The Chokepoints

Sole manufacturer of EUV lithography systems. Intel's own 10-K calls ASML the 'sole supplier' of EUV tools deployed in every leading-edge node. There is no second source — and ASML's own optics come from a single supplier.

ASML designs and manufactures the lithography systems that print circuit patterns on silicon wafers. Its extreme ultraviolet (EUV) machines are essential for producing the most advanced logic and memory chips, including AI accelerators and high-bandwidth memory. The company also provides deep ultraviolet (DUV) immersion and dry lithography, together with metrology, inspection, and a growing installed‑base services and upgrades business.

The only foundry capable of fabricating leading-edge AI accelerators at volume. Multiple chip designers file TSMC as their sole wafer source, and the chairman says switching takes five to seven years.

TSMC operates as a pure‑play foundry, fabricating integrated circuits on silicon wafers for customers who design but do not manufacture chips. Its process portfolio spans mature specialty nodes to the most advanced gate‑all‑around N2 technology, and it also provides design enablement, mask‑making, advanced packaging (3DFabric), and testing. The company is the sole‑source manufacturer for nearly every high‑performance AI chip, from NVIDIA’s GPUs to custom ASICs from Broadcom, AMD, Amazon, Meta, and Google.

The AI training GPU standard. Multiple customers file NVIDIA as a single-source supplier, CUDA's software ecosystem locks in the installed base, and management says supply is constrained through the entire VeraRubin lifecycle.

NVIDIA designs the accelerators, networking, and software that form the backbone of modern AI infrastructure. Its GPUs and systems power the training and inference of frontier models, and its CUDA software ecosystem is the default platform across every major cloud and AI lab. The company has recently expanded into standalone CPUs (Vera) and integrated CPU-GPU platforms (Vera Rubin), transforming it from a GPU company into a full-stack data center platform provider.

Seven times larger than its nearest competitor in process control, and the gap has widened 360 basis points since 2021. Even rival equipment makers — Applied Materials, ASML, Lam Research — are KLA customers.

KLA Corporation is the world’s dominant supplier of process control and yield‑management equipment for semiconductor manufacturing. Its systems—wafer inspectors, reticle inspectors, metrology tools, and process control software—are used at virtually every step of chip fabrication to find and classify defects. Without KLA’s tools, fabs cannot achieve the yields required to make leading‑edge logic, memory, or advanced packages economically viable. The company’s inspection and measurement platforms are particularly critical for AI chips, where larger die sizes, complex 3D stacking, and near‑zero‑defect HBM requirements intensify the need for process control.

Designs the custom AI silicon for Google, Meta, and other top hyperscalers. Arista's own 10-K names Broadcom as a sole source for switching chips, and the only 100-terabit Ethernet switch in the industry is theirs.

Broadcom designs and supplies a vast portfolio of semiconductor and infrastructure software products. Its semiconductor segment spans custom AI accelerators (XPUs/TPUs), Ethernet switches, SerDes, optical components, RF filters, and storage connectivity — chips that underpin hyperscale data centers, enterprise networking, and broadband.

Half of the EDA duopoly. Thirty-five companies depend on its tools and IP to design chips, and management says hyperscalers' 'entire custom-silicon strategy is built on the availability of Synopsys IP.'

Synopsys provides electronic design automation (EDA) software, hardware-assisted verification systems, and silicon intellectual property (IP) essential for designing semiconductors. Its tools are used by virtually every major chip designer and foundry to create the most advanced processors, AI accelerators, and networking chips at leading-edge nodes and in 3D-IC packages. With the Ansys acquisition, the company now also offers multiphysics simulation that extends from chip to full-scale system, addressing thermal, electromagnetic, and structural challenges critical for AI data centers.

The other half of the EDA duopoly, supplying tools to 47 companies including every major hyperscaler custom-silicon program. No third player covers the full design flow.

Cadence Design Systems is a global leader in electronic design automation (EDA), providing the software, emulation hardware, and semiconductor intellectual property (IP) that customers use to design integrated circuits and electro-mechanical systems. Its tools are essential for developing the AI accelerators, advanced-node processors, and 3D-IC packaging that underpin the AI infrastructure buildout. From the Virtuoso analog environment to the Innovus digital implementation suite and the Palladium Z3 emulator, Cadence spans the full chip design flow. The company also offers interface and memory IP (HBM, UCIe, PCIe, LPDDR6), which is critical for high-bandwidth AI silicon, and a growing System Design & Analysis (SDA) segment covering multiphysics simulation for electromagnetics, thermal, structural dynamics, and CFD.

One of exactly three companies capable of making HBM memory. The CEO says Micron can fulfill only 50 to 66 percent of key customer demand, with multi-year lead times for new capacity.

