Ben Bajarin: 2027 is the peak year of AI compute constraint, supply relief comes 2028
BenBajarin · x · 2026-09-04
Analyst Ben Bajarin publishes his "2027: Peak Year of Constraint" thesis on AI infrastructure supply:
- AI demand sits roughly 115-120% above annual capacity, with bottlenecks moving deeper into the supply chain (foundry capacity, components, power).
- In 2027 demand keeps growing, but capacity funded in 2026 and installed through 2027 won't deliver full-year mature-yield output, creating a visible growth cap.
- By early 2028, foundry/component capacity and data center projects should provide real relief, enabling faster unit growth through 2028-2030.
- Once constraints ease, specific vendors not fully modeled in most scenarios could see outsized re-acceleration.
Bottom line: the industry grows only as fast as its most constrained layer; 2027 is the tightest year, with the opportunity window opening after 2028.
Related event: Analyst: 2027 Will Be the Peak Year of Compute Constraint(3 posts)→
More from Infra
- SGLang merges megakernel low-latency mode, hitting ~1000 tok/s/user on GB300 NVL72 — ying11231 · 2026-10-21
- The topology of an LLM agent: where the model, loop, and tools run shapes cost and security — davidcrawshaw · 2026-10-21
- Chris Lattner shows how MAX and Mojo extend PyTorch across vendor hardware at PyTorchCon 2026 — PyTorch · 2026-10-21
- General Compute partners with Positron AI for cluster-scale deployment, live Q1 2027 — eptwts · 2026-10-21
- MemTensor's MemT-1.0 makes persistent memory part of model compute: 1.86x faster inference, 66x less storage — _akhaliq · 2026-10-21
- LithosAI sweeps 12 #1 spots on Artificial Analysis, GLM 5.3 Flash at 1,543 tok/s — JiaZhihao · 2026-10-21