Divergent Compute.AI Economic Think Tank

Markets / Industries / Semiconductors

Industry view · Semiconductors

The picks-and-shovels trade became the whole gold rush

Semiconductors stopped being a cyclical commodity business and became the physical chokepoint of the AI economy. Global chip sales hit $772 billion in 2025 and are tracking toward roughly $975 billion in 2026, with essentially all of the upward revision driven by AI demand while non-AI end markets stayed soft.

$772B
2025 global semiconductor sales (+22.5%)
WSTS/SIA
$193.7B
Nvidia FY2026 data-center revenue
Nvidia 10-K
~$500B
Est. 2026 AI-chip market
Deloitte
71%
TSMC global foundry share, Q3 2025
Counterpoint

01 · The thesis

Compute is the new supply chain, and it runs through three or four companies

The AI cycle has concentrated value, not spread it. A single foundry (TSMC) fabricates the accelerators for all five US hyperscalers plus Broadcom; a single equipment maker (ASML) holds 100% of EUV lithography; and a memory triopoly (SK Hynix, Samsung, Micron) controls the high-bandwidth memory that every AI chip needs — with all 2026 HBM capacity sold out. The binding constraint is no longer wafer starts but advanced packaging: TSMC's CoWoS lines have been fully booked with 52-78 week lead times even as capacity roughly tripled in two years.

The second-order story is that AI is now designing the chips that run AI. Synopsys and Cadence have pushed reinforcement-learning and agentic tools into place-and-route and verification, with customers reporting 5-10x productivity gains and AI methods now touching 20-40% of leading-edge designs. That compresses design cost and time-to-market — and lowers the barrier for hyperscalers building custom silicon to escape Nvidia's margin.

1Design / EDA

AI designs the silicon now

Reinforcement-learning and generative tools automate floorplanning, optimization and verification, cutting workflows from days to minutes.

Synopsys DSO.ai, Cadence Cerebrus
2IP / Architecture

Custom beats general-purpose

Hyperscalers commission bespoke accelerators (XPUs/TPUs) to cut cost and dependence on Nvidia, reshaping who captures design value.

Broadcom, Marvell, Arm, Annapurna
3Manufacturing

The physical chokepoint

Leading-edge logic (3nm/2nm) and AI accelerator volume concentrate in one foundry, making capacity allocation a geopolitical lever.

TSMC, Samsung Foundry, Intel
4Memory

HBM is the new bottleneck

AI accelerators are memory-bound; HBM supply is sold out through 2026 and pricing is in a structural shortage.

SK Hynix, Samsung, Micron
5Packaging / Equipment

Advanced packaging gates everything

CoWoS interposers and EUV/High-NA tools are the gating capacity; lines stay booked even as output expands fast.

TSMC CoWoS, ASML, Applied Materials
Pace of AI disruption by stage — Divergent Compute assessment

02 · Public players & exposure

Who routes through, who gets routed around

We plot the listed players on two editorial axes — how exposed each is to AI disruption, against how ready its data, brand and position are to be the answer. The figures in the table are sourced; the placement is our read.

