AI Infrastructure Stocks

View APP profile

Role

AppLovin operates an end-to-end AI-powered advertising platform whose Axon recommendation engine runs real-time machine-learning inference at massive scale to match advertiser demand with publisher ad supply.

Inclusion rationale

Per its FY2025 Form 10-K, substantially all of AppLovin's revenue comes from Axon Ads Manager, powered by the Axon AI recommendation engine that performs real-time machine-learning predictions across vast ad auctions at microsecond speeds, making it a public AI-platform/software whose AI products generate sustained inference and compute demand as of 2026-06.

Confidence

medium

Sector

Technology

50D distance

-30%

52-week-high proximity

40%

A/D

E

RVOL

0.6x

50D dollar volume

$2.7B

View CRM profile

Role

Salesforce builds the Agentforce 360 enterprise AI agent platform layered on its CRM and Data 360 stack, whose autonomous agents and generative features generate large-scale LLM inference and compute demand.

Inclusion rationale

Per Salesforce's Q4 FY2026 results, Agentforce ARR reached $800M (up 169% Y/Y) across 29,000+ deals, and the platform has processed more than 19 trillion tokens to date (up 5x Y/Y) via its LLM gateway, positioning Salesforce as a public AI-platform vendor whose deployed agents directly create inference and compute consumption as of 2026-06.

Confidence

high

Sector

Technology

50D distance

19%

52-week-high proximity

78%

A/D

C

RVOL

0.7x

50D dollar volume

$2.4B

View META profile

Role

Operates Meta AI products that create large-scale AI inference demand.

Inclusion rationale

Meta markets Meta AI as a personal AI assistant for asking, chatting, and creating across its consumer surfaces. That makes META a platform-side demand driver for AI infrastructure.

Confidence

high

Sector

Technology

50D distance

-6%

52-week-high proximity

71%

A/D

C

RVOL

0.7x

50D dollar volume

$10.7B

View NOW profile

Role

ServiceNow is an enterprise software platform whose Now Assist generative-AI and AI-agent products are embedded across IT, customer service, HR, and security workflows, driving large-scale LLM inference demand.

Inclusion rationale

ServiceNow's FY2025 10-K describes Now Assist as its integrated AI offering, letting customers deploy thousands of out-of-the-box, third-party, or custom AI agents that orchestrate workflows autonomously with human-in-the-loop governance on the Now Platform. As an enterprise AI platform monetizing generative AI and agentic features at scale (its Now Assist business is a key growth driver as of 2026-06), its products generate sustained inference and compute demand, placing it squarely in the AI Platforms stage.

Confidence

high

Sector

Technology

50D distance

17%

52-week-high proximity

66%

A/D

B

RVOL

0.4x

50D dollar volume

$2.4B

View PLTR profile

Role

Sells an enterprise AI platform that helps customers operationalize AI workflows.

Inclusion rationale

Palantir's AIP documentation describes an AI Platform integrated with Foundry for enterprise AI applications. PLTR fits this stage as a software platform that can drive downstream compute and inference needs.

Confidence

high

Sector

Technology

50D distance

26%

52-week-high proximity

85%

A/D

B

RVOL

0.7x

50D dollar volume

$6.1B

View SNOW profile

Role

Snowflake operates the Snowflake AI Data Cloud, a public data/software platform whose Cortex AI and LLM/inference features let enterprises run AI workloads directly on their data, generating demand for GPU-backed inference and compute.

Inclusion rationale

Snowflake's own SEC filing states its cost of product revenue includes "third-party cloud infrastructure expenses, including those related to AI inference and graphics processing units (GPUs), incurred in connection with our customers' use of our platform," and names Snowflake Cortex AI as a core strategy. As of 2026-06 this places SNOW in the AI Platforms stage: its AI software products directly create inference and compute demand.

Confidence

high

Sector

Technology

50D distance

15%

52-week-high proximity

94%

A/D

B

RVOL

0.8x

50D dollar volume

$1.5B

View AMZN profile

Role

Rents high-performance cloud instances for AI training and inference workloads.

Inclusion rationale

AWS markets Amazon EC2 P5 instances for accelerated computing using NVIDIA H100 GPUs. AMZN is a primary cloud-platform buyer, host, and renter of AI compute capacity.

Confidence

high

Sector

Consumer Discretionary

50D distance

5%

52-week-high proximity

91%

A/D

C

RVOL

0.6x

50D dollar volume

$12.8B

View CRWV profile

Role

CoreWeave builds and operates large-scale GPU data centers and rents that AI compute capacity (Nvidia-accelerated servers plus storage, networking, and software) to AI labs and enterprises on a cloud platform.

Inclusion rationale

CoreWeave is a purpose-built AI hyperscale cloud whose platform sits "between the models and the silicon," providing managed GPU infrastructure for training and serving AI models; as of 2026-06 it had surpassed 1 GW of active power, over 3.5 GW of contracted capacity, and a $99.4B revenue backlog from committed customer contracts, squarely placing it in the Cloud Platforms stage that buys, hosts, and rents AI compute.

Confidence

high

Sector

Technology

50D distance

-4%

52-week-high proximity

56%

A/D

E

RVOL

0.6x

50D dollar volume

$2.8B

View GOOGL profile

Role

Operates Google Cloud AI infrastructure for cloud AI workloads.

Inclusion rationale

Google Cloud markets AI infrastructure for building and running AI applications. GOOGL is a primary cloud-platform provider because it sells hosted AI infrastructure capacity directly.

Confidence

high

Sector

Technology

50D distance

-1%

52-week-high proximity

85%

A/D

D

RVOL

1.0x

50D dollar volume

$10.9B

View MSFT profile

Role

Operates Azure AI infrastructure for training and inference at cloud scale.

Inclusion rationale

Microsoft Azure markets AI infrastructure for high-performance computing and AI workloads. MSFT is therefore a primary hyperscale provider of rented AI compute capacity.

Confidence

high

Sector

Technology

50D distance

16%

52-week-high proximity

88%

A/D

A

RVOL

0.5x

50D dollar volume

$16.3B

View NBIS profile

Role

Nebius operates a full-stack AI cloud that buys, hosts, and rents large-scale GPU compute capacity to AI developers, startups, and enterprises.

