distribution equipment that connects utility power to hyperscale data center loads, with data centers already representing 42 percent of its revenue and 47 percent of its backlog. Exposure hinges on hyperscaler capital spending staying strong enough to keep pulling grid equipment orders ahead of its legacy industrial and utility business.
SUMCO's pure play wafer exposure gives it direct leverage to the leading edge logic and HBM DRAM capacity being built for AI. The thesis depends on 300mm wafer demand and pricing, which swing with the whole semiconductor cycle.
AI and self service products, led by CXone Mpower agents, are NICE's growth engine and are reported as a distinct revenue line. Most revenue still comes from the established CX and financial crime cloud base, so the thesis depends on enterprises adopting AI agents at scale.
Csquare leases colocation and data center space to hyperscalers and enterprises, with AI workloads now the fastest-growing part of its leasing pipeline on top of a large legacy colocation base. Exposure hinges on AI capacity demand continuing to outpace the legacy enterprise colocation business it was built from.
Centrus enriches uranium and is the only US-licensed producer of HALEU, the fuel required by the advanced small modular reactors data centers are contracting. Exposure hinges on the SMR wave materializing, which is itself an AI power demand story.
Sify's growth thesis and capital spending center on AI-ready data centers and GPU colocation for the India buildout. Current revenue still leans on legacy network and ICT services, so the thesis depends on Indian enterprises and cloud players actually filling that new capacity.
Nano Nuclear is developing microreactors aimed squarely at on-site power for AI data centers, with its KRONOS construction permit application accepted for review by the NRC. The company is pre-revenue with prototype operation targeted around the end of the decade, so the score is capped at development stage and the thesis depends on licensing and execution rather than current sales.
Deep Fission is developing the Gravity Reactor, a small modular nuclear design installed underground, with a co-development agreement to power Endeavour Energy's edge data centers and non-binding letters of intent covering a large amount of potential capacity aimed largely at AI load growth. The company has no reactor revenue and its first units are not expected online until 2029, so the score reflects development stage rather than delivered power.
Aeluma is a development-stage company whose commercial thesis centers on photonics for AI data center optical interconnects, funded today mostly by government contracts from NASA and the Department of Defense. The thesis depends on turning its quantum dot laser on silicon technology into hyperscaler supply chain wins before meaningful commercial revenue exists.
Infineon is one of the clearest power semiconductor beneficiaries of the AI server buildout, supplying PSUs, power stages, and SiC and GaN devices with a named NVIDIA collaboration on 800 volt HVDC architecture. Automotive remains the bulk of revenue, so the thesis depends on car and industrial cycles, with AI power as the fastest-growing line.
Robotics revenue is large but most units shipped are traditionally programmed, with AI adaptive robots pursued through i3-Mechatronics and an NVIDIA alliance. The thesis depends on AI native robotics becoming the revenue base rather than just the growth vector, similar to peer Fanuc.
Hinge Health's TrueMotion computer vision and AI care automation genuinely deliver the product, replacing much of the labor in physical therapy. Revenue comes from employer digital health contracts, so the thesis depends on benefits adoption, with AI as the margin engine rather than the offering itself.
Certara's biosimulation platform is built on modeling and machine learning, and Certara.AI extends it with generative AI for regulatory writing and drug development. Much of the business is classical simulation and consulting, so the thesis depends on pharma R&D software spending, with the modern AI layer as the growth driver.
BUZZ HPC gives HIVE a genuine operating GPU cloud building sovereign AI capacity in Canada. It remains a small share of a mining-dominated revenue base, so the thesis depends on scaling that segment while bitcoin economics fund the transition.
Sells watsonx AI software, Granite models, and AI consulting engagements to enterprises, alongside hybrid cloud and mainframes. A growing AI book of business anchors the growth story, but most revenue is legacy software, infrastructure, and services.
Sells Agentforce AI agents and Einstein features as paid add-ons across its CRM, service, and marketing clouds to enterprises. AI is the key growth story, but the vast majority of revenue is still core subscription software.
Sells Snapdragon processors with NPUs that run AI on smartphones, PCs, and cars, plus patent licensing royalties. The on-device AI upgrade cycle is a key part of the thesis, though handset demand still dominates revenue.
Sells a cloud data platform where enterprises store and query the data feeding AI, plus Cortex AI for models and applications. The thesis increasingly hinges on AI workloads driving consumption, though core analytics remains the revenue base.
Nokia acquired Infinera to attack data center optical interconnect and sells IP routing and optical gear to hyperscalers and neoclouds, with an NVIDIA equity investment cementing an AI-RAN and AI networking collaboration. The thesis depends on that pivot outgrowing the legacy mobile networks business that still dominates revenue.
Carrier launched its QuantumLeap data center suite, spanning chillers, coolant distribution units, and liquid cooling integration, to chase the AI buildout. It entered the vertical later than Vertiv, Johnson Controls, and Trane, and data centers remain a small slice of a residential and commercial HVAC base, so exposure is Growing.
Flex manufactures electronics for other companies, and building AI servers, racks, and data center power products for hyperscalers is now its fastest-growing business. The AI data center segment increasingly sets the growth rate, though autos and consumer manufacturing remain large.
Cameco mines uranium and owns a stake in Westinghouse, supplying fuel and services to the nuclear fleet. AI data center power demand is now the marginal force behind reactor restarts and new builds, making it a top driver of the uranium thesis.
FANUC sells industrial robots and CNC systems worldwide, adding AI for servo tuning, bin picking, and predictive maintenance. As a leading robot maker it is a direct play on factory automation, though demand today tracks auto and electronics capex rather than AI itself.
nVent sells electrical enclosures, connection systems, and liquid cooling infrastructure into data center construction. Data centers are its fastest-growing end market and a top-two driver, alongside a broad industrial electrical base.
Hubbell's utility segment sells transmission, distribution, and substation equipment directly into grid modernization driven by data center load growth. The broad electrical portfolio dilutes pure AI linkage, so the thesis depends on sustained utility capex rather than hyperscaler spending alone.
Seagate sells high-capacity hard drives to cloud and AI operators storing the massive datasets behind model training and inference. Nearline drive demand from AI data lakes and HAMR-driven capacity gains are now the central growth story, though HDDs serve general cloud storage too.
SiTime sells MEMS precision timing chips into AI accelerators, data-center networking, and 5G infrastructure. AI data-center design wins are its fastest-growing segment and the core of the growth story, though diversified electronics demand remains significant.
Data center distribution has become a meaningful share of sales and WESCO's primary growth engine, supplying electrical gear, racks, and network infrastructure to hyperscale projects. The base business is broad industrial and electrical distribution, so exposure is Growing.
Sells test equipment used to validate complex AI accelerators and high-bandwidth memory, plus Universal Robots cobots and mobile robots. AI chip test demand is a key growth driver, with a direct robotics business on top.
Kratos builds low cost autonomous combat drones like Valkyrie plus target drones, satellite comms, and hypersonic systems for defense customers. The upside case centers on AI enabled collaborative combat aircraft programs, but today's revenue is mostly conventional defense hardware and services.
Hyperscaler funded long haul and middle mile fiber tied to AI data center connectivity is a fast growing demand stream on top of Dycom's telecom construction base. The thesis depends on data center interconnection continuing to drive new route construction beyond the carrier fiber-to-the-home cycle.
MaxLinear's Keystone PAM4 DSPs and interconnect silicon compete for sockets in AI data center optics against Marvell, Broadcom, and Credo, making AI infrastructure its strategic growth bet. The thesis depends on winning those designs at scale, since broadband and connectivity chips still generate most revenue.
Powell builds custom switchgear and integrated power control rooms for data centers, utilities, and oil and gas facilities. Data center electrical orders are a fast-growing top-two driver, but the petrochemical and industrial base still supplies most revenue.
RadNet pairs the largest US outpatient imaging network with a DeepHealth AI segment selling cancer screening software, consolidated further by the 2025 iCAD acquisition. AI is a fast-growing but still small piece of the business, so the thesis depends on scan volumes and how quickly AI screening becomes standard across its centers.
Pega sells enterprise decisioning and workflow automation software, with GenAI Blueprint used to accelerate application design and modernize legacy processes. AI is the centerpiece of its current growth pitch, layered onto a mature workflow and CRM base.
MYR is positioned on both the building side and the grid side of data center electrical demand, wiring facilities and constructing the transmission and substations that connect them. The thesis depends on data center work growing within a revenue base that remains broad utility and commercial electrical construction.
iRhythm sells the Zio cardiac monitoring patch, whose deep-learning ECG analysis is core to how it diagnoses arrhythmias at scale. AI is central to the product's economics and accuracy, but demand comes from cardiac care volumes, not AI infrastructure spending.
Bitdeer runs a real but small NVIDIA GPU cloud and can convert parts of its large global power portfolio to AI hosting. The thesis depends on executing that conversion, since mining and related services still produce most revenue.
Veeco's annealing, ion beam, and wet processing niches gain from leading edge AI node transitions and EUV mask blank production. The thesis depends on those niches holding share, with compound semiconductor and data storage markets tempering the AI signal.