Positioning — editorial assessment, not a sourced metric. Bubble = approximate relative scale.
CompanyStanceThe sourced fact
NvidiaNVDADominant incumbentFY2026 revenue hit $215.9B (+65%) with data-center revenue of $193.7B and 71.1% full-year gross margin; holds roughly 80-90% of the AI accelerator market (Nvidia FY2026 results; Counterpoint/Silicon Analysts).
TSMCTSMIndispensable foundryQ3 2025 revenue of $33.1B (+30.3% YoY) with HPC at 57% of sales and ~71% global foundry share; AI revenue ~$30B, about 24% of 2025 sales (TSMC investor materials; Counterpoint).
ASMLASMLEUV monopolyRecord €32.7B net sales in 2025 with 100% EUV market share and 2026 guidance raised to €36-40B on AI-driven demand (ASML Q4 2025 results).
BroadcomAVGOCustom-silicon kingFY2025 AI revenue of $20B (+65%); Q1 FY2026 AI revenue $8.4B (+106%) with a $73B backlog and a stated target above $100B in AI chip revenue by 2027 (Broadcom earnings; Tom's Hardware).
SK Hynix000660.KSHBM leaderRecord 2025 revenue of 97.1T won (+~50%) and 47.2T won operating profit (doubled YoY); ~57-62% HBM market share with all 2026 capacity sold out (CNBC; Counterpoint).
AMDAMDCredible #2Q3 2025 record revenue of $9.2B (+36% YoY); Instinct MI-series generated an estimated $7-8B in 2025 for roughly 5-7% accelerator share, with an OpenAI multi-gigawatt deal signed (AMD earnings; Silicon Analysts).
MicronMUMemory thirdHolds ~21% of HBM revenue share, expects to sell out 2026 HBM capacity and guided toward an ~$8B HBM annualized run-rate (Counterpoint; Reuters).
IntelINTCTurnaround bet2025 revenue of $52.9B (roughly flat); 18A ramped into volume but external foundry revenue was only $307M, and custom AI processors reached just a >$1B run-rate against a $100B TAM (Intel 10-K/annual report; Investing.com).
CerebrasCBRSWafer-scale challengerRaised $1.1B at an $8.1B valuation in 2025, then priced a 2026 IPO at $185/share raising $5.55B at a fully diluted valuation near $56B (IntuitionLabs; Metaintro/EE Times).
GroqGROQ-pvtInference upstartRaised a large 2025 round lifting its valuation to about $6.9B on an LPU architecture built for low-latency inference (IntuitionLabs).
The map is Divergent Compute’s editorial positioning, offered as a lens, not a measurement. Every figure in the right-hand column is drawn from a named source — see Sources.

03 · The two clocks

The spend, and the payoff

Three timers running against the sector — and against each other.

Sources: Nvidia FY2026 10-K; Broadcom earnings; TSMC investor materials; WSTS/SIA. Figures are latest disclosed annual or run-rate.

Capacity clock: TSMC's CoWoS advanced-packaging demand roughly tripled from ~370,000 wafers in 2024 to ~670,000 in 2025 and is approaching ~1.0 million in 2026, yet lines stay booked with 52-78 week lead times — the gating constraint on every AI accelerator (Silicon Analysts).

Memory clock: HBM is sold out through 2026, DRAM market revenue grew 260% YoY in Q1 2026, and analysts warn of a memory pricing shortage that could persist for years as AI data centers absorb supply (Counterpoint; Deloitte).

Concentration clock: Nvidia's data-center revenue grew from $47.5B (FY2024) to $193.7B (FY2026) in two years, now ~90% of company revenue — even as hyperscaler custom ASICs, a ~$30B market growing 30%+ annually, emerge as the faster threat (Nvidia 10-K; JPM).

04 · Private flagships

The AI-native challengers

The companies attacking this industry AI-first, with disclosed funding where available:

Nvidia

AI compute platform

Blackwell and the coming Rubin generation anchor the accelerator market; Jensen Huang cites $500B+ in booked Blackwell/Rubin orders and partnerships to deploy 10GW for OpenAI alone.

Public (NVDA); FY2026 revenue $215.9B, net income $120.1B

TSMC

Leading-edge foundry

Mass-producing 2nm and ramping a $165B US expansion while managing the geopolitical exposure of fabricating nearly all advanced AI silicon in Taiwan.

Public (TSM); 2025 AI revenue ~$30B, ~71% foundry share

Broadcom

Custom ASIC architect

Co-designs TPUs/XPUs for Google, Meta, OpenAI and Anthropic; signed a Google supply agreement running through 2031 and a 10GW OpenAI accelerator deal.

Public (AVGO); $73B AI backlog, $100B 2027 target

Cerebras

Wafer-scale systems

Pursues training and inference with the largest single chip ever built; 2026 IPO was the year's biggest tech debut, with shares up 68% on day one.

IPO priced $185/sh, $5.55B raised, ~$56B FD valuation

Groq

Inference accelerator

Deterministic LPU architecture targets ultra-low-latency LLM serving, the highest-volume recurring AI workload as inference overtakes training in spend.

Private; ~$6.9B valuation (2025 round)

Tenstorrent

Open RISC-V challenger

Led by chip-design veteran Jim Keller, it bets on an open, CUDA-free RISC-V plus AI-accelerator stack manufactured on Samsung 4nm.