Inclusion rationale

As of 2026-06, Nebius is an AI-native cloud platform that provisions and leases hyperscale GPU infrastructure — its Q1 2026 results describe it as "the AI cloud company" and detail securing up to 1.2 GW of power for a new owned AI factory, underpinning multibillion-dollar compute deals with Microsoft and Meta. That function — acquiring, hosting, and renting AI compute capacity — places it directly in the Cloud Platforms stage.

Confidence

medium

Sector

Technology

50D distance

-5%

52-week-high proximity

70%

A/D

D

RVOL

0.6x

50D dollar volume

$5.1B

View ORCL profile

Role

Provides Oracle cloud infrastructure for enterprise AI workloads.

Inclusion rationale

Oracle markets AI infrastructure as part of Oracle Cloud Infrastructure. ORCL fits this stage as an enterprise cloud operator renting AI compute and related services.

Confidence

high

Sector

Technology

50D distance

-1%

52-week-high proximity

41%

A/D

D

RVOL

0.5x

50D dollar volume

$4.7B

View AMD profile

Role

AMD designs and sells Instinct-series data-center GPU/APU accelerators (MI300, MI350, and the upcoming MI450) that perform AI model training and inference at hyperscale.

Inclusion rationale

AMD is one of the two dominant merchant suppliers of AI accelerator silicon, and its Instinct GPU ramp is now a primary driver of company revenue: Q1 2026 Data Center segment revenue hit $5.8B (+57% YoY) on continued Instinct shipments, with customers including Meta committing to deploy up to 6 gigawatts of AMD Instinct GPUs. This places AMD squarely in the AI-accelerator silicon stage as of 2026-06.

Confidence

high

Sector

Technology

50D distance

-10%

52-week-high proximity

78%

A/D

C

RVOL

0.6x

50D dollar volume

$14.2B

View AVGO profile

Role

Broadcom designs and supplies custom AI accelerator silicon (XPU ASICs) for hyperscalers' model training and inference workloads, alongside the AI networking chips that connect those accelerators into clusters.

Inclusion rationale

Broadcom's Q1 FY2026 results (ended Feb 1, 2026) reported $8.4B of AI semiconductor revenue, up 106% year-over-year, which CEO Hock Tan attributed to "robust demand for custom AI accelerators and AI networking," with Q2 AI revenue guided to $10.7B. As the leading designer of bespoke training/inference ASICs for hyperscale customers, Broadcom sits at the core of the AI accelerator silicon stage as of 2026-06.

Confidence

high

Sector

Technology

50D distance

-8%

52-week-high proximity

72%

A/D

C

RVOL

0.9x

50D dollar volume

$8.9B

View CBRS profile

Role

Cerebras designs and sells wafer-scale AI accelerator silicon (the Wafer-Scale Engine WSE-3) and CS-3 systems purpose-built for large-model training and inference.

Inclusion rationale

Cerebras's IPO prospectus describes the company as an AI compute platform built around its proprietary WSE-3 chip and CS-3 systems for both AI training and inference, placing it directly in the accelerator-silicon stage as of June 2026 following its May 2026 Nasdaq debut. The chip is positioned as a GPU alternative (cited as having far more compute and on-chip memory than the leading commercial GPU), confirming its function as model-training/inference accelerator silicon.

Confidence

medium

Sector

Technology

RS unavailable for this snapshot

50D distance

-11%

52-week-high proximity

—

A/D

D

RVOL

0.8x

50D dollar volume

$1.5B

View MRVL profile

Role

Marvell designs custom AI accelerator silicon (XPUs/ASICs) for hyperscalers, taping out purpose-built training and inference compute chips for cloud customers' data centers.

Inclusion rationale

Marvell's AI infrastructure page confirms it "designs custom silicon tailored for any application" and has "developed a new custom HBM compute architecture that optimizes custom XPU performance," placing it squarely in the AI-accelerator silicon stage; as of 2026-06 it is a leading merchant designer of custom AI XPUs/ASICs for hyperscalers (e.g., Amazon Trainium, Microsoft Maia, Google).

Confidence

high

Sector

Technology

50D distance

-2%

52-week-high proximity

70%

A/D

B

RVOL

0.8x

50D dollar volume

$8.4B

View NVDA profile

Role

NVIDIA designs and sells the data-center GPU accelerators (the Blackwell/Blackwell Ultra platform plus Hopper) that perform the compute for large-scale AI model training and inference.

Inclusion rationale

NVIDIA's fiscal 2026 Form 10-K describes its Data Center accelerated-computing GPUs as purpose-built for "AI model training and inference," and the filing details the launch and scaling of the Blackwell Ultra platform for agentic, reasoning, and physical AI. As of 2026-06 it is the dominant supplier of merchant GPU silicon to hyperscalers and AI-factory buildouts, placing it squarely in the AI accelerator silicon stage.

Confidence

high

Sector

Technology

50D distance

0%

52-week-high proximity

88%

A/D

D

RVOL

1.2x

50D dollar volume

$26.6B

View QCOM profile

Role

Qualcomm designs and sells the AI200 and AI250 rack-scale data center inference accelerators, Hexagon-NPU-based accelerator silicon (chips, PCIe cards, and full liquid-cooled racks) built to run large-scale generative AI inference workloads.

Inclusion rationale

In October 2025 Qualcomm unveiled the AI200 (commercial in 2026) and AI250 (2027) as purpose-built data center inference accelerators, explicitly positioning them against rack-scale solutions from Nvidia and AMD, which places QCOM directly in the AI accelerator silicon stage as of 2026-06. Confidence is medium because the products are still ramping to commercial availability rather than shipping at volume.

Confidence

medium

Sector

Technology

50D distance

-11%

52-week-high proximity

61%

A/D

E

RVOL

0.9x

50D dollar volume

$2.7B

View MPWR profile

Role

Monolithic Power Systems designs the high-integration analog/mixed-signal DC-to-DC power-conversion ICs and power-management solutions that convert and regulate the voltages delivered to AI accelerators, CPU servers, and other silicon inside AI data-center racks.

Inclusion rationale

Per its FY2025 10-K, MPWR's differentiated DC-to-DC products are used to convert and control voltages within AI systems and cloud/on-premises CPU servers, with its "Enterprise Data" end market (CPU servers, workstations, and AI systems) at 25.2% of 2025 revenue, placing it squarely in the power-delivery/conversion silicon stage feeding AI racks. The filing emphasizes high integration, high-voltage operation, and high load-current capability, the exact characteristics required for dense 48V/high-voltage AI server power delivery as of 2026-06.

