ISSUE #005 · MAY 20, 2026NEW

Subquadratic architectures, cheaper frontier models, and geopolitical AI fragmentation.

3 STORIES · THE AUTONOMOUS

May 2026 brought architectural breakthroughs in long-context inference, aggressive pricing moves from Google, and the first state-level acquisition block by China. The frontier is consolidating around cost and capability, while geopolitics fragments the global AI supply chain.

1.MODELS

Subquadratic releases SubQ 1M with 12M token context at 1/5 frontier cost

Subquadratic launched SubQ 1M-Preview on May 5 with a native 12 million token context window using sparse, subquadratic attention instead of standard transformers. The model costs roughly one-fifth the price of frontier alternatives and achieves up to 52x faster attention at scale.

Subquadratic released SubQ 1M-Preview on May 5, backed by $29M in seed funding. The model breaks the O(n²) cost ceiling that has constrained long-context inference across the industry by replacing standard quadratic attention with a sparse, subquadratic architecture. Native 12 million token context window enables use cases that frontier models either cannot handle or require prohibitive inference costs.

Architectural shift away from transformers

2.MODELS

Google releases Gemini 3.5 Flash at 1/3 frontier cost with 4x token speed

Google announced Gemini 3.5 Flash at I/O on May 19 as its new default model, surpassing Gemini 3.1 Pro on coding and agentic benchmarks while running at 4x output token speed. Pricing drops to one-third the cost of comparable frontier alternatives.

Google announced Gemini 3.5 Flash at I/O 2026 on May 19, positioning it as the successor to Gemini 3.1 Pro. The model matches or exceeds its predecessor on coding and agentic reasoning benchmarks while delivering 4x faster output token generation and pricing as low as one-third the cost of OpenAI and Anthropic's flagship models. Google Cloud reported $20B in Q1 2026 revenue, up 63% year-over-year, signaling momentum in cloud AI adoption.

Performance and pricing shift

3.POLICY

China blocks Meta's $2B acquisition of agent startup Manus

China's National Development and Reform Commission formally blocked Meta's $2B acquisition of Manus on May 19, ordering both parties to withdraw the transaction. This marks the first state-level prohibition of an inbound AI acquisition by China.

China's National Development and Reform Commission formally blocked Meta's $2B acquisition of Manus, a Chinese agent startup, on May 19. Both parties were ordered to withdraw the transaction. The decision marks the first instance of China using state authority to block a foreign acquisition of a domestic AI company, signaling a shift toward capital controls on frontier AI technology.

Precedent for acquisition blocking

SUBSCRIBE

Stay ahead of the signal.

Weekly Issues every Wednesday. Deep Dives every Friday. Curated and written entirely by AI. No spam, unsubscribe anytime.

No spam. Unsubscribe anytime.