Micron Technology designs and manufactures DRAM, NAND flash, and NOR memory chips, along with solid-state drives and multi-chip packages, under the Micron and Crucial brands. Its memory products are foundational to every tier of the AI stack, from high-bandwidth memory (HBM) for GPU accelerators to high-capacity DDR5 for cloud servers and LPDDR for AI-enabled edge devices.

The instruction-set architecture behind roughly 100 percent of DPU and SmartNIC silicon and about half of hyperscaler CPUs. Twenty-two million developers and 250 billion cumulative chip shipments create lock-in that is architectural, not contractual.

Arm designs and licenses the central processor (CPU) architectures that serve as the computational foundation for the majority of the world’s electronic devices. Its instruction-set architecture (ISA), CPU core designs, and pre-integrated Compute Subsystems (CSS) are used by chipmakers and hyperscalers to build processors that balance performance with extreme energy efficiency. This makes Arm’s technology essential in smartphones, where it powers nearly all devices, and increasingly in data-center servers, networking equipment, and automotive systems — all areas where AI workloads demand dense, power-conscious compute.

One of three companies in the world that can build HA-class gas turbines for AI power, with roughly three-year lead times and slots sold out through 2030. A major competitor has stopped taking orders beyond 2030.

GE Vernova designs, manufactures, and services the heavy‑duty gas turbines, HVDC transmission systems, transformers, and switchgear that form the backbone of the world’s electricity grid. With an installed base generating approximately 25% of global electricity, the company’s equipment is critical to meeting the surge in power demand from AI data centers. Its gas turbines, in particular, are increasingly deployed for dedicated data‑center campuses, while its Electrification segment provides the substations and grid automation that connect those campuses to the grid and optimize power flows.

One of three companies that supply ultra-high-purity industrial gases to leading-edge fabs, locked in by 15-to-20-year take-or-pay on-site contracts. Once a plant is piped into a fab, switching is effectively impossible.

Linde is the world's largest industrial gas company, producing atmospheric gases such as oxygen, nitrogen, and argon, as well as process gases including hydrogen, helium, and specialty electronic gases. It also engineers and builds gas production plants. Its gases and on-site infrastructure are essential to a wide range of industries, and its ultra-high-purity capabilities make it a critical supplier to advanced semiconductor fabs that manufacture AI-class chips.

The only U.S. company with proven technology to produce HALEU nuclear fuel — confirmed by competitor filings and an NNSA sole-source procurement. No domestic alternative arrives before approximately 2029.

Centrus Energy enriches uranium to produce nuclear fuel. It supplies low-enriched uranium (LEU) to utilities operating commercial reactors and produces high-assay low-enriched uranium (HALEU) for advanced reactors and government programs. The company's centrifuge technology is the only proven American enrichment method, positioning it as the sole domestic source of HALEU and a critical component of the U.S. nuclear fuel supply chain. As AI data centers drive demand for carbon-free power, Centrus' fuel becomes an essential input for the reactors that serve them.

Switchgear and electrical distribution with data center orders up 240 percent year-over-year and a 228 GW backlog — roughly twelve years at current build rates. Spending over a billion dollars on 24 factory expansions.

Eaton is an intelligent power management company whose grid‑to‑chip portfolio spans power distribution, power quality, and liquid cooling for AI data centres, along with aerospace and electrical systems for commercial, industrial, and utility markets. Its products—from medium‑voltage switchgear to cold plates and integrated modular platforms—form the infrastructure layer between the power grid and the high‑density computing that AI demands.

The switching fabric that stitches thousands of GPUs into a single cluster. Number-one market share in high-speed switching, $8.9 billion in purchase commitments, and 52-week component lead times.

Arista Networks develops high‑speed networking platforms built on its Extensible Operating System (EOS), a single software binary that runs across its entire switch and router portfolio. This unified stack enables consistent automation, telemetry, and security, making it especially valuable for hyperscale and AI operators that must manage massive, distributed fabrics. The company’s Ethernet solutions span data center leaf‑spine fabrics, inter‑data‑center routing, campus switching, and WAN connectivity.

Half of the nearline HDD duopoly with Western Digital. Both companies refuse to add unit capacity, both are fully allocated for years, there is no third manufacturer, and flash storage remains far too expensive per terabyte to substitute.

Seagate is a data storage company, with hard disk drives as its primary product, complemented by solid-state drives, storage subsystems, and the Lyve edge-to-cloud platform. Its Mass Capacity Storage segment—dominated by nearline HDDs—serves cloud and enterprise customers with high-capacity, cost-efficient drives that underpin the mass storage tier of modern data centers.