Confidence

high

Sector

Technology

50D distance

-6%

52-week-high proximity

75%

A/D

D

RVOL

1.1x

50D dollar volume

$1.3B

View ON profile

Role

onsemi designs and manufactures the power-conversion and power-delivery silicon — silicon and silicon carbide devices, solid state transformers, power supply units, and processor-level voltage regulation — that steps energy down across the full 800VDC-to-core power tree feeding AI data center GPU racks.

Inclusion rationale

In its July 2025 collaboration with NVIDIA, onsemi is positioned to supply high-efficiency, high-density power conversion across every stage of the AI data center power journey — from high-voltage AC/DC conversion at the substation to precise voltage regulation at the processor level — using its silicon and SiC portfolio for 800 VDC distribution, solid state transformers, and power supply units. That spanning role across the AI-rack power tree as of 2026-06 places it squarely in the power-semiconductors stage.

Confidence

medium

Sector

Technology

50D distance

-22%

52-week-high proximity

53%

A/D

D

RVOL

0.9x

50D dollar volume

$1.0B

View MU profile

Role

Supplies high-bandwidth memory for AI accelerators and clusters.

Inclusion rationale

Micron markets HBM3E memory, a key input for AI accelerator performance. MU is a primary memory-stage supplier for AI infrastructure.

Confidence

high

Sector

Technology

50D distance

-5%

52-week-high proximity

73%

A/D

D

RVOL

0.8x

50D dollar volume

$43.8B

View P profile

Role

Everpure (formerly Pure Storage) designs and sells flash-based storage and data-management platforms — FlashBlade//EXA, FlashArray, and Purity OS — that feed and store data for AI training/inference clusters, hyperscalers, and enterprise AI pipelines.

Inclusion rationale

As of June 2026, Everpure positions its FlashBlade//EXA system as purpose-built for large-scale AI and HPC storage (citing the highest-ever SPECstorage 2020 AI_Image benchmark score) and is growing a hyperscaler segment alongside AI-data tooling like Purity DeepReduce and Everpure Data Stream, placing it firmly in the Memory & Storage stage supplying storage for AI clusters. Its Q1 FY2027 results (May 27, 2026) frame the company as helping enterprises "activate their data for the AI era," confirming current relevance to AI infrastructure.

Confidence

medium

Sector

Industrials

RS unavailable for this snapshot

50D distance

20%

52-week-high proximity

—

A/D

B

RVOL

1.2x

50D dollar volume

$249.8M

View SNDK profile

Role

Sandisk designs and manufactures NAND flash storage — including enterprise/data-center SSDs and the emerging High Bandwidth Flash (HBF) memory tier — that supplies the high-capacity, high-bandwidth storage feeding AI training and inference clusters.

Inclusion rationale

As an independent NAND-flash maker (spun out of Western Digital in Feb 2025), Sandisk supplies enterprise SSDs and is co-developing/standardizing High Bandwidth Flash (HBF) with SK hynix as a new memory tier purpose-built for AI inference, placing it directly in the Memory & Storage stage for AI clusters as of 2026-06. Its data-center segment is the company's primary AI-driven growth engine, underscoring its role in storage for AI infrastructure.

Confidence

high

Sector

Industrials

50D distance

-9%

52-week-high proximity

63%

A/D

D

RVOL

0.8x

50D dollar volume

$23.2B

View STX profile

Role

Supplies high-capacity drives for AI data-center storage demand.

Inclusion rationale

Seagate announced 30TB drives to meet global data-center AI storage demand. STX is a primary storage supplier within the AI infrastructure chain.

Confidence

high

Sector

Industrials

50D distance

-11%

52-week-high proximity

69%

A/D

B

RVOL

0.6x

50D dollar volume

$4.6B

View WDC profile

Role

Western Digital designs and manufactures high-capacity hard disk drives (HDDs) that provide the bulk mass-storage layer for the data lakes and training/inference datasets running in hyperscale AI data centers.

Inclusion rationale

As a post-flash-spinoff pure-play HDD maker, Western Digital is repositioning its drives explicitly for the AI-driven data economy, unveiling AI-targeted products (40TB UltraSMR shipping, 60TB ePMR and 100TB+ HAMR roadmap, High Bandwidth Drive and power-optimized "cold data" drives) at its February 2026 Innovation Day. With roughly 90% of revenue tied to AI and cloud and its 2026 HDD production sold out to hyperscalers, it is a core storage supplier for AI clusters as of 2026-06.

Confidence

high

Sector

Industrials

50D distance

-20%

52-week-high proximity

54%

A/D

C

RVOL

0.9x

50D dollar volume

$5.1B

View ALAB profile

Role

Provides PCIe connectivity products for advanced AI and cloud infrastructure.

Inclusion rationale

Astera Labs announced production ramp of a PCIe 6 connectivity portfolio for advanced AI and cloud deployments. ALAB is a primary listed interconnect supplier for AI clusters.

Confidence

high

Sector

Technology

50D distance

-21%

52-week-high proximity

56%

A/D

D

RVOL

0.8x

50D dollar volume

$1.9B

View ANET profile

Role

Provides Ethernet AI networking platforms for data-center clusters.

Inclusion rationale

Arista announced Etherlink AI networking platforms for AI workloads. ANET is a primary networking supplier for cluster fabrics inside AI infrastructure.

Confidence

high

Sector

Industrials

50D distance

6%

52-week-high proximity

88%

A/D

B

RVOL

0.7x

50D dollar volume

$1.4B

View APH profile

Role

Amphenol designs and manufactures high-speed connectors, copper and fiber-optic cable assemblies, and backplane interconnect systems that physically move data between compute, switching, and storage nodes inside data centers and AI clusters.

Inclusion rationale

Amphenol's IT datacom interconnect products (high-speed connectors, near-chip cable, and fiber-optic connectivity) are core hardware for data movement inside AI data centers, and the company is expanding this position via its $10.5B acquisition of CommScope's Connectivity and Cable Solutions business to add fiber-optic interconnect for artificial intelligence and other data center applications. As of 2026-06, AI infrastructure buildout is driving rapid growth in this interconnect portfolio, placing APH squarely in the networking and interconnect stage.

Confidence

high

Sector

Technology

50D distance

-3%

52-week-high proximity

87%

A/D

C

RVOL

0.8x

50D dollar volume

$1.2B

View CIEN profile

Role

Provides optical networking technology for AI data-center connectivity.