Bitzero Holdings leases its full 110 megawatt Norway data center campus to AI infrastructure operator OneQode under a 15 year, roughly 2.6 billion dollar contract, and is deploying Nvidia Blackwell GPUs to enter AI and HPC hosting directly. Exposure hinges on converting that large contracted backlog into recognized revenue, since most of its current revenue still comes from Bitcoin mining rather than AI hosting.
Fluence is a pure play on grid-scale battery storage, and storage attached to renewables and grid capacity serving data center load growth is a primary driver of its backlog. Most contracts come through utilities and developers rather than hyperscalers directly, so the thesis depends on the broader storage cycle.
Corning sells optical fiber and connectivity solutions that hyperscalers buy in volume to wire AI data centers, alongside display and specialty glass. The optical segment's AI orders are now the clearest growth driver, though legacy glass businesses remain large.
Data center generator sets are the standout growth story inside Cummins, with AI demand producing record orders and multi year backlogs for its Power Systems segment. The company is still dominated by truck and off highway engine markets, so exposure is Growing.
GE HealthCare holds more FDA authorizations for AI-enabled devices than any other company, and deep learning features like AIR Recon DL increasingly sell its scanners. Revenue remains predominantly imaging hardware and services, so the thesis depends on hospital capital spending, with AI as a share gainer rather than the core driver.
Western Digital sells enterprise hard drives and platforms into cloud and AI data centers that need cheap bulk storage for training data. AI-driven nearline demand underpins the growth story, though the business also serves general cloud and client storage cycles.
Lumen's Private Connectivity Fabric deals with Microsoft and other hyperscalers made AI data center interconnection the company's central growth story, backed by a major fiber supply agreement with Corning. The thesis depends on converting that contracted AI demand into durable growth while legacy telecom services continue to decline.
Primoris has real and growing data center and power delivery exposure, with management citing data center related contracts as a key growth driver. The thesis depends on that work scaling inside a diversified energy and utility construction base that still sets the company's overall trajectory.
Denison is a pre-production uranium developer levered to the nuclear expansion that AI electricity demand is driving, with the high grade Phoenix deposit as its flagship. Exposure runs through uranium prices and utility contracting rather than direct AI contracts, and carries commodity and construction risk.
American Superconductor's grid interconnection and power quality systems are winning orders tied directly to data center and semiconductor fab connections, and management cites AI-driven grid demand in record bookings. Small scale and a mixed wind and defense business add variability, so the thesis depends on that order momentum holding.
Apple sells devices whose silicon carries neural engines for on-device AI, and Apple Intelligence is its play to make AI a reason to upgrade iPhones. AI is now a top strategic driver of the upgrade cycle and services, but the massive hardware franchise stands on much more than AI.
Tencent builds its Hunyuan foundation model and sells AI cloud services, while embedding AI across WeChat, games, and its advertising engine. AI is a top-two growth driver, especially in ad targeting and cloud, but gaming and social platforms remain the core of the business.
Sells the Falcon security platform with AI-driven detection and the Charlotte AI analyst to enterprises. It benefits from AI-era attacks and budgets, but growth tracks cybersecurity spending rather than the AI infrastructure buildout itself.
Sells consulting and IT services that help enterprises deploy generative AI, with billions in GenAI bookings alongside broader transformation work. AI is the fastest-growing practice, but most revenue remains general consulting and outsourcing.
Sells Firefly generative AI features and credits inside Creative Cloud and Acrobat subscriptions to creative and enterprise users. AI is both its growth engine and its biggest competitive risk, while the core remains mature creative software subscriptions.
ST supplies power conversion silicon for AI data centers, including 800 volt HVDC work aligned with NVIDIA's architecture, plus edge AI microcontrollers like the STM32N6. Automotive and industrial dominate revenue, so the thesis depends on those cyclical markets, with AI power and edge AI as growing lines.
EMCOR provides electrical and mechanical construction services, with data centers among its largest and fastest-growing end markets. The buildout is a top driver of backlog growth, but healthcare, manufacturing, and commercial work keep the base broad.
GlobalFoundries does not make leading-edge AI compute, but its specialty processes in silicon photonics, power management, and RF feed the data center bill of materials around every accelerator. The thesis depends on AI driving volume through those platforms and on its US capacity, backed by CHIPS Act support, gaining strategic value.
Dover owns genuine picks and shovels liquid cooling components: CPC quick disconnect couplings used in direct to chip loops and SWEP heat exchangers, which management cites among its fastest growing product lines. They sit inside a large diversified industrial portfolio, so exposure stays Growing.
NRG sells merchant power and retail electricity, having doubled its generation fleet with LS Power's gas assets as data center load tightens Texas and PJM markets. Exposure hinges on AI demand lifting power prices and landing large data center supply deals.
Sells ultra-pure materials, filters, and handling products required throughout advanced chip fabrication, including AI logic and memory. AI-driven leading-edge capacity growth lifts demand, though revenue follows the whole semiconductor cycle.
Hyve Solutions is a genuine hyperscale AI hardware ODM, designing and racking servers for large cloud buyers. The thesis depends on Hyve winning more hyperscaler programs, since the much larger distribution business tracks broad IT spending rather than the buildout.
Kuaishou earns short-video advertising and live-commerce revenue, and sells Kling, one of the leading AI video generation models. Kling and AI ad tools are the incremental growth story, but the core remains a consumer video ad business.
Galaxy is a crypto financial services firm building the 1.6 GW Helios AI campus in Texas with CoreWeave as anchor tenant. Helios is the growth story and rerating case, but trading, asset management, and crypto markets still drive most current results.
Silicon Motion's MonTitan platform puts it in enterprise SSDs for AI servers, where controller performance requirements are rising fastest. The thesis depends on winning hyperscale and enterprise designs, since client PC and smartphone storage controllers still generate most revenue.
Sells search and observability software, with Elasticsearch vector search powering RAG and AI-driven applications for developers. GenAI search workloads are a real growth driver, though most revenue remains core search, logging, and security.
Sells the Singularity endpoint and cloud security platform with autonomous AI detection and the Purple AI analyst. The product is AI-native, but demand comes from security budgets, so exposure to the buildout is indirect.
Planet operates a satellite fleet imaging Earth daily and sells AI-derived analytics like change detection to defense, agriculture, and government customers. AI turns its imagery archive into scalable products, but demand is driven by geospatial intelligence budgets rather than AI infrastructure.
Uranium Energy Corp produces and develops US uranium, selling into a domestic fuel market tightened by reshoring policy. AI data center power demand is the key force reviving American nuclear, so the thesis is increasingly tied to the buildout, though uranium prices set the outcome.
Novanta supplies vision, photonics, and motion components to robot OEMs rather than AI products of its own. The thesis depends on unit growth in surgical robots, warehouse automation, and industrial robotics pulling through more content per machine.
Five9 sells cloud contact center software where AI virtual agents and agent assist are the main new bookings driver for enterprise customers. The thesis increasingly depends on Five9 monetizing AI automation faster than AI shrinks the seat count it bills on.
Sprinklr sells a customer experience platform to large enterprises, with AI agents for customer service and social listening as its main upsell. AI-powered CCaaS is the growth bet, but the thesis depends on execution in a crowded market, not AI infrastructure demand.
Xometry runs a custom manufacturing marketplace where an AI quoting engine prices parts instantly and ML matches jobs to suppliers. The AI engine is its core moat and margin driver, but demand comes from industrial production cycles, not the buildout itself.
CEVA licenses semiconductor IP including NeuPro AI processor cores and sensor hub DSPs to chipmakers building edge devices. Royalties from on-device AI inference are the key growth story, but much of current revenue comes from connectivity and audio IP.
enCore is a producing US uranium miner whose demand outlook is materially driven by AI-linked nuclear restarts and new build commitments. It has no direct hyperscaler contracts, so the exposure stays indirect and rides the uranium price cycle.
Nano-X sells low-cost digital X-ray systems paired with AI software that flags disease in scans, sold to imaging providers. The thesis depends heavily on adoption of its AI radiology offering, but the company is pre-scale and commercially unproven.
Dominion's Virginia utility serves Loudoun County's Data Center Alley, the largest data center cluster in the world, and data center projects dominate its interconnection pipeline. The regulated model caps the upside, but data center load is the defining driver of its capex and growth plan.
Hoya holds two chokepoint positions tied to AI: it dominates EUV photomask blanks used to pattern leading-edge chips and leads in glass substrates for hyperscale hard drives. The thesis depends on those two niches growing, since eyeglass lenses and medical optics generate the majority of revenue.
NetApp is positioning its ONTAP software, all-flash arrays, and first-party cloud storage services as the data foundation for enterprise AI, including NVIDIA-validated AIPod systems. The thesis depends on AI workloads driving new storage purchases, since most revenue still comes from traditional enterprise storage refresh cycles.
Actuator modules and speed reducers for humanoid robots plus water cooling for AI servers are real but small slices of a diversified motor giant. The thesis depends on humanoids and liquid cooled data centers scaling enough to move a very large revenue base.