Private; ~$2.6-3.2B valuation (Series D)

05 · Signals

What moved, and what to watch

2025

AI absorbs the entire forecast revision

WSTS raised the 2026 global chip outlook by ~$175B in December 2025 — driven entirely by AI demand, with weakness in non-AI markets (Deloitte).

Oct 2025

Anthropic-Google-Broadcom mega-deal

Anthropic committed to up to 1M TPU chips, then a further multi-gigawatt TPU deal for 2027, the single largest external AI-ASIC arrangement to date (Hashrate Index).

Jan 2026

SK Hynix overtakes Samsung on profit

SK Hynix posted 47.2T won 2025 operating profit, beating Samsung for the first time, as HBM shortages pushed memory pricing into a structural up-cycle (CNBC).

H2 2025

2nm enters high-volume manufacturing

TSMC began N2 production with monthly capacity scaling toward 200,000 wafers by 2028; 2nm orders surged despite ~$30,000 wafer pricing (DigiTimes/Jukan).

2026

Big debuts reopen the chip-IPO window

Cerebras' blockbuster IPO and Groq's mega-round signaled the market is again ready to fund large semiconductor challengers (EE Times).

06 · The exposure read

Who’s defensible, who’s at risk

AI rewards clean, structured advantage and punishes friction. The line runs through who owns the data, the brand and the customer — and who is merely a step the technology can route around.

Defensible

  • The chokepoint owners — TSMC, ASML and the HBM triopoly capture structural rents because their capacity is the binding constraint; demand can be satisfied only as fast as they expand.
  • Nvidia, for now — full-stack lock-in (CUDA, NVLink networking, rack-scale systems) and 71% gross margins let it monetize the build-out even as customers seek alternatives.
  • Custom-silicon enablers — Broadcom and Marvell turn every hyperscaler's escape attempt from Nvidia into their own multi-year ASIC revenue, with Broadcom targeting $100B by 2027.
  • EDA incumbents — Synopsys and Cadence sit on a >90%-share oligopoly that gets more valuable as AI makes chips more complex and as AI-designed-AI accelerates the design refresh cycle.

At risk

  • Intel — 18A reached volume but as largely an internal product; external foundry revenue of $307M and a sub-$1B custom-AI run-rate leave its turnaround dependent on design wins it hasn't yet landed.
  • Pure-play accelerator startups — Groq, SambaNova and peers face Nvidia's software moat and hyperscaler in-house silicon simultaneously; SambaNova's down-round to ~$2.2B shows valuations can compress fast.
  • The whole sector, if the build-out pauses — Nvidia at ~90% of revenue from data center and HBM sold out on AI demand means a slowdown in hyperscaler capex would hit foundry, memory and equipment at once.
  • Geographic concentration — with leading-edge AI silicon fabricated almost entirely in Taiwan, a single supply shock or export-control escalation threatens the entire downstream AI economy.
The semiconductor industry has bound its fortunes to a single demand vector. That made 2025-2026 the best two years in its history — and concentrated the risk into the same handful of chokepoints. Whoever controls advanced packaging, EUV and HBM controls how fast AI can scale; everyone downstream, Nvidia included, builds on rented constraints.

Sources

Where this comes from

WSTS / SIA — 2025 global sales $772B, 2026 ~$975B  ·  Deloitte — 2026 Global Semiconductor Industry Outlook (AI-chip ~$500B)  ·  Nvidia — Q4 & Fiscal 2026 financial results ($215.9B, $193.7B DC)  ·  Nvidia — Q3 Fiscal 2026 results  ·  Silicon Analysts — AMD vs Nvidia AI GPU market share 2026  ·  TSMC at the center of the AI boom — Q3 2025 (LongYield)  ·  Silicon Analysts — TSMC CoWoS sold out, 2nm booked  ·  ASML — Q4 2025 financial results (€32.7B, 100% EUV)  ·  Tom's Hardware — Custom AI ASIC state of play (Broadcom, TPU, MTIA)  ·  Yahoo/Motley Fool — Broadcom FY2025 AI revenue $20B (+65%)  ·  CNBC — SK Hynix record 2025 profit, overtakes Samsung  ·  Counterpoint — Global DRAM & HBM market share  ·  Intuition Labs — Cerebras vs SambaNova vs Groq (funding/valuations)  ·  Cadence — Chip design AI inflection point (5-10x productivity)