Inclusion rationale

Ciena discusses optical innovation for the future of data-center networking in AI environments. CIEN is a primary optical networking exposure in this supply-chain stage.

Confidence

high

Sector

Technology

50D distance

-11%

52-week-high proximity

58%

A/D

E

RVOL

0.8x

50D dollar volume

$1.1B

View CRDO profile

Role

Supplies optical AI connectivity products for low-power data movement.

Inclusion rationale

Credo announced plans to showcase high-performance, low-power optical AI connectivity solutions. CRDO fits the interconnect stage through specialized connectivity for AI infrastructure.

Confidence

high

Sector

Technology

50D distance

-8%

52-week-high proximity

72%

A/D

D

RVOL

0.9x

50D dollar volume

$1.7B

View CSCO profile

Role

Supplies data-center AI networking solutions and switches.

Inclusion rationale

Cisco describes AI networking for data centers with NVIDIA Spectrum-X. CSCO fits this stage as a networking and interconnect supplier for AI clusters.

Confidence

high

Sector

Industrials

50D distance

-5%

52-week-high proximity

85%

A/D

D

RVOL

0.5x

50D dollar volume

$2.7B

View COHR profile

Role

Coherent designs and manufactures the optical transceivers, lasers, silicon-photonics modules, and optical components/cables (DACs/AOCs) that move data between nodes in AI datacenter clusters.

Inclusion rationale

Coherent's own networking site positions its pluggable optical transceivers, lasers/photodiodes, active optical cables, and optical circuit switches as the products that "power next-generation AI networks, hyperscale datacenters, and global communications," with a 1.6T silicon-photonics transceiver and an NVIDIA optics partnership — placing it squarely in the optical interconnect stage as of 2026-06. Its fiscal Q3 2026 results show Datacenter & Communications as the primary growth driver, confirming this is a core, current line of business.

Confidence

high

Sector

Healthcare

50D distance

-16%

52-week-high proximity

63%

A/D

C

RVOL

0.9x

50D dollar volume

$2.0B

View FN profile

Role

Fabrinet is a contract manufacturer that provides advanced optical packaging, precision assembly, and test services to OEMs of optical transceivers, modules, and lasers used to move data between AI cluster nodes.

Inclusion rationale

Fabrinet manufactures the high-speed optical communications components, modules, and subsystems (including 400G/800G/1.6T datacom transceivers and lasers) that form the optical interconnect layer of AI data centers, with datacom now its fastest-growing segment as of its Q3 FY2026 results (quarter ended March 27, 2026). It belongs in this stage because it is the dominant outsourced builder of the advanced transceivers that link AI compute nodes.

Confidence

high

Sector

Technology

50D distance

-18%

52-week-high proximity

56%

A/D

E

RVOL

1.1x

50D dollar volume

$520.4M

View GLW profile

Role

Corning supplies the optical fiber, cables, connectors, and co-packaged-optics systems that physically carry data between nodes in AI compute clusters and across data-center campuses.

Inclusion rationale

Corning's March 2026 announcement details an end-to-end optical interconnect portfolio purpose-built for AI networks — multicore fiber with 4x density, Contour Flow micro cable, MMC connectors, and "fiber to the chip" co-packaged optics — squarely placing it in the Optical Interconnect stage as of 2026-06. This is reinforced by its May 2026 multiyear optical-connectivity partnership with NVIDIA and ongoing AI data-center work with Meta.

Confidence

medium

Sector

Materials

50D distance

-16%

52-week-high proximity

54%

A/D

E

RVOL

0.8x

50D dollar volume

$2.6B

View LITE profile

Role

Lumentum designs and manufactures the high-speed optical transceivers, EML lasers, and photonic components that convert electrical signals to light to move data between nodes inside AI/cloud data center clusters.

Inclusion rationale

Lumentum's FY2025 10-K describes its Cloud & Networking segment supplying optical/photonic chips, components, modules, and subsystems (including EML chips, pump lasers, narrow linewidth laser assemblies for data center interconnect, and 800G modules) to cloud data center and AI/ML infrastructure providers. As of 2026-06 it is a leading merchant supplier of the 200 Gbps-per-lane EMLs and 800G/1.6T transceivers that carry traffic between AI cluster nodes, placing it primarily in the optical interconnect stage.

Confidence

high

Sector

Technology

50D distance

2%

52-week-high proximity

76%

A/D

B

RVOL

0.7x

50D dollar volume

$4.1B

View CDNS profile

Role

Cadence Design Systems develops electronic design automation (EDA) software, verification/emulation systems, and semiconductor design IP that chipmakers use to architect, design, and verify advanced AI silicon before manufacturing.

Inclusion rationale

Cadence's FY2025 10-K describes the company as a leading provider of EDA software, hardware, and semiconductor IP, with $5.3B revenue (up 14%) driven by demand across software, hardware, and IP product offerings. Its tools (e.g., digital implementation, logic simulation, and formal verification, plus AI-driven flows partnered with NVIDIA and TSMC) are the design-stage software layer used to create AI chips, placing it squarely in the chip-design EDA stage as of 2026-06.

Confidence

high

Sector

Technology

50D distance

-10%

52-week-high proximity

76%

A/D

D

RVOL

0.9x

50D dollar volume

$815.2M

View SNPS profile

Role

Synopsys is the leading provider of electronic design automation (EDA) software and semiconductor design IP that engineers use to architect, optimize, verify, and tape out the advanced silicon underlying AI accelerators and high-performance computing chips.

Inclusion rationale

Synopsys self-describes as "#1 in EDA" and "#2 in Silicon IP," supplying the AI-enabled design-automation tools and pre-built IP blocks that chip designers rely on to build AI silicon at advanced nodes, placing it squarely in the chip-design EDA stage. As of 2026 this remains its core business, with Design Automation revenue reaching $2.0B in Q1 FY2026 amid surging AI-driven design complexity.

Confidence

high

Sector

Technology

50D distance

-6%

52-week-high proximity

64%

A/D

E

RVOL

0.9x

50D dollar volume

$776.7M

View ARM profile

Role

Provides processor IP used by semiconductor designers.

Inclusion rationale

Arm markets microprocessor cores and processor technology as silicon IP. ARM fits the foundry and IP stage as a design-IP provider for advanced chips.