MasTec builds power delivery, generation, and data center site infrastructure, and management points to AI driven demand as a tailwind across its Clean Energy and Power Delivery segments. The thesis depends on that work becoming dominant within a broad mix that still spans telecom and pipeline construction.
Jacobs provides engineering, design, and construction management for data centers, semiconductor fabs, and power infrastructure, and has been named on AI campus builds. Data centers are a growing share of a portfolio still led by water, transport, and government programs.
AI powers Check Point's threat detection through ThreatCloud AI and it sells GenAI security products like Infinity AI Copilot, but the core business is a broad legacy security franchise. The thesis depends on enterprise security budgets and platform consolidation, not on the AI buildout directly.
NexGen is developing Rook I in Saskatchewan, the largest development-stage uranium deposit in the Western world. Its value rests on future uranium prices, which AI-driven nuclear demand underpins, but it is a pre-production single-asset bet, not a direct AI supplier.
ACM Research's cleaning and plating tools benefit from global fab construction, including AI driven leading edge and packaging capacity. The thesis depends on broad capacity buildout, with China concentration and mature node weighting capping the AI purity.
Benchmark builds HPC and AI compute hardware and supplies precision components to semiconductor capital equipment makers, both growing lines. The thesis depends on continued outsourcing wins in computing and semi-cap, while medical, industrial, and defense manufacturing still anchor most of the business.
FuelCell Energy formed a dedicated data center power venture with Diversified Energy and TESIAC and pitches its fuel cells as bridge power for AI campuses, giving it a concrete AI channel. The company is small and execution stage, so the thesis depends on converting that pipeline into operating projects.
Eos makes long-duration storage whose pipeline and DOE-backed capacity expansion are levered to grid storage demand driven by data center load growth. The company is early revenue with execution risk, so the thesis depends on converting its pipeline into delivered, profitable systems.
Willdan's grid planning and energy program work scales with data center driven load forecasting, and management cites AI load growth as a core demand driver. It is a small services firm with a diversified public sector base, so the thesis depends on utility and municipal budgets continuing to fund that planning wave.
Curtiss-Wright supplies critical components for nuclear plants, so the AI-driven push for new reactors, restarts, and SMRs expands its commercial nuclear pipeline and aftermarket. Defense and industrial segments remain the majority of the business, keeping the thesis tied to defense budgets as much as the buildout.
Hyperscaler contracts anchor Brookfield Renewable's development pipeline, led by its global framework agreement with Microsoft covering more than 10 gigawatts of renewable capacity for data centers. The large diversified hydro, wind and solar base keeps the AI linked share a minority of cash flows.
SK Telecom has set up SK Hyper as a dedicated AI data center company, develops its own A.X foundation models, and sits inside SK Group alongside memory maker SK Hynix. Mobile telecom still supplies the bulk of its revenue.
Fluor builds semiconductor fabs and data centers, including work on TSMC's Arizona campus, and those awards are a growing share of backlog. The thesis depends on continued advanced-technology construction spending, while energy, mining, and government projects still drive most of the business.
Atkore's conduit, cable tray, and raceway products go into every data center build, and management cites data centers as a key demand vertical. Products are commodity-adjacent and end markets span broad construction, so the thesis depends on nonresidential construction volumes as much as the buildout.
Uranium Royalty offers clean levered exposure to the uranium cycle that AI-era nuclear demand is tightening, without mine operating risk. The linkage is indirect and the company is small, which caps the score.
SpaceX earns revenue from Starlink connectivity and Falcon and Starship launch services, and gained Grok and frontier model exposure by absorbing xAI. AI and orbital compute are a growing pillar, but the valuation still rests primarily on Starlink and launch dominance.
Sells AI-driven security platforms like Cortex XSIAM to enterprises, and secures AI applications and cloud infrastructure. Growth comes from security spending broadly, with AI as product differentiation more than a direct buildout revenue stream.
Sells ERP and business software, embedding the Joule copilot and Business AI features into its cloud suites for enterprises. AI supports cloud migration and pricing, but growth is driven by the core ERP transition.
Sells edge cloud services including Workers AI inference, AI Gateway, and vector search on its global network, alongside security and delivery. AI developer services are a fast-growing new act, while most revenue remains core security and performance products.
Freeport mines copper, which data centers, grid upgrades, and electrification consume in far greater quantities than conventional construction. AI demand is a growing pillar of the copper deficit thesis, but pricing still depends on global industrial cycles and China.
Sells cloud monitoring subscriptions, now including LLM observability and Bits AI, to companies running modern applications. AI-native customers and AI workload monitoring are a growing driver, but the core business tracks general cloud usage.
Shin-Etsu's dominant positions in silicon wafers, photoresists, and mask blanks give it real leverage to AI driven wafer starts. The thesis depends on chip materials demand, while PVC and industrial chemicals keep the group's AI share modest.
L3Harris develops AMORPHOUS drone swarm autonomy software, AI-enabled electronic warfare, and autonomous vessels, giving it real defense AI programs. These remain a small share of a large diversified prime, so the thesis depends on overall defense budgets, with autonomy as a growth vector consistent with peer primes.
Sells Tasers, body cameras, and cloud software to police, with AI products like Draft One report writing driving premium plans. AI features are a growing upsell in its software bundle, though hardware and evidence subscriptions dominate revenue.
Iron Mountain earns most revenue from records storage and information management while building a multi-gigawatt data center colocation platform on land it already owns. Data centers are the growth story and rerating case, but the legacy business still dominates revenue.
Sells the Atlas database platform, now with vector search that lets developers build AI applications on operational data. AI apps are a growing consumption driver, but revenue mostly tracks general software workloads running on MongoDB.
MACOM sells optical and RF components, including laser drivers and high-speed analog parts for data center interconnects. AI-driven datacom demand is its fastest growing segment, but telecom, industrial, and defense markets still supply most revenue.
Dynatrace sells observability software built around its Davis causal AI engine, monitoring cloud applications for large enterprises. AI is genuinely core to the product and complex AI workloads add monitoring demand, yet spending follows cloud budgets more than AI capex directly.
Akamai is repositioning its global edge network as AI inference infrastructure through Akamai Inference Cloud, built with NVIDIA. AI revenue is small today, so the thesis depends on security and delivery demand while distributed inference adoption plays out.
Booz Allen sells consulting and technology services to US defense and intelligence agencies, with AI deployment a flagship offering. Federal AI adoption is its clearest growth vector, but revenue depends on government budgets and contract cycles across many non-AI programs.
Aptiv sells ADAS compute, sensors, software, and vehicle architecture to automakers building automated driving features. Its growth thesis rests on rising software and AI content per vehicle, though today's revenue is still dominated by conventional auto parts cycles.
Riot mines bitcoin at massive powered sites and is repurposing land, power, and cooling capacity toward AI and HPC hosting. Exposure hinges almost entirely on converting that power portfolio into signed AI tenant deals, since current revenue is still bitcoin mining.
MP Materials mines and refines rare earths at Mountain Pass and manufactures magnets used in servers, robotics, EVs, and defense, with DoD and Apple offtakes. Robotics and data center demand is a growing pillar, but EV and defense markets still dominate volumes.
Duolingo sells language learning subscriptions, with AI generating lessons, powering conversation practice, and gating its premium Max tier. AI is central to product velocity and pricing, but revenue rides consumer subscription growth rather than AI infrastructure demand.
Data center backup power, served through UPS battery lines like DataSafe, is a real growth market inside EnerSys' Energy Systems segment. Most revenue still comes from motive power for forklifts and broad reserve power markets, so exposure is Growing.
EXL sells data analytics and AI-led operations services to insurers, banks, and healthcare firms, embedding AI agents into client workflows. Its repositioning from BPO to AI services is the growth story, though much revenue still comes from traditional process management.
Mercury sells rugged processing subsystems that put sensor, radar, and electronic warfare compute onto military platforms. Defense AI processing is a real growth vector, but revenue depends on defense program schedules more than on the commercial AI buildout.
Klaviyo sells marketing automation to e-commerce brands, using AI for segmentation, predictive analytics, and campaign generation. AI features deepen the product and pricing, but growth ultimately tracks e-commerce marketing spend rather than the AI buildout.
Kulicke and Soffa's core wire bonding serves mainstream automotive and consumer chips, while thermocompression bonding placements for HBM stacking give it a real AI angle. The thesis depends on TCB adoption scaling from a minority of revenue into a major line.
COPT leases data center shell buildings in Northern Virginia to a hyperscale cloud tenant while its core business remains defense office and secure campus space. The shells add steady AI-linked growth, but the thesis depends on US defense spending and government leasing, not the buildout.
Enchanted Rock sells onsite natural gas microgrids and backup generation, historically to grocery, healthcare, and industrial customers, and is now pitching utilities and data center developers on fast, flexible power connections for AI facilities. Exposure hinges on AI data center power deals becoming a material share of a business still rooted in general commercial backup power.
