Sovereignty is more than chip design

The stack, layer by layer

Useful domestic AI compute is advancing faster than full supply-chain independence.

01

Chinese design

Who designs and controls the chip architecture and software?

02

Domestic fabrication

Where are the compute dies, memory, and packages manufactured?

03

Replenishable supply chain

Can production expand without foreign tools, components, materials, or replacement parts?

The dependency map

Where catch-up comes first

The Observatory’s forecasts, dated 2026-09-11. These are judgments with uncertainty, not reported milestones.

Focused inference

2026–2029

Substantial domestic substitution and selective cost/performance wins on established workloads.

Confidence: High domestically; medium globally

HBM and advanced packaging

2027–2030

Useful domestic volume can improve before parity with the fastest contemporary memory.

Confidence: Medium

Frontier pretraining

Late 2020s

Viable domestic training expands, while efficiency, software flexibility and reliability remain disadvantages.

Confidence: Medium

EUV and frontier fabrication

Into the 2030s

First successful chips can precede commercially competitive manufacturing by years.

Confidence: High on direction; low on dates

Evidence that would change the outlook

  • Customer-qualified domestic HBM at scale
  • Domestic scanners publishing sustained yield, overlay, throughput, and uptime
  • Independent large-cluster training results with time-to-quality and energy use