Confidence

high

Sector

Technology

50D distance

-20%

52-week-high proximity

53%

A/D

D

RVOL

0.6x

50D dollar volume

$2.3B

View GFS profile

Role

GlobalFoundries is a scaled pure-play semiconductor foundry that manufactures complex integrated circuits on differentiated process platforms for fabless chip designers and IDMs, supplying silicon for end markets including communications infrastructure and data center.

Inclusion rationale

As of June 2026, its FY2025 Form 20-F describes GlobalFoundries as "one of the world's leading semiconductor foundries, manufacturing complex ICs" with a highly differentiated technology portfolio serving Communications Infrastructure / Data Center customers, and the filing reflects its 2025 acquisition of MIPS (processor design IP) and InfiniLink, placing it squarely in the foundry and manufacturing-partner layer behind advanced silicon. It is a specialty/mature-node foundry rather than a leading-edge AI-accelerator fab, which warrants medium confidence for this stage.

Confidence

medium

Sector

Technology

50D distance

-27%

52-week-high proximity

50%

A/D

E

RVOL

0.8x

50D dollar volume

$225.9M

View INTC profile

Role

Intel operates the Intel Foundry segment, an integrated device manufacturer that fabricates advanced AI/compute silicon on its own leading-edge nodes (e.g., 18A with RibbonFET/PowerVia) and offers wafer fabrication, advanced packaging, chiplet integration, and design-enablement services to external foundry customers.

Inclusion rationale

Intel's FY2025 10-K states that its Intel Foundry operating segment "offers semiconductor design and manufacturing services to external customers" across wafer fabrication, advanced packaging, chiplet integration, and design enablement, and that it "ramped Intel 18A into high-volume production" in 2025 as the first U.S.-made gate-all-around/backside-power node. As a U.S.-based foundry, design-IP provider, and manufacturing partner for advanced AI silicon, Intel sits squarely in the Foundry & IP stage as of 2026-06.

Confidence

high

Sector

Technology

50D distance

-19%

52-week-high proximity

61%

A/D

D

RVOL

0.8x

50D dollar volume

$12.6B

View TSM profile

Role

Manufactures advanced semiconductors for fabless chip designers.

Inclusion rationale

TSMC describes itself as a semiconductor foundry company. TSM is a primary manufacturing partner behind advanced AI silicon.

Confidence

high

Sector

Technology

50D distance

-3%

52-week-high proximity

86%

A/D

D

RVOL

0.9x

50D dollar volume

$5.8B

View AMKR profile

Role

Amkor is the largest U.S.-headquartered OSAT provider, delivering outsourced advanced-packaging assembly and test — including 2.5D/3D heterogeneous integration and high-density fan-out for HBM-to-logic integration — that enables AI accelerators and data-center compute silicon.

Inclusion rationale

Amkor's May 2026 investor presentation frames advanced packaging (2.5D/3D, HBM integration) as the company's core growth engine amid an "Advanced Packaging-Led Era," with computing/AI accelerators a key end market, placing it squarely in the advanced-packaging/OSAT stage behind AI accelerators as of 2026-06. As a primary OSAT integrator of HBM into 2.5D/3D packages for AI chips, it occupies a load-bearing node in this stage.

Confidence

high

Sector

Technology

50D distance

-27%

52-week-high proximity

50%

A/D

D

RVOL

0.8x

50D dollar volume

$378.4M

View KLIC profile

Role

Kulicke & Soffa designs and sells the thermo-compression and hybrid bonding equipment used to stack HBM memory and assemble 2.5D/3D advanced packages, the back-end interconnect tooling that physically integrates memory and logic dies in AI accelerator packages.

Inclusion rationale

As of 2026-06, K&S sits squarely in the advanced-packaging/OSAT stage: its expanded memory portfolio spans Ball, Vertical Wire, Advanced Thermo-Compression, and Hybrid Bonding, the very interconnect processes used to build HBM stacks and 2.5D/3D packages, and management cites roughly 70% sequential TCB growth in fiscal 2026 plus its first HBM system shipment, tying the business directly to AI-driven advanced packaging demand. Confidence is medium because K&S is a back-end equipment supplier (not an OSAT or packaging house itself), so its revenue is a derived, lagging proxy for end-market AI advanced-packaging volume.

Confidence

medium

Sector

Technology

50D distance

-20%

52-week-high proximity

61%

A/D

C

RVOL

0.5x

50D dollar volume

$123.4M

View ONTO profile

Role

Onto Innovation supplies the optical metrology and inspection systems (e.g., the Dragonfly platform and 3Di technology) that semiconductor and HBM makers use for process control and defect detection across 2.5D/3D advanced-packaging and HBM stacking flows.

Inclusion rationale

Per Onto Innovation's March 2026 IR release, a leading HBM manufacturer selected its Dragonfly G5 inspection system for the HBM4 ramp with double-digit orders of G5 and 3Di tools, and the company expects advanced-packaging revenue to grow over 30% in 2026 — placing it squarely in the advanced-packaging/HBM integration stage behind AI accelerators as of 2026-06.

Confidence

high

Sector

Healthcare

50D distance

-7%

52-week-high proximity

73%

A/D

D

RVOL

0.6x

50D dollar volume

$397.5M

View AMAT profile

Role

Provides materials engineering equipment for semiconductor manufacturing.

Inclusion rationale

Applied Materials' 10-K is a primary company filing for its semiconductor equipment business. AMAT fits this stage through equipment used in advanced chip production.

Confidence

high

Sector

Technology

50D distance

-13%

52-week-high proximity

65%

A/D

D

RVOL

1.2x

50D dollar volume

$5.6B

View ASML profile

Role

Supplies lithography systems used to manufacture advanced chips.

Inclusion rationale

ASML describes its role as a semiconductor equipment company. ASML is a primary semicap supplier because advanced AI silicon depends on lithography capacity.

Confidence

high

Sector

Industrials

50D distance

-2%

52-week-high proximity

87%

A/D

C

RVOL

0.8x

50D dollar volume

$3.3B

View KLAC profile

Role

Provides process control and yield management systems for chip manufacturing.

Inclusion rationale

KLA describes itself as a leader in process control and yield management. KLAC fits this stage because advanced chip production depends on inspection and yield tools.

Confidence

high

Sector

Healthcare

50D distance

-17%

52-week-high proximity

59%

A/D

D

RVOL

0.8x

50D dollar volume

$2.9B

View LRCX profile

Role

Supplies wafer fabrication equipment for semiconductor process steps.