Bel Fuse supplies power conversion modules, magnetics, and integrated connector modules used in AI servers and the networking gear that links them, and management cites data center power as a growth driver. The thesis depends on a diversified components portfolio spanning aerospace, industrial, and consumer markets, so AI is one driver among many.
Aehr sells wafer-level burn-in and test systems, historically for silicon carbide EV chips and now expanding into AI processor burn-in. The AI test opportunity is the main new growth lever, but the base business remains tied to the EV power semiconductor cycle.
Energy Fuels produces US uranium and separates rare earths at its White Mesa mill, touching both the nuclear fuel and magnet supply chains. AI power and robotics demand support both markets, but commodity prices and ramp execution drive the stock.
Cohu supplies back end test handlers, contactors, and inspection systems with an emerging HBM and AI device test opportunity. The thesis still depends on automotive, mobile, and industrial chip demand, which dominate its revenue today.
Globant sells AI development studios and services to enterprises adopting generative AI, billing for design and build work in the software layer. It has repositioned as AI native and AI subscriptions drive growth, yet most revenue is still broad digital services.
Data centers are a named growth vertical where Limbach's mechanical and cooling expertise applies directly, but healthcare, industrial, and institutional buildings still dominate the mix. The thesis depends on the company converting its mission critical push into a larger share of revenue over time.
Data center power distribution, led by busbars, is Methode's clearest growth engine, with management repeatedly citing record data center power product sales on AI demand. The company remains majority automotive and is working through operational restructuring, so exposure is Growing.
Agora sells real-time voice and video APIs to developers, and conversational AI agents that need live audio infrastructure are its emerging growth driver. Exposure hinges on voice AI applications scaling on its network, on top of a mature video engagement base.
Mirion supplies the radiation detection and instrumentation that nuclear restarts, uprates, and SMR programs all require, making it a picks-and-shovels play on the AI-driven nuclear revival. Medical and defense segments dilute the linkage, so the thesis depends on the pace of nuclear construction rather than AI demand directly.
Meta's Hyperion data center in Richland Parish, Louisiana, one of the largest single data center power commitments in the US, anchors Entergy's capital plan along with Amazon Web Services load in Mississippi. The thesis depends on regulated returns from building the plants and grid to serve that load, across a broad legacy utility base.
Cisco sells networking switches, optics, and security into data centers, and AI infrastructure orders from hyperscalers and enterprises have become a headline growth driver. Exposure hinges on winning Ethernet-based AI cluster deployments against Arista and whitebox rivals, atop a large mature networking base.
Data center power demand from hyperscalers is a primary driver of NextEra's renewables and storage backlog and of its push to restart the Duane Arnold nuclear plant. The thesis still rests on a diversified regulated utility plus a broad independent power business, so AI-linked demand is one driver among several.
The Socrates projects make Williams one of the few midstream companies with directly contracted data center power revenue, on top of Transco expansions driven by power demand. The core pipeline business remains broad based gas transport, keeping AI a modest slice of cash flow.
AEP's load growth inflection is explicitly data-center-driven, with customer commitments for roughly 20 gigawatts of incremental demand across Ohio, Texas, and Indiana and the largest transmission network in the country to serve it. The regulated structure keeps AI-linked revenue modest, so the thesis depends on translating load growth into approved grid investment.
UMC's mature-node foundry business supplies the dozens of supporting chips inside every AI server, from power management to networking silicon, and its 12nm partnership with Intel adds a US-capacity angle. The thesis depends on overall semiconductor volume growth rather than direct AI accelerator demand, keeping direct exposure low.
EQT pairs the largest US gas production base with midstream reach into PJM, where data center power demand is concentrating, and has supply arrangements tied to new data center linked power projects. Commodity natural gas remains the core business, so AI is a demand tailwind rather than the revenue base.
Reddit earns advertising revenue and sells its conversation corpus through data-licensing deals with major AI labs. Licensing and AI-powered search are the incremental thesis, but the core remains an ad-supported social platform.
Tower runs a specialty foundry making analog, power management, and silicon photonics chips, including components used in data-center optical transceivers. AI-linked photonics and power chips are a growing driver, but diversified industrial and RF demand still dominates.
Belden supplies the cabling and connectivity that physically wire data centers, with management flagging hyperscale and AI facilities as a growth vector. The thesis depends mostly on industrial automation and enterprise network spending, keeping AI a modest though genuine slice.
Sells CPUs with AI-focused NPUs and Gaudi accelerators, and is building a foundry that could manufacture AI chips for others. Exposure is mostly optionality, since it has minimal accelerator share and the core PC and server business is under pressure.
Caterpillar's Energy & Transportation segment has posted record demand for reciprocating engines and gas turbines that management attributes to data center power. Construction and mining equipment dominate total revenue, so the thesis still rides global heavy industry cycles more than the buildout.
ABB sells industrial robots, electrification gear, and automation software, and its electrification unit supplies power equipment into data centers. Robotics and data center power are genuine growth drivers, but most revenue still comes from broad industrial demand.
Southern Copper produces copper at some of the lowest costs in the industry from Peruvian and Mexican mines. Data center and grid electrification demand supports the long-term copper story, but the business remains a broad commodity play, not an AI-specific one.
Prologis is converting parts of its land bank into powered data center developments with a multi-gigawatt pipeline. The program is strategically significant, but the thesis still depends on logistics rents and global goods flows, which generate essentially all of its revenue.
CoreSite gives American Tower genuine data center and interconnection exposure that benefits from AI and cloud traffic, but it sits inside a REIT dominated by communications towers. The thesis depends on mobile data growth and tower leasing economics far more than on the AI buildout.
Intuit sells tax and accounting software and is pushing Intuit Assist and AI agents across TurboTax and QuickBooks to automate filing and bookkeeping. AI is central to its product strategy and pricing power, but the core franchises would thrive without the buildout.
Atlassian sells Jira and Confluence subscriptions, layering Atlassian Intelligence and Rovo AI agents on top to lift seat prices and upsell. AI is becoming a monetization lever, but the thesis still rests on core developer-collaboration seat growth.
Oncor's Texas transmission buildout is a major pillar of Sempra's capital plan, driven by an ERCOT interconnection queue holding more than 100 gigawatts of requests that include large AI campuses. Diversification across California utilities and LNG infrastructure keeps AI-linked demand a minority driver of the overall business.
PSEG pairs New Jersey's largest regulated utility with the Salem and Hope Creek nuclear fleet, and has held publicized talks about supplying data centers directly from those units in the capacity-constrained PJM market. The thesis depends mostly on regulated utility growth today, with nuclear-to-data-center deals the optionality on top.
First Solar makes thin film modules for utility-scale solar plants, and hyperscaler power procurement for data centers supports its multi-year sold-out backlog. AI is one demand driver among utility, policy, and grid factors, so the thesis rests on broad solar deployment rather than the buildout alone.
Sells credit scores and ML-based decisioning software that lenders use for underwriting and fraud detection. Analytics is the product, but growth follows lending activity and score pricing rather than the AI infrastructure buildout.
Sells access to trapped-ion quantum computers through cloud providers and research contracts, pitched partly at future AI and optimization workloads. Exposure is speculative optionality, since useful quantum advantage for AI remains years away.
Rents cloud compute to developers and startups, adding GPU droplets and managed AI infrastructure after acquiring Paperspace. AI compute is its fastest-growing segment, but the core remains general-purpose SMB cloud hosting.
Quantinuum sells trapped-ion quantum computers, cloud access to them, and quantum software to enterprises, governments, and research institutions. Its AI exposure is optionality rather than demand: the machines are pitched at future AI and optimization workloads, so the connection depends on quantum advantage arriving for problems classical hardware cannot already handle.
Figma sells collaborative design software subscriptions and is embedding generative AI for design creation and prototyping across the platform. AI is both its biggest product opportunity and a competitive threat from prompt-to-app tools, so exposure is strategic rather than revenue-led today.
Element Solutions supplies plating chemistry for the copper pillar and wafer-level packaging in AI accelerators plus materials for high-layer data center boards. Industrial and automotive end markets still dilute the mix, so the thesis depends on broad electronics production, not just AI.
IDACORP is a small utility where two AI linked customers, Meta's Kuna data center and Micron's Boise fab expansion, visibly move total load, driving record growth and capex. The regulated model caps upside, but proportional exposure is higher than most large-cap peers.
Netskope sells cloud security and SASE subscriptions, with SkopeAI positioned to govern and secure enterprise use of AI applications and data. Securing AI usage is a genuine growth vector, but the core thesis remains network security displacement of legacy appliances.
EPAM sells engineering and consulting services to enterprises building generative AI applications, capturing spend in the AI software layer. AI projects are its clearest growth engine after a services slowdown, but most revenue remains general digital transformation work.
Portland General Electric already carries one of the highest existing data center load shares among US utilities, anchored by the Hillsboro cluster. Regulated returns cap the upside from that demand, keeping the score at the low end of the growing band.
