Semiconductor Fabrication Capacity
The upstream constraint on AI chip supply — foundry (TSMC) wafer output, process-node capacity, and memory (HBM/DRAM) fabrication — which caps how many AI accelerators can be produced regardless of downstream demand, power, or data-center buildout.
Also known as: TSMC, wafer fab capacity, wafer production, foundry capacity, process node, leading-edge node, 3 nanometer, 2 nanometer, lithography equipment, chip fabrication, fab capacity, wafer allocation, HBM production, DRAM wafer capacity
Part of: Compute & Energy Infrastructure
Documents (1)
- Presentation: From Fab To Token - The State Of The Market Jordan Nanos · 2026-08-18 source ↗ Reader 🖥️
The binding constraint on AI's growth is not GPU chip design, data-center power, cooling, construction, or end-user demand — all of which are being solved fast enough not to bind — but TSMC's wafer-fabrication capacity, and because that supply-side constraint (not a collapse in demand or profitability) is what is actually rationing AI compute, the current AI buildout reflects real, constrained demand rather than a speculative bubble.
Propositions touching this concept (6)
By publication month, oldest first (extraction month when no publication date is known).
2026-08
- explicitcontestedTSMC wafer-fabrication capacity — not GPU chip demand, data-center power/construction, or end-user demand — is the binding 'choke point' on AI's growth; every other commonly cited bottleneck is being solved fast enough not to bind.
- explicitTSMC's capital expenditure and wafer output growth lag hyperscaler AI-accelerator spending because TSMC takes a more measured capex approach than the hyperscalers, creating a persistent wafer-supply shortfall.
- explicitAI accelerator demand has reversed the historical order of leading-edge silicon allocation: AI chips, not smartphones/PCs, now get first claim on TSMC's most advanced (3nm, soon 2nm) process capacity, at the expense of smartphone supply.
- explicitNVIDIA will continue to command the overwhelming majority of TSMC's 3nm wafer allocation among AI accelerator vendors, so competitors (AMD, custom silicon, startups) cannot meaningfully take share from NVIDIA regardless of their own product quality, because output is capped by wafer allocation rather than by design competitiveness.
- explicitMemory fabrication capacity is being reallocated to AI at an accelerating rate: AI's share of DRAM wafer capacity is forecast to rise from about 12% in 2023 to over 60% within the next year.
- explicitBy the end of 2028, the industry will have more empty data-center capacity waiting for chips than chips to fill it, because data-center buildout is outpacing TSMC's wafer output growth.
Open gaps (3)
- incompleteThe entire 'not a bubble' argument rests on TSMC's wafer-fabrication constraint persisting, but the talk treats this constraint as a durable structural fact rather than examining its own time horizon — what happens once 2nm ramps, TSMC's Arizona/Japan fabs mature, or Samsung/Intel foundry capacity catches up. If the fab constraint eases while demand growth assumptions don't hold, the 'no bubble' conclusion could reverse quickly.Presentation: From Fab To Token - The State Of The Market — read next: Look for SemiAnalysis or other analyst coverage of TSMC's multi-year capex/capacity roadmap and competing foundry ramp timelines, to assess how durable the fab-capacity bottleneck actually is.
- incompleteEvery named company example is one whose centering choice is, so far, associated with a good outcome (Novartis, Fujifilm, TSMC, Samsung, Toss, DBS, Nvidia, Shopify, early Amazon). Kodak and Southwest appear only as foils establishing other companies' successes, not as tested instances of 'had a clear, deliberately chosen single center and it still failed.' No company that centered and underperformed is discussed, so the base rate of centering's success is unknowable from the article alone.The Power of Strategic Centering — read next: Survivorship-bias literature (Mlodinow's 'The Drunkard's Walk', Taleb) on retrospective success narratives fitted to winners; Rosenzweig's 'The Halo Effect' critique of the 'In Search of Excellence' genre of business writing, which has essentially the same evidentiary structure as this article.
- uninformedNo attempt is made to separate the causal contribution of 'centering' from contemporaneous confounds: Novartis's radioligand bet coincided with a broadly hot oncology/precision-medicine investment cycle; TSMC's and Samsung's ecosystem success is explicitly bound up with decades of targeted state industrial policy (named in the text itself), yet the article still credits 'centering' as the operative variable rather than state support; Fujifilm's successful diversification depended on pre-existing chemical/materials-science IP whose availability to Kodak is never assessed.The Power of Strategic Centering — read next: Regression-to-the-mean and confound-control literature in strategy research; comparative case-study methodology critiques (e.g., Geoffrey Jones or business-history methodology essays) on isolating a single explanatory variable from multi-causal corporate histories.