Inclusion rationale

Lam Research presents products used across semiconductor manufacturing processes. LRCX is a primary semicap equipment supplier for advanced chip production.

Confidence

high

Sector

Industrials

50D distance

-8%

52-week-high proximity

71%

A/D

D

RVOL

0.8x

50D dollar volume

$4.0B

View TER profile

Role

Supplies semiconductor test equipment used in chip production.

Inclusion rationale

Teradyne markets semiconductor testing products. TER is a semicap equipment exposure through test systems used to validate chips and devices.

Confidence

high

Sector

Healthcare

50D distance

-5%

52-week-high proximity

75%

A/D

D

RVOL

0.7x

50D dollar volume

$1.7B

View AAON profile

Role

AAON, through its BASX subsidiary, designs and manufactures custom liquid-cooling and free-cooling chiller systems and thermal-management equipment installed inside AI data centers to reject the heat from high-density GPU/compute racks.

Inclusion rationale

AAON/BASX supplies purpose-built data-center cooling systems for AI workloads — for example its customized free-cooling chiller deployment for Applied Digital's Polaris Forge 1 AI campus, which rejects up to 100% of the IT load with zero water consumption for power densities 15-30x traditional data centers. With BASX-branded data-center sales surging in 2025-2026 on liquid-cooling demand, the company is a direct provider of the thermal/cooling layer of AI data-center physical infrastructure as of 2026-06.

Confidence

medium

Sector

Industrials

50D distance

-28%

52-week-high proximity

51%

A/D

E

RVOL

1.4x

50D dollar volume

$114.8M

View CARR profile

Role

Carrier Global supplies the cooling backbone of AI data centers—centrifugal chillers (e.g., AquaEdge 30CF) and direct-to-chip liquid cooling integrated through its QuantumLeap thermal-management suite—to remove heat from high-density compute.

Inclusion rationale

As of 2026-06, Carrier is actively scaling AI-data-center thermal infrastructure: Carrier Ventures expanded its 2025 investment in ZutaCore (April 2026) to deliver interoperable single- and two-phase, waterless direct-to-chip liquid cooling under its QuantumLeap suite, and launched the AquaEdge 30CF chiller (Feb 2026) for data-center uptime. This places it squarely in the cooling/physical-infrastructure stage as chip power densities push thermal limits.

Confidence

high

Sector

Industrials

50D distance

-11%

52-week-high proximity

78%

A/D

D

RVOL

0.6x

50D dollar volume

$403.7M

View ETN profile

Role

Provides modular power infrastructure for rapid AI factory deployment.

Inclusion rationale

Eaton announced modular data-center offerings for AI factories from grid to chip. ETN is a primary supplier of power distribution and electrical infrastructure for AI data centers.

Confidence

high

Sector

Industrials

50D distance

-2%

52-week-high proximity

85%

A/D

B

RVOL

0.7x

50D dollar volume

$946.5M

View JCI profile

Role

Johnson Controls designs and manufactures the integrated cooling and thermal-management systems inside AI data centers — Silent-Aire coolant distribution units, York air- and water-cooled chillers, and CRAH/CRAC air handlers that capture, transfer, and reject heat from high-density GPU racks.

Inclusion rationale

Johnson Controls' data-center thermal-management portfolio spans heat capture (Silent-Aire CDUs delivering 500 kW–10 MW of liquid cooling to cold plates), transfer (York/M&M Carnot CRAH/CRAC units), and rejection (York YDAM/YVAM magnetic-bearing chillers), positioning it squarely in the cooling and physical-infrastructure layer of AI data centers as of 2026-06. Recent moves — the Accelsius direct-to-chip investment and the agreement to acquire Alloy Enterprises for next-gen liquid cooling — reinforce its active leadership in this stage.

Confidence

high

Sector

Industrials

50D distance

-2%

52-week-high proximity

91%

A/D

C

RVOL

0.6x

50D dollar volume

$584.3M

View MOD profile

Role

Modine designs and manufactures the in-data-center thermal stack — chillers, dry coolers, precision air handlers, CRAC/CRAH units, fan walls, rear-door heat exchangers, coolant distribution units (CDUs), and immersion cooling — through its Data Centers business (Airedale by Modine).

Inclusion rationale

Per its FY2026 10-K, Modine's Data Centers business sells data-center cooling products (chillers, dry coolers, CRAC/CRAH, fan walls, rear-door heat exchangers, CDUs, and immersion solutions) explicitly to serve AI-driven high-performance-computing demand, and it signed a long-term capacity agreement to supply more than $4 billion of these products in 2027-2029. This places Modine squarely in the cooling and physical-infrastructure layer inside AI data centers as of 2026-06.

Confidence

high

Sector

Industrials

50D distance

-18%

52-week-high proximity

58%

A/D

C

RVOL

0.8x

50D dollar volume

$371.8M

View NVT profile

Role

nVent Electric designs and manufactures the cooling, power-distribution, and rack infrastructure inside AI data centers — coolant distribution units (CDUs), liquid-cooling manifolds, intelligent PDUs, and racks that remove heat and deliver power at the rack level.

Inclusion rationale

At SC25 in November 2025 nVent unveiled a modular liquid-cooling and power portfolio (row- and rack-based CDUs, TCS manifolds, next-gen PDUs, and reference-design racks) built to chip-makers' cooling specs and to hyperscale AI workloads, including a Project Deschutes CDU based on Google's OCP specification and a liquid-cooling/power reference architecture with Siemens. This places nVent squarely in the thermal, cooling, power-distribution, and physical-infrastructure layer inside AI data centers as of 2026-06.

Confidence

high

Sector

Industrials

50D distance

-5%

52-week-high proximity

83%

A/D

B

RVOL

1.1x

50D dollar volume

$343.8M

View TT profile

Role

Provides thermal management reference designs for AI factories.

Inclusion rationale

Trane Technologies announced thermal management reference designs for AI factories. TT fits this stage through cooling infrastructure for high-density AI facilities.

Confidence

high

Sector

Healthcare

50D distance

-4%

52-week-high proximity

90%

A/D

D

RVOL

1.1x

50D dollar volume

$573.4M

View VRT profile

Role

Provides converged physical infrastructure for AI factories and data centers.

Inclusion rationale

Vertiv announced infrastructure support for NVIDIA Vera Rubin DSX AI factories. VRT is a primary thermal and power infrastructure supplier for AI data centers.