SOLV Energy is one of the largest utility-scale solar and battery storage contractors in the country, building projects mostly for utilities and independent power producers that are racing to add capacity as electricity demand rises, including from AI data centers. Exposure is indirect and depends on how much of its multi-year backlog is ultimately driven by data center load growth rather than the broader renewable buildout it has served for over a decade.
MARA mines bitcoin and is converting part of its power and data center footprint into AI and HPC hosting capacity sold to compute tenants. Exposure hinges on how much capacity actually shifts to AI leases, since bitcoin still drives nearly all revenue.
USA Rare Earth is building a US mine-to-magnet chain whose NdFeB magnets feed hard drives, data center cooling, and robotics, backed by major federal funding. It is early in commercial ramp, so the thesis depends on executing Stillwater and Round Top, not current AI sales.
TransAlta has a concrete, named data center power arrangement at its Keephills site with CPP Investments and Brookfield, starting at roughly 230 megawatts with room to scale toward a gigawatt. It remains an early stage opportunity layered on a diversified merchant power fleet.
Sells quantum-inspired optimization software and photonic computing chips, marketed toward AI, logistics, and security workloads. Revenue is minimal and the AI connection is largely speculative positioning rather than proven demand.
Photronics supplies the mask sets every chip design requires, so AI driven design activity lifts high end demand. The thesis depends on overall design starts, with revenue weighted to mature nodes and display masks rather than AI specific work.
SkyWater is a US foundry fabricating specialty chips, including radiation-hardened processors and advanced packaging aimed at defense and AI-adjacent computing. Exposure grows with domestic chip reshoring and defense AI programs, but most current volume is not AI-specific.
Named hyperscaler projects in NIPSCO's northern Indiana territory, including announced AWS and Google campuses, make the data center linkage concrete, and that load is a stated pillar of its growth plan. The regulated gas and electric base keeps the AI linked share of the business modest.
Concrete NovaLT turbine orders for data center power make Baker Hughes' AI linkage real rather than thematic. Oilfield services and LNG equipment still dominate revenue, so the thesis depends mostly on traditional energy spending.
Sells cloud-delivered zero trust security subscriptions with AI-based threat detection and risk scoring to enterprises. AI enhances the product and pitch, but growth is driven by network security transformation, not the AI buildout.
Teck is now a copper-focused miner after exiting coal, feeding electrification and data center demand. AI supports the long-run copper deficit story, but the stock trades on commodity prices and its Anglo American merger, not buildout contracts.
DTE serves Michigan territory that has attracted large data center projects, including an AI campus at Saline Township, and it has raised its capital plans around that load growth. Its earnings still depend on regulated rate base and on Michigan electricity and gas demand overall.
PPL's Pennsylvania territory sits in the middle of the PJM data center boom, with tens of gigawatts of interconnection requests disclosed and a joint venture with Blackstone Infrastructure to build gas generation for data centers. The thesis still rests on regulated rate-base growth, which keeps the AI linked share of earnings modest for now.
Named wins from Google, Meta and Panasonic plus a disclosed multi gigawatt large-load queue anchor Evergy's growth story and its largest ever capital plan. The regulated model keeps the AI linked share of revenue in the single digits for now.
Lattice sells low-power FPGAs used for control, security, and sensor processing in servers, edge devices, and industrial systems. AI server attach and edge AI inference are growth vectors, but demand still swings with broad communications and industrial cycles.
Li Auto sells extended-range and battery EVs, bundling its AD Max autonomous driving stack and Mind GPT assistant into the vehicles. AI driving capability is a key differentiator, but value still rests on vehicle volumes and margins in China.
JFrog sells software supply chain tools into the AI software layer, adding ML model registry and AI model scanning to Artifactory for enterprise DevOps teams. Exposure hinges on MLOps becoming a paid expansion driver, but the core artifact management business is not AI dependent.
Varonis sells data security software that classifies sensitive data and detects threats, increasingly positioned as the prerequisite for safe enterprise AI rollouts like Copilot. That AI readiness angle is a real demand driver, though the base business is conventional data governance.
Materion's deposition targets, specialty alloys, and clad metals flow into semiconductor fabrication, its largest end market, with AI adding demand. Defense, aerospace, and industrial markets dominate the rest, so the thesis depends on diversified precision materials demand.
Xanadu sells photonic quantum computers, cloud access to them, and the PennyLane software developers use to build quantum applications, including machine learning workloads. Its exposure depends on quantum advantage arriving for AI relevant problems before its cash runs out, the same forward looking bet carried by its quantum computing peers.
Infleqtion sells full-stack neutral atom quantum computers, quantum sensing and clock systems, and the Superstaq software platform mainly to government and enterprise research customers such as defense agencies and national labs. Its AI exposure is speculative optionality tied to whether quantum computing eventually becomes useful for AI workloads, the same basis on which the universe's other quantum computing pure plays are scored.
Pasqal designs and manufactures neutral atom quantum computers and sells cloud access to them, with customers including Aramco, CMA CGM, OVHcloud, Thales, and Sumitomo using its systems mainly for research and industrial computing rather than AI workloads specifically. Its AI exposure is speculative optionality tied to whether quantum machine learning becomes commercially useful, the same basis on which the universe's other quantum computing pure plays are scored.
Georgia Power sits in one of the country's biggest data center growth markets, with tens of gigawatts of requests in its interconnection pipeline and the newly completed Vogtle nuclear units prized by hyperscalers. The thesis still depends on regulated rate-base growth across a large diversified utility, where AI load is an accelerant rather than the core business.
Data center driven power generation is a named driver of Kinder Morgan's project backlog, with sanctioned expansions in Texas and the Southeast tied to that demand. Its vast legacy transport base keeps the AI-linked share of the business small.
WEC's Wisconsin territory hosts Microsoft's Mount Pleasant AI campus and other announced data center projects, which anchor its expanded capital plan. The thesis depends on regulated rate-base growth, and data center load is still a modest slice of a broad utility.
Announced hyperscaler projects in its Iowa and Wisconsin territory, including Google in Cedar Rapids, are the stated driver of Alliant's raised capital and load forecasts. As a smaller regulated utility, the exposure is real but capped by the rate-regulated model.
Phoenix is one of the largest US data center markets, and large-load customers, including the TSMC semiconductor ecosystem and hyperscaler campuses, dominate APS's disclosed growth pipeline. The regulated structure keeps the AI linked slice of earnings modest despite strong load growth.
Valmont's utility structures are a picks-and-shovels play on grid expansion for data center load, and management cites data centers as a demand driver. Agriculture and infrastructure segments dominate the company, so the thesis depends mostly on irrigation and infrastructure cycles rather than the buildout.
OGE Energy serves Oklahoma territory where data center projects have added large load requests to its interconnection pipeline, supporting new generation and grid spending. Its earnings are set by regulated rates and regional demand rather than by AI spending directly.
Clearway is signing power purchase and co-location agreements with hyperscaler and data center customers across its renewables and flexible gas fleet, and highlights that channel as a core growth driver. The activity is real but early, on top of a broadly contracted clean power base.
Black Hills has an outsized, well documented data center niche in Cheyenne, Wyoming, where Microsoft campuses and new AI projects make data centers its fastest growing load segment. The small scale and regulated model keep the overall score in the peripheral band.
Duke has signed agreements with Amazon, Google, Microsoft, and Nucor to accelerate carbon-free generation for data center load in the Carolinas, where its interconnection queue includes large hyperscaler campuses. The thesis depends on regulated rate-base growth across a very large utility base, with AI-linked demand still a small share of revenue.
Energy Transfer signed the first major midstream gas supply deal for a dedicated data center and discloses a large pipeline of similar inquiries. Its massive diversified midstream base means AI-linked demand remains a small slice of the total business.
Exelon's wires-only utilities, led by ComEd in the Chicago area, have disclosed double-digit gigawatts of data center interconnection requests that are lifting transmission capex plans. The thesis depends on regulated delivery rates rather than power sales, so AI load raises grid investment without selling energy directly to hyperscalers.
Data center linked gas power demand is a disclosed driver of DT Midstream's new expansion projects in Appalachia and the Midwest. Its contracted pipeline model is steady but limits how much direct AI revenue can move the needle.
Sells firewalls and security subscriptions with FortiAI and ML-driven threat intelligence to enterprises. AI is embedded product technology, and growth follows network security spending rather than AI infrastructure demand.
Xcel has disclosed a multi-gigawatt data center pipeline across its Minnesota, Colorado, and Texas panhandle territories and raised its capital plan on that load growth. The thesis depends on conventional regulated rate-base economics, with AI demand supporting growth without changing the business model.
The Trade Desk sells programmatic ad buying, with its Kokai AI engine optimizing bids and audiences for advertisers. AI is core to its product edge, but revenue tracks digital ad spend rather than AI infrastructure demand.
Sells IT services and outsourcing, using its Topaz platform to deliver generative AI projects for enterprise clients. AI is a growing share of deal flow, but revenue depends on overall enterprise IT spending, and AI could deflate traditional service work.
