AsOf

Trends

Compute and infrastructure

Chips, data centres, cloud capacity, inference serving. Counted from the record's checked claims, one column per quarter since 2023. /trends/compute-infra.md

45
2023 Q12026 Q4*
By quarter
2026 Q4*9 of 212 updates (4%)
2026 Q314 of 340 updates (4%)
2026 Q29 of 257 updates (4%)
2026 Q13 of 128 updates (2%)
2025 Q42 of 98 updates (2%)
2025 Q33 of 77 updates (4%)
2025 Q23 of 47 updates (6%)
2024 Q32 of 37 updates (5%)
all quarters2024 Q32025 Q22025 Q32025 Q42026 Q12026 Q22026 Q32026 Q4

1–14 of 14 claims in 2026 Q3, newest first

2026-09-29valid
OpenAI · feature · Ultrafast mode
OpenAI feature: Ultrafast mode, claim clm_7802CD9870BD · valid · 2 sources (2 primary) · conf 0.70
2026-09-29valid
OpenAI · feature · Ultrafast mode for GPT-6 Astra
OpenAI feature: Ultrafast mode for GPT-6 Astra, claim clm_988C1118D629 · valid · 1 source (1 primary) · conf 0.70
2026-09-28unverified
GPT-6 Luna · available in · Amazon Bedrock
GPT-6 Luna available in: Amazon Bedrock, claim clm_EB9AE334E7EC · new claim · 1 source · conf 0.50
2026-09-23valid
Fireworks AI · released · ARCv3
Fireworks AI released: ARCv3, claim clm_0D788A888BB2 · valid · 1 source (1 primary) · conf 0.70
2026-09-21valid
Meta Platforms · launched · Petal
Meta Platforms launched: Petal, claim clm_F46CF3C21DF9 · valid · 1 source (1 primary) · conf 0.70
2026-09-17valid
Amazon EC2 T8i · released · Amazon EC2 T8i
Amazon EC2 T8i released: Amazon EC2 T8i, claim clm_BF5A39F6FB48 · valid · 1 source (1 primary) · conf 0.50
2026-09-08valid
Cohere · released · North Mini Code Megakernel Serving Engine
Cohere released: North Mini Code Megakernel Serving Engine, claim clm_2BDD08F75797 · valid · 1 source (1 primary) · conf 0.70
2026-09-02valid
OpenAI · feature · slow_down
OpenAI feature: slow_down, claim clm_5C1F68F40D60 · valid · 1 source (1 primary) · conf 0.70
2026-08-24valid
Vercel · feature · Turborepo cache hits
Vercel feature: Turborepo cache hits, claim clm_0B22C44837F9 · valid · 1 source (1 primary) · conf 0.70
2026-08-24valid
Nvidia · released · NVIDIA Groq 3 LPX
Nvidia released: NVIDIA Groq 3 LPX, claim clm_AB426FA1C938 · valid · 1 source (1 primary) · conf 0.70
2026-08-13valid
OpenAI · launched · Ultrafast mode
OpenAI launched: Ultrafast mode, claim clm_068B51DDCAAE · valid · 1 source (1 primary) · conf 0.70
2026-07-30valid
OpenAI · feature · Fast mode
OpenAI feature: Fast mode, claim clm_FD20D5B9F714 · valid · 2 sources (2 primary) · conf 0.70
2026-07-27valid
Fireworks AI · available in · US-only Serverless
Fireworks AI available in: US-only Serverless, claim clm_34E3E67BC648 · valid · 1 source (1 primary) · conf 0.70
2026-07-09valid
OpenAI · feature · explicit prompt caching controls
OpenAI feature: explicit prompt caching controls, claim clm_5839E6576586 · valid · 1 source (1 primary) · conf 0.70