Confidence

high

Sector

Technology

50D distance

-13%

52-week-high proximity

67%

A/D

D

RVOL

0.6x

50D dollar volume

$1.8B

View ACM profile

Role

Provides engineering and infrastructure services for data centers and digital infrastructure.

Inclusion rationale

AECOM markets data centers and digital infrastructure services within high-tech markets. The exposure is project and infrastructure services rather than owned grid equipment, so confidence is medium.

Confidence

medium

Sector

Industrials

50D distance

-6%

52-week-high proximity

48%

A/D

C

RVOL

0.9x

50D dollar volume

$135.0M

View GEV profile

Role

GE Vernova designs and manufactures grid equipment, power transmission, and electrification systems (grid solutions, transformers, power conversion, grid automation/software) that connect, transmit, and stabilize electricity for utilities and data centers.

Inclusion rationale

GE Vernova's Electrification segment supplies the transmission and grid-equipment hardware (including Prolec GE transformers) that AI data centers and utilities need to expand capacity, booking $2.4B in equipment orders to support data centers in Q1 2026 alone — more than all of the prior year — placing it squarely in the grid-electrification layer exposed to AI power demand. Its full acquisition of grid-equipment supplier Prolec GE in Q1 2026 added ~$5B to backlog, reinforcing its central role in grid equipment and transmission as of 2026-06.

Confidence

high

Sector

Technology

50D distance

-9%

52-week-high proximity

79%

A/D

D

RVOL

0.8x

50D dollar volume

$2.9B

View HUBB profile

Role

Provides electrical and utility products for data-center markets.

Inclusion rationale

Hubbell lists data centers as a market for its products. HUBB is a primary electrification supplier exposed to data-center and grid equipment demand.

Confidence

high

Sector

Technology

50D distance

-6%

52-week-high proximity

83%

A/D

D

RVOL

1.0x

50D dollar volume

$285.5M

View MYRG profile

Role

MYR Group is one of the largest U.S. specialty electrical contractors building, upgrading, and maintaining high-voltage transmission lines, substations, and distribution networks for electric utilities across North America.

Inclusion rationale

Its Transmission & Distribution segment (~$2.0B of 2025 revenue) directly constructs and modernizes the grid equipment and substation infrastructure that utilities are expanding to meet rising electrification and data-center power demand; the FY2025 10-K cites growing investor-owned utility transmission investment ($178B projected 2025-2028) and data-center-driven electrification as core demand tailwinds. This places it squarely in Grid & Electrification, with medium confidence as a contractor/installer rather than a grid-equipment OEM.

Confidence

medium

Sector

Industrials

50D distance

-22%

52-week-high proximity

61%

A/D

E

RVOL

0.9x

50D dollar volume

$121.2M

View POWL profile

Role

Powell Industries designs and manufactures custom-engineered medium-voltage switchgear and E-House modular substations that connect the utility grid to power-hungry loads, supplying the electrical distribution and control equipment that energizes data centers, substations, and grid-modernization projects.

Inclusion rationale

Powell's medium-voltage switchgear and modular substation packages are the interface between the transmission/distribution grid and large electrical loads, making it a direct grid-equipment and electrification supplier. As of 2026-06, its order book is being driven by AI-led power demand — its first data center megaproject booking (over $100M of data center orders in fiscal Q1 2026) alongside 35% YoY growth in electric utility revenue — confirming active exposure to grid build-out and AI power demand.

Confidence

high

Sector

Technology

50D distance

-21%

52-week-high proximity

58%

A/D

E

RVOL

0.6x

50D dollar volume

$187.8M

View PWR profile

Role

Provides power generation and grid infrastructure solutions for large-load customers.

Inclusion rationale

Quanta announced its selection by NiSource to provide power generation and grid infrastructure solutions for a large-load customer. PWR fits the grid-electrification stage through infrastructure work tied to data-center-driven demand.

Confidence

high

Sector

Industrials

50D distance

-8%

52-week-high proximity

78%

A/D

D

RVOL

1.3x

50D dollar volume

$856.2M

View BE profile

Role

Bloom Energy manufactures and deploys on-site solid oxide fuel cell power systems (Energy Servers) that supply around-the-clock baseload electricity contracted for AI data centers and AI infrastructure.

Inclusion rationale

Bloom belongs to the broader AI-baseload generation theme as of 2026-06 because its FY2025 10-K confirms it sells dispatchable, grid-independent baseload power into AI/data-center customers, anchored by multi-gigawatt deals such as Oracle's Project Jupiter (up to 2.8 GW). Note a stage-fit caveat: Bloom is a fuel cell generator, not a merchant nuclear or SMR operator, and is widely positioned as the near-term alternative to nuclear/SMR for AI baseload, hence medium confidence in this nuclear-and-SMR stage.

Confidence

medium

Sector

Technology

50D distance

-15%

52-week-high proximity

58%

A/D

E

RVOL

0.9x

50D dollar volume

$3.6B

View CEG profile

Role

Constellation Energy operates the largest U.S. nuclear fleet plus hydro, wind, solar, and (via Calpine) gas generation, supplying around-the-clock electricity to growing data-center and AI computing load through long-term power purchase agreements.

Inclusion rationale

CEG's Q1 2026 results emphasize that its nuclear fleet and flexible generation are being contracted to meet accelerating power demand from data centers and the AI/data economy, with disclosed gigawatt-scale agreements (e.g., the CyrusOne/Calpine relationship) and PPAs with hyperscale customers. As of June 2026 it is a direct upstream electricity supplier to the data-center buildout, placing it squarely in the energy-supply stage.

Confidence

high

Sector

Utilities

50D distance

4%

52-week-high proximity

66%

A/D

C

RVOL

1.0x

50D dollar volume

$828.1M

View DUK profile

Role

Duke Energy is a regulated electric utility and power generator that supplies electricity to its Southeast/Midwest service territories, including large data-center customers connecting to its grid under Electric Service Agreements.

Inclusion rationale

Duke Energy's Q1 2026 SEC earnings release states it has secured a total of 7.6 GW of economic development projects under Electric Service Agreements and identifies data-center electricity usage as a primary load-growth driver, placing it squarely in the energy-supply stage as a generator/utility serving growing data-center power demand as of 2026-06. As a regulated electric utility, it directly generates and delivers the electricity that data centers consume.