Sells electric trucks and SUVs while developing its own autonomy platform and software stack, partly funded by its Volkswagen software joint venture. Autonomy and software are future thesis points, while current revenue is vehicle sales.
BCE is building Bell AI Fabric, a set of Canadian data centers offering sovereign GPU capacity to domestic customers. That buildout is small next to a telecom business whose results depend on Canadian wireless and broadband competition.
Nutanix sells hybrid cloud infrastructure software, with GPT-in-a-Box packaging GPU and AI workload support on its platform. AI is an emerging attach opportunity, but growth is driven mainly by VMware displacement and core cloud platform renewals.
CDW resells and integrates IT hardware and services, helping enterprises procure GPU servers, AI networking, and related infrastructure. AI-driven hardware refresh is a genuine tailwind, but it is one demand stream inside a broad, low-margin distribution business.
Leidos sells IT modernization, intelligence analysis, and autonomous systems to US defense and civilian agencies, embedding AI into large government contracts. Exposure hinges on Pentagon AI budgets growing, but the bulk of the business is conventional government services.
PTC sells CAD and PLM software to manufacturers, adding AI to generative design, IoT analytics, and product data management. AI strengthens the portfolio, but growth is driven by digital engineering budgets rather than AI infrastructure spending.
CACI sells AI enabled signals intelligence, electronic warfare, and cyber tools to US defense and intelligence agencies. Growing military AI budgets are a real tailwind, though revenue is spread across many conventional government technology programs.
SailPoint sells identity security software using AI to automate access recommendations and governance decisions for enterprises. Securing machine and AI agent identities is an emerging demand driver, but the core business is established identity governance spend.
Sells a DevSecOps platform with the GitLab Duo AI add-on for code suggestions, security explanations, and reviews. AI is an upsell on developer seats, and AI coding tools are both an opportunity and a competitive threat.
Ormat's 24/7 carbon free geothermal power and growing storage business fit exactly what hyperscalers want for data centers, and management cites data center driven load growth in the western US as a demand driver. Direct contracted data center revenue is still small, keeping the exposure peripheral with optionality.
Sells superconducting quantum processors and cloud access, mainly to government and research customers, with AI positioned as a future workload. Revenue is tiny and the AI link is speculative optionality rather than a current driver.
Parsons sells engineering and technology services to defense and intelligence customers, including AI for cyber operations and missile defense. AI-enabled contracts are a growing slice, but broad government infrastructure and defense spending drives the business.
Box sells enterprise content management seats and monetizes Box AI add-ons that extract answers and metadata from stored documents. The thesis is that AI on top of a company's content raises price per seat, though the underlying business is steady content storage and governance.
IQM sells full-stack superconducting quantum computers and cloud access mainly to national labs, universities, and supercomputing centers, not directly to AI buyers. The AI link is speculative optionality, resting on quantum computing eventually becoming useful for AI-scale optimization and machine learning problems.
Axcelis sells ion implantation equipment to chip fabs, with strength in silicon carbide power devices and mature-node logic. AI benefits it indirectly through power semiconductors and advanced logic capacity, but implant demand is mostly tied to EV and general chip cycles.
Cellebrite sells digital forensics and investigative analytics software to law enforcement, using AI to extract and analyze evidence from devices. AI accelerates investigations and supports pricing, but the value rests on government forensics budgets, not the buildout.
Plug Power markets stationary fuel cell power to data center operators seeking fast, clean capacity, but that pipeline is early stage while material handling and hydrogen remain the actual business. The thesis depends on hydrogen economics improving, with AI data center power more aspiration than revenue today.
AvePoint sells data governance and management software for Microsoft 365, positioning itself as the data-readiness layer for Copilot and enterprise AI deployments. Exposure hinges on AI adoption forcing companies to clean and govern data, a real but still minority tailwind.
Rogers' curamik substrates sit inside the power semiconductor modules used in data center power systems, EVs, and grid equipment scaling with AI electricity demand. EV and industrial markets dominate today, so the thesis depends on power electronics adoption broadly.
VIAVI sells test instruments and optical components for networks, with growing demand from testing 800G links and data center interconnects. AI networking buildout is a tailwind for its lab and production test lines, but telecom carrier spending still dominates results.
AIRO sells military drones, avionics, and training services to defense customers, with autonomy software embedded in its unmanned systems. Exposure hinges on autonomous drone procurement accelerating, but it remains a small diversified defense supplier rather than an AI pure play.
Vertical Aerospace develops the VX4 eVTOL aircraft, with AI in flight control, navigation, and design optimization. The thesis is electric aviation certification and orders, with autonomy a longer-term option rather than the current driver.
TELUS operates sovereign AI compute capacity in Canada and rents GPU capacity to domestic customers that need data kept in country. The AI business is a small line next to wireless, broadband, and its health and agriculture units.
CenterPoint has disclosed a multi gigawatt pipeline of data center and industrial electrification load in the Houston area, supporting a record capital plan. The thesis depends on regulated rate-base growth in ERCOT's fastest growing metro, and the AI linked share remains small and early.
Sells automotive and industrial chips, including S32 processors and i.MX parts that run AI at the edge in cars and devices. Edge AI is a growing content story, but demand follows auto and industrial cycles.
Keysight sells test and measurement systems used to validate AI data center interconnects, high-speed chips, and networks. AI and data center testing is a growing driver, but revenue spans broad electronics, telecom, and aerospace demand.
Siemens Healthineers sells imaging scanners, radiotherapy systems, and lab diagnostics, with AI-Rad Companion and AI treatment planning embedded in the workflow. AI differentiates the equipment, while revenue is driven by hospital capital spending, not the buildout.
Elbit sells defense electronics, UAVs, and electronic warfare systems with growing AI autonomy and battlefield-intelligence content. Backlog growth is driven by global rearmament, with AI an enabler inside the products rather than the demand source.
Teledyne's sensors and cameras feed semiconductor inspection tools and AI-enabled vision systems in defense and industry, giving it real but minority exposure to the buildout. The thesis depends on broad instrumentation, aerospace, and defense spending rather than AI demand itself.
PG&E has disclosed a multi gigawatt pipeline of data center interconnection applications in Northern California and a program under which new large loads help offset bills for other customers. The thesis depends on wildfire safety spending and regulated grid investment, with data centers a small slice of a very large utility.
Lynas produces separated rare earths at scale outside China, feeding the magnets inside drives, cooling systems, and motors. Most demand still comes from EVs, wind, and defense, so the thesis depends on rare earth prices and Western supply policy more than AI.
NIO sells premium EVs featuring its NOMI assistant, autonomous driving stack, and in-house driving chips. AI capability shapes the product, but the investment case hinges on vehicle deliveries and path to profitability, not AI demand itself.
SAIC integrates AI and digital engineering into US government systems through platforms like Tenjin, billing on federal contracts. AI raises win rates and margins on modernization work, but the company remains a broad government services integrator.
Horizon Quantum sells compilers and developer tools that let programmers write quantum software for different hardware, and it now owns and operates its own quantum computer testbed rather than reselling someone else's access. Its AI exposure is speculative optionality tied to whether quantum computing eventually becomes useful for AI workloads, scored slightly below its hardware-owning peers since its core product is still the software layer rather than a machine it sells directly.
Linde supplies ultra-high-purity gases to semiconductor fabs under long-term on-site contracts, and AI-driven fab construction keeps adding wins. Electronics is a modest part of a very diversified gas business, so the thesis depends on industrial gas volumes and pricing overall.
Uses AI for pricing, matching, and routing in its rides and delivery marketplace, and partners with Waymo and others on robotaxis. Autonomy is optionality that could cut driver costs or disrupt the model, not a current revenue driver.
Air Liquide's Electronics division supplies carrier gases, precursors, and advanced materials to leading-edge fabs, with AI-driven construction its fastest growth vector. Electronics is a small share of a diversified gas group, so the thesis depends on broad industrial gas demand.
Rheinmetall sells ammunition, vehicles, and defense electronics, embedding AI in autonomous ground vehicles, drone control, and situational awareness. European rearmament, not AI, drives the growth, with AI content a rising but secondary element.
F5 sells application delivery and security products, adding AI gateways, bot defense, and API protection for AI-era traffic. AI workloads create new demand for its load balancing and security, but the core is mature enterprise networking.
Rubrik sells data security and backup subscriptions, using AI for ransomware recovery, data classification, and its Ruby assistant. It also benefits from enterprises needing to secure data feeding AI systems, but the core driver is cyber resilience spend, not the buildout.
BWXT manufactures naval nuclear reactors and fuel for the US government, with advanced reactor and microreactor programs offering data center power optionality. AI matters mainly through future nuclear demand for compute power, while the defense base is independent of AI.
Arrow distributes electronic components and enterprise IT, sourcing AI chips and helping customers deploy edge and data center AI systems. AI lifts volumes through the components cycle, but thin-margin distribution economics keep exposure modest.
Globus Medical sells spine implants and the ExcelsiusGPS robotic navigation system to hospitals for AI-assisted surgery. Robotics adoption drives implant pull-through and is a genuine growth lever, but most revenue remains conventional musculoskeletal hardware.
