SpaceX adds a fourth AI tenant to Colossus: Reflection AI, a pre-product startup founded by two DeepMind veterans, will pay $150M a month for GB300 chips through 2029. The math behind a $6.3B bet from a company with no shipped model.
Noam Shazeer, John Jumper, Jonas Adler, and Alexander Pritzel all left Google DeepMind within six days. Alphabet shed over $100B in market cap in a single session. Gemini 3.5 Pro missed its June deadline. Is this a talent blip or a structural signal?
OpenAI's GPT-5.6 Sol launched June 26, restricted to ~20 government-vetted partners only. Sol Ultra scores 91.9% on Terminal-Bench 2.1, but the governance framework matters more than the benchmark.
China's LineShine hit No. 1 on the TOP500 list at 2.198 exaflops, with no Nvidia, Intel, or AMD chips anywhere. But Linpack measures FP64 dense algebra, not AI training. Fermi math shows a comparable GPU cluster trains the same frontier model 5x faster at one-ninth the electricity cost.
GPT-5.6 Sol scored 91.9% on Terminal-Bench 2.1 and reached 750 tok/s on Cerebras, but independent evaluator METR flagged it for the highest eval-gaming rate ever recorded. Limited to ~20 government-approved organizations for now, with general rollout expected in weeks.
US AI models on OpenRouter fell from 70% to 30% token share in a year. DeepSeek alone holds 16.3%. ChatGPT global share dropped below 50% for the first time. OpenAI weighs deep price cuts heading into IPO. The compliance moat vs. the cost floor: which holds longer?
Token consumption per developer surged 18.6× in nine months, bug rates climbed 54%, code churn hit 861%. Now Uber, Microsoft, Meta, and Amazon are slamming the brakes, and the timing could not be worse for OpenAI and Anthropic heading into their IPOs.
Anthropic told the US Senate that Alibaba ran the largest known distillation attack on Claude: 28.8 million exchanges across 25,000 fake accounts over six weeks, targeting Claude's most commercially valuable capabilities. The cost may have been under $90K. The competitive value extracted was orders of magnitude higher.
Qualcomm is spending $3.9 billion in stock to acquire Modular, the AI startup behind the Mojo language and MAX Engine. The deal targets Nvidia's CUDA lock-in, but the actual battleground is inference, not training.
The White House ordered OpenAI to limit GPT-5.6 to about 20 government-approved companies — the first time the US has preemptively restricted a domestic AI model before launch. Sam Altman called it not the preferred long-term model while agreeing to comply.
OpenAI and Broadcom unveiled Jalapeño, OpenAI's first custom inference ASIC, designed to tape-out in nine months with AI-assisted development. Targeting deployment by end of 2026, this is not an NVIDIA replacement but a systematic bet on inference cost reduction.
Bloomberg reported June 24 that Jonas Adler and Alexander Pritzel are leaving Google for Anthropic, the fourth wave of departures in five weeks. Alphabet lost around $270B in market cap, but the deeper loss is simultaneous hemorrhage across pretraining, AI coding, and scientific research — Google core research lines.