Confidence

high

Sector

Utilities

50D distance

-3%

52-week-high proximity

91%

A/D

D

RVOL

0.7x

50D dollar volume

$531.9M

View NEE profile

Role

NextEra Energy is a major US power generator and utility that develops and operates electricity generation (renewables, storage, and natural gas) to supply large-scale loads, including data centers.

Inclusion rationale

As of 2026-06, NextEra received federal approval to develop up to 10 GW of natural-gas generation in Texas and Pennsylvania explicitly designed to serve growing electricity demand from large-scale users including data centers, and the company has publicly targeted 15-30 GW of new data-center power-supply capacity by 2035. This places it squarely in the Energy Supply stage as a power generator/utility feeding growing data-center electricity demand.

Confidence

high

Sector

Utilities

50D distance

-3%

52-week-high proximity

85%

A/D

C

RVOL

0.7x

50D dollar volume

$989.6M

View NRG profile

Role

NRG Energy is an independent power producer and integrated retail electricity provider that generates and sells wholesale and retail power, increasingly contracting new gas-fired generation and long-term retail power agreements to supply the growing electricity demand of data centers across the ERCOT and PJM markets.

Inclusion rationale

As of 2026-06, NRG operates a generation fleet supplying surging Texas and PJM power demand and is advancing gas-fired generation development (a ~5.4 GW project with GE Vernova/Kiewit and data-center LOIs scaling toward 6.5 GW) plus signed data-center retail power agreements, placing it squarely in the energy-supply stage feeding data-center electricity needs. The Q1 2026 earnings filing reaffirms 2026 guidance, cites strong ERCOT fleet availability, and notes leadership's emphasis on growing demand for reliable, affordable power generation.

Confidence

high

Sector

Utilities

50D distance

-15%

52-week-high proximity

59%

A/D

D

RVOL

1.5x

50D dollar volume

$377.1M

View SO profile

Role

Operates regulated utility and generation assets that can support load growth.

Inclusion rationale

Southern Company frames energy innovation and investment as central to powering growth. The AI linkage is indirect through regional load growth and utility supply, so confidence is medium.

Confidence

medium

Sector

Utilities

50D distance

-5%

52-week-high proximity

89%

A/D

C

RVOL

1.5x

50D dollar volume

$559.6M

View VST profile

Role

Independent power producer that generates and sells nuclear and natural-gas electricity, supplying carbon-free power directly to hyperscaler data centers under long-term power purchase agreements.

Inclusion rationale

Vistra owns and operates nuclear and gas generation and, as of 2026-06, has contracted that capacity to AI/data-center demand — including a 20-year PPA announced 2026-01-09 to supply Meta with 2,609 MW of carbon-free nuclear power from its Perry and Davis-Besse plants, alongside ~3,800 MW of nuclear supply to Amazon at Comanche Peak. This makes it a core electricity supplier feeding growing data-center load, the defining role of the energy-supply stage.

Confidence

high

Sector

Utilities

50D distance

-11%

52-week-high proximity

62%

A/D

D

RVOL

1.1x

50D dollar volume

$683.6M

View OKLO profile

Role

Oklo designs, builds, owns, and operates its Aurora fast-fission small modular reactors (15-75 MW each) to sell carbon-free baseload electricity directly to AI data center, energy, and industrial customers under long-term power purchase agreements.

Inclusion rationale

Per its FY2025 Form 10-K, Oklo's core business is deploying Aurora powerhouses under a build-own-operate model and selling baseload power to hyperscale data centers, anchored by a ~14 GW pipeline including a 12 GW Master Power Agreement with Switch, a 500 MW Equinix LOI (with $25M prepayment), and a 1.2 GW Meta-backed Ohio campus. It belongs in Nuclear & SMR Generation as a merchant SMR generator contracting capacity for AI baseload, though confidence is medium because the pipeline is largely non-binding and its first reactor (Aurora-INL) is not yet operational as of 2026-06.

Confidence

medium

Sector

Utilities

50D distance

-16%

52-week-high proximity

20%

A/D

D

RVOL

0.8x

50D dollar volume

$446.2M

View TLN profile

Role

Talen Energy is an independent power producer that operates the 2.2 GW Susquehanna nuclear plant as merchant baseload and is contracting that carbon-free generation—plus planned small modular reactors—to supply Amazon's AI/cloud data centers.

Inclusion rationale

Talen signed a long-term power purchase agreement to deliver up to 1,920 MW of Susquehanna nuclear output to Amazon Web Services data centers in Pennsylvania (running through 2042), explicitly to support AI and cloud computing, and the two parties also agreed to explore building new Small Modular Reactors within Talen's footprint. This places TLN squarely in merchant-nuclear and SMR generation contracted for AI baseload as of 2026.

Confidence

high

Sector

Utilities

50D distance

-17%

52-week-high proximity

68%

A/D

C

RVOL

0.9x

50D dollar volume

$329.4M

View CCJ profile

Role

Cameco mines and mills uranium and provides refining, conversion, and enrichment (via its Global Laser Enrichment interest) and fuel-services capacity that turns natural uranium into the fuel utilities use to generate nuclear power.

Inclusion rationale

Cameco is one of the world's largest uranium producers and is invested across the nuclear fuel cycle — mining/milling, refining and conversion, and enrichment through Global Laser Enrichment — supplying the fuel that utilities rely on to generate emissions-free nuclear power, which is the generation source increasingly tapped for AI data-center load as of 2026. This positions it as a primary upstream input to nuclear generation feeding AI power demand.

Confidence

high

Sector

Materials

50D distance

6%

52-week-high proximity

76%

A/D

D

RVOL

0.7x

50D dollar volume

$310.4M

View LEU profile

Role

Centrus Energy operates the American Centrifuge Plant in Piketon, Ohio, enriching uranium into low-enriched uranium (LEU) for the existing reactor fleet and high-assay low-enriched uranium (HALEU) for next-generation reactors, occupying the front end of the nuclear fuel cycle.

Inclusion rationale

Centrus is a U.S.-owned uranium enrichment supplier and the only company with an NRC-licensed facility producing HALEU; as of 2026-06 it is scaling commercial LEU and HALEU output backed by a January 2026 $900M DOE award and a multi-billion-dollar enrichment backlog, directly supplying the nuclear fuel that powers generation for surging AI/data-center electricity demand.

Confidence

high

Sector

Materials

50D distance

2%

52-week-high proximity

38%

A/D

D

RVOL

0.9x

50D dollar volume

$130.6M