Joby builds electric air taxis with autonomy software as a longer term layer on top of piloted operations. Near term value hinges on FAA certification and manufacturing, with AI powered autonomous flight a later stage option rather than the current driver.
Impinj sells RAIN RFID chips and readers that tag physical items, feeding inventory data to retail and logistics customers. AI-driven supply chain analytics is a use case for that data, but demand is tied to retail tagging adoption, not AI compute.
OSI Systems sells security screening equipment with AI threat detection for airports and cargo, plus patient monitoring hardware. AI algorithms differentiate its Rapiscan systems, but demand follows security and healthcare procurement budgets rather than AI infrastructure.
Power Integrations sells high-voltage power conversion chips, mostly into appliances, phones, and industrial gear, with newer GaN parts aimed at AI server power supplies. Data center power is a growth option, not yet a major revenue contributor.
Standard Nuclear manufactures TRISO fuel and supplies it under a multi-year agreement to Radiant Industries for its Kaleidos portable microreactor, which is marketed toward data centers alongside several other customer types, plus separate fuel work for the Department of Defense. Its AI exposure hinges on how much of the broader advanced reactor demand it eventually serves is actually driven by AI data center load growth rather than defense, industrial, or remote power use.
Nomad Power Solutions sells transportable battery storage units to a customer base that includes AI data centers alongside utilities, industrial sites, government agencies, and disaster response, with fiscal 2025 revenue of 9.4 million dollars and a growing deferred revenue balance. Exposure hinges on how much of its future order book comes specifically from data center customers, since no named AI data center contract has been disclosed yet.
Stryker sells orthopedic implants and medical devices, with the Mako surgical robot and AI planning software a growing franchise. Surgical robotics is a real growth vector, but most revenue comes from conventional devices and hospital equipment.
Emerson sells process automation systems and industrial software, including AspenTech, with AI applied to plant optimization. It benefits indirectly from power and LNG projects tied to data centers, but core demand comes from traditional process industries.
Air Products supplies on-site and specialty gases to leading-edge fabs, and new AI-era fab construction supports its electronics backlog. Electronics remains a minority of a diversified gas and hydrogen company, so the thesis depends on industrial demand and its hydrogen strategy.
Rockwell sells factory automation hardware and software, embedding AI in FactoryTalk, predictive maintenance, and digital twins. It benefits from reshoring and smarter factories, but demand tracks industrial capex cycles more than AI specifically.
Wipro sells IT outsourcing and consulting with AI woven in through its ai360 initiative, monetizing enterprise AI adoption projects. AI is a growth talking point rather than the driver, and the vast majority of revenue is traditional IT services.
Sells power semiconductors and image sensors into cars and industrial systems, enabling ADAS sensing and electrification. Exposure comes from vehicle intelligence content growth, while revenue is driven by auto and industrial demand cycles.
Samsara sells connected operations subscriptions, using AI video analytics on dash cams and equipment data to reduce accidents and downtime. AI is core to the product's value proposition, but customers buy fleet safety and operations outcomes, not AI infrastructure.
Archer sells the Midnight eVTOL for urban air taxi routes, with AI autonomy planned for future generations of the aircraft. The investment case rests on certification, production, and launch partners, so AI exposure today is mostly optionality.
Kyndryl sells IT infrastructure management services, using its Bridge AIOps platform and AI consulting signings to modernize enterprise estates. AI-related services are a growing signings mix, but legacy managed infrastructure still dominates revenue.
Qualys sells cloud based vulnerability management and compliance subscriptions, using AI to prioritize risk through TruRisk. AI improves prioritization inside an established security workflow, so exposure is a feature level story rather than a driver of the thesis.
Tenable sells exposure management software that scans and prioritizes cyber risk, with ExposureAI adding generative guidance for analysts. AI enhances the product and AI systems add new attack surface, but the core vulnerability assessment business drives the value.
Rapid7 sells security analytics and managed detection, using AI for threat detection and vulnerability prioritization. AI features keep it competitive in the SOC market, but the investment case rests on security spending generally, not the AI buildout.
BHP frames data center and grid electrification demand from AI as a core driver for its copper growth assets like Escondida and Copper South Australia. Iron ore still dominates earnings, so the thesis depends on steel markets and commodity cycles far more than the buildout.
Sells e-commerce software and payments to merchants, bundling AI tools like Sidekick and Magic for content and support. AI improves the product, but growth tracks merchant sales volume, not the AI buildout.
Rio Tinto's copper ramp at Oyu Tolgoi, its aluminum business, and its new lithium arm all feed AI-era electrification demand. Iron ore remains the earnings engine, so the thesis depends on bulk commodity markets rather than data center growth.
Sells fighter jets, missiles, and defense systems, embedding AI in autonomy, targeting, and intelligence programs for the Pentagon. AI is a capability layer inside huge hardware programs, not a distinct revenue driver.
Glencore is a top-tier copper and cobalt producer whose output and recycling business are levered to AI data center and grid electrification demand. Coal and the trading arm still drive much of earnings, so the thesis depends on broad commodity cycles.
Sells bombers, satellites, missile defense, and autonomous aircraft, with AI embedded in battle management and unmanned systems. Autonomy provides real exposure, but revenue is driven by large defense hardware programs.
Honeywell sells aerospace, building, and industrial automation products, with Forge software and warehouse automation as its AI-adjacent lines. Automation is one growth pillar among many, so buildout exposure is real but clearly a minority of the thesis.
Merck KGaA's Electronics unit supplies deposition materials, CMP slurries, and gas delivery systems for leading-edge chipmaking tied to AI node transitions. Healthcare and life science dominate the group, so the thesis depends on pharma and lab demand more than the buildout.
Anglo American is centering its portfolio on copper through assets like Collahuasi and Quellaveco and a pending merger with Teck to form Anglo Teck. Copper's AI-linked demand is still a minority earnings driver, so the thesis depends on diversified commodity markets and merger execution.
Sells genetic tests like Signatera and Panorama, using ML algorithms to analyze cell-free DNA for cancer and prenatal screening. AI is core to assay accuracy, but growth is driven by diagnostic test volumes, not the AI buildout.
Dassault Systemes sells 3D design and simulation software, adding generative design and AI-driven virtual twins across CATIA and SIMULIA. AI enriches an engineering software franchise, but license and subscription growth from manufacturing R&D drives the thesis.
Sells communications APIs and customer data tools, adding AI assistants and predictive features for engagement workflows. AI is a product enhancement, with revenue driven by messaging and voice usage volumes.
Madison Air Solutions sells indoor air quality and airflow equipment across residential, commercial, and industrial markets, with a dedicated Nortek Data Center Cooling line serving data center customers. AI exposure is limited to that one product line inside a much larger business built on HVAC, filtration, and ventilation demand that has little to do with AI.
Align sells Invisalign aligners and iTero scanners, using AI in treatment planning and smile simulation. AI improves clinical workflow and conversion, but revenue depends on consumer orthodontic demand, leaving AI as a supporting tool.
Unity sells its real-time 3D engine and ad monetization tools to game developers, with AI in asset creation, its Vector ad models, and digital twins. AI improves both engine and ads businesses, but the thesis is gaming ecosystem recovery, not the buildout.
Dropbox sells cloud storage and collaboration subscriptions, with Dash adding AI universal search across a user's tools. Dash is the main growth bet on a mature core, so AI matters to the story, but current value rests on the legacy storage base.
Fastly sells edge delivery, security, and compute, positioning its network for AI inference caching through the AI Accelerator. AI at the edge is early and small, so exposure is mostly optionality on top of a competitive CDN business.
Redwire sells space infrastructure including satellite components, robotic assembly systems, and Earth observation analytics to government and commercial customers. AI and robotics are embedded capabilities, but the thesis is space and defense budgets, not the AI buildout.
Twist sells synthetic DNA to biotech, pharma, and antibody discovery customers, using ML to guide library and antibody design. Exposure comes partly from AI driven drug discovery demand for DNA and long shot DNA data storage, but the core is a synthesis tools business.
Voyager sells defense and space technology and leads the Starlab commercial space station program with NASA support. AI features in its defense systems, but the thesis is space station execution and defense budgets, not the buildout.
Genius Sports sells official sports data, betting analytics, and computer vision player tracking to sportsbooks and leagues. ML powers its products, but revenue is driven by sports betting growth, making AI a capability rather than the thesis.
Wolfspeed makes silicon carbide power semiconductors, mostly for EV drivetrains, with some relevance to power-dense AI data center supplies. Its fate hinges on EV demand and its own balance sheet restructuring, with AI only a peripheral demand source.
Johnson & Johnson sells pharmaceuticals and medical devices, developing the Ottava surgical robot and applying AI to drug discovery and medtech analytics. Surgical robotics gives real but small exposure, and the thesis rests on drug and device franchises, not AI.
Analog Devices sells analog and mixed-signal chips into industrial, automotive, and communications markets, including power and connectivity parts that touch data centers and edge AI. Broad industrial cycles drive the business, with AI a modest but growing slice.
Motorola Solutions sells public safety communications, video security, and command center software, with AI in video analytics and alerting. Government radio and security demand drives the business, so AI is a product enhancer with minor buildout exposure.
Nasdaq earns exchange fees and sells financial software, with Verafin anti-fraud and AI market surveillance as growth products. AI strengthens its fintech segment, but listings, trading, and index revenue drive most of the value.
Autodesk sells design and construction software subscriptions, adding generative AI for CAD, BIM, and project intelligence. AI defends and extends the franchise, but customers buy design tools regardless, so buildout exposure stays minor.
Sells rocket launches, satellites, and space systems, using AI mainly for mission software and autonomous flight operations. Its growth is driven by space and defense demand, with AI a supporting technology.
Roblox monetizes virtual currency and ads on its gaming platform, using generative AI for content creation tools and moderation at scale. AI lowers creation friction and costs, but the thesis is user growth and monetization, not the buildout.
Sells identity and access management subscriptions, using AI for threat detection and access decisions, and beginning to secure AI agent identities. AI is a feature and an emerging niche, not the growth engine.
Penumbra sells thrombectomy devices for stroke and clot removal, with AI in imaging analysis and an emerging robotics program. Revenue depends on procedure volumes and device adoption, so AI and robotics are supporting features, not the thesis.
Generac sells backup generators and home energy systems, applying AI to grid services and predictive maintenance. There is optionality in power resiliency demand around strained grids, but core results ride weather, housing, and outage cycles rather than AI.
Hudbay produces copper with gold and zinc by-products across Peru and Canada, with the Copper World project set to add US supply. AI matters only as one driver of broad electrification demand, so the thesis depends on copper prices and mine execution, not the buildout.
Snap sells ads on Snapchat, using AI for AR lenses, content ranking, and its My AI chatbot. AI raises engagement and ad efficiency but also inflates infrastructure costs, and the thesis remains ad market share, not the buildout.
Silicon Labs sells low power wireless SoCs for IoT devices, adding ML inference blocks for on-device intelligence. Edge AI is a feature that helps win sockets, while the business is driven by smart home and industrial IoT demand cycles.
Karman supplies missile, hypersonics, and space propulsion components to defense primes. Its growth rides munitions restocking and hypersonics budgets, where AI-enabled weapons are a demand backdrop rather than what Karman itself sells.
Ero offers close to pure copper exposure from its Brazilian mines, with the Tucuma ramp driving growth. AI data center and grid demand support the copper market broadly, but the thesis depends on copper prices and operational delivery, not AI itself.
NetScout sells network performance monitoring and DDoS protection appliances, applying AI to threat detection and analytics. AI sharpens the product, but the business tracks carrier and enterprise network spending rather than the AI buildout.
Calix sells software platforms and access equipment to broadband service providers, with AI features in marketing and support clouds. Its growth follows rural broadband and fiber spending, so AI is a product garnish rather than a thesis driver.
10x Genomics sells single cell and spatial biology instruments and consumables to researchers, with AI powering the analysis software. Value tracks life science tools demand and instrument placements, with AI as an enabling feature rather than the driver.
Ferroglobe produces silicon metal, the upstream feedstock for polysilicon and semiconductor-grade silicon, with strategic value as a Western supplier. Most volume goes to aluminum and chemical customers, so the thesis depends on commodity silicon prices, not AI demand.
N-able sells IT management, backup, and security software to managed service providers, adding AI for threat detection and endpoint automation. Growth tracks MSP seat expansion, so AI features defend the product rather than drive the thesis.
OneSpan sells authentication and e-signature software to banks, using AI for fraud detection and biometric verification. AI strengthens the security product, but the value rests on banking authentication contracts rather than the AI buildout.
Varex sells X-ray tubes, detectors, and imaging components to medical imaging OEMs and industrial inspection customers. AI-enhanced imaging software is a feature, but revenue tracks medical equipment and cargo screening demand, keeping AI exposure minor.
Eli Lilly sells pharmaceuticals, led by GLP-1 drugs, and applies AI to discovery, trial design, and manufacturing. AI may speed the pipeline and it has partnered with AI drug discovery firms, but the value driver is drug demand, not AI.
Sells aerospace engines, missiles, and defense electronics, applying AI to radar, guidance, and predictive maintenance. AI exposure is embedded within broader defense programs and remains peripheral to the revenue base.
Sells analog and embedded chips into industrial, automotive, and consumer markets, including edge processors for vision and sensing. AI is a small application niche, with demand driven by broad industrial cycles.
Vale's copper and nickel output from Vale Base Metals feeds grid buildout and the storage batteries tied to data center power demand. Iron ore is the overwhelming majority of revenue, so the thesis depends on steel demand, not AI.
Microchip sells microcontrollers and analog chips into industrial, auto, and IoT markets, offering edge AI development tools. Its cycle is driven by broad embedded demand, with AI a small emerging attach rather than a driver.
Tradeweb operates electronic trading marketplaces for rates, credit, and money markets, applying AI to pricing and execution analytics. AI refines the platform, but revenue is driven by trading volumes and electronification of fixed income, not AI demand.
Pembina is building the Greenlight Electricity Centre, a dedicated natural gas power plant that will supply Meta's Sturgeon County AI data center under a long term tolling agreement, developed jointly with Morgan Stanley Infrastructure Partners and Kineticor Asset Management. That project is a real, named commitment to the buildout, but it is a small piece of a diversified midstream business built on pipelines, gas processing, and marketing across Western Canada that has nothing to do with AI.
Gen Digital sells consumer security subscriptions under Norton, Avast, and LifeLock, using AI for threat and scam detection. AI both powers its products and creates new scam threats to defend against, but exposure to the buildout itself is minor.
Albemarle produces lithium for batteries, including grid-scale storage increasingly co-located with data centers. The business is overwhelmingly tied to EV demand and lithium prices, so AI exposure is a small emerging slice.
Alcoa smelts the aluminum used in transmission lines and data center construction, both stretched by AI power demand. That demand is a small slice of a cyclical commodity business, so the thesis depends on global aluminum prices and smelting economics.
Tetra Tech sells environmental and water consulting services to governments and industry, applying AI to water modeling and climate analytics. Data center water and power permitting work offers some tailwind, but AI is marginal to a government-services-driven thesis.
Workiva sells compliance and financial reporting SaaS, adding generative AI for document drafting and data linking. AI features aid retention and upsell, while regulatory reporting demand drives the business, not the buildout.
Diodes sells commodity analog, discrete, and power management chips into consumer, automotive, and industrial markets. Some parts land in edge AI and server boards, but exposure is diffuse and AI does not meaningfully set the growth rate.
Yelp monetizes local business advertising, using AI for recommendations, review analysis, and ad targeting. AI is an internal efficiency tool while generative search threatens its traffic, so buildout exposure is peripheral and partly negative.
Workhorse sells electric last-mile delivery vans and drone systems, with software for route optimization and fleet management. The thesis is commercial EV adoption and survival, with AI and autonomy peripheral options.
Chevron is an integrated oil and gas major that signed a 20 year, multibillion dollar gas supply and power agreement to build a dedicated complex fueling a Microsoft AI data center in West Texas. That project is a real, named commitment to the buildout, but it is a small piece of a company whose revenue overwhelmingly comes from oil and gas production, refining, and marketing that have nothing to do with AI.
AstraZeneca sells pharmaceuticals and uses AI in drug discovery, trial design, and precision medicine to improve R&D productivity. AI is a tool that may raise pipeline odds over time, but value is driven by drug launches and patents, not AI demand.
Walmart sells groceries and general merchandise, using AI for inventory forecasting, search, and supply chain automation. AI improves margins as a buyer of the technology, so buildout exposure is operational rather than a revenue driver.
Robinhood earns trading, interest, and subscription revenue from retail investors, layering AI features like Cortex onto the platform. Growth depends on retail engagement and asset gathering, keeping AI exposure peripheral.
Sells submarines, combat vehicles, business jets, and IT services, with AI applied inside GDIT programs and combat systems. AI exposure is peripheral, since revenue depends on major defense platform budgets.
Sells payments and financial services through Square and Cash App, using AI for fraud detection, underwriting, and support. AI is an internal efficiency tool, not a revenue line tied to the buildout.
Illumina sells gene sequencers and consumables, with its DRAGEN AI software accelerating genomic analysis. Revenue depends on sequencing volumes and instrument cycles, so AI is an enabling feature rather than a buildout thesis.
DexCom sells continuous glucose monitors, with AI algorithms for glucose prediction and dosing insights. Revenue depends on sensor adoption and reimbursement, so AI is a product feature with peripheral buildout exposure.
Viasat sells satellite bandwidth for government, aviation, and consumer broadband, using AI internally for network and bandwidth optimization. AI improves how it runs the network, but demand for its capacity has little to do with the AI buildout.
Firefly builds Alpha rockets and Blue Ghost lunar landers, selling launch and lander missions to NASA and commercial customers. It is a space access story with only peripheral connection to AI or robotics demand.