415.tech
AI & tech, from the frontlines of Silicon Valley

Past editions

The last seven days of 415.tech, newest first. Older briefs roll off after a week — subscribe so you never miss one.

This week
Thu
11
Jun
● Today's edition
Apollo leads $35B private financing for Broadcom's AI compute platform, targeting 20GW by 2028
Plus: Editorial polish — DiffusionGemma · Google ships Gemini 3.5 Live Translate — continuous speech-to-speech across 2,000+ language pairs
13 stories
Tue
9
Jun
Editorial polish: Anthropic Fable 5 / Mythos 5
Plus: Apple builds new Siri on Google Gemini at $1B/year as Tim Cook names September successor · Google orders 3M+ TPUs from Intel for 2028, the first confirmed large-scale foundry diversification from TSMC
10 stories
Mon
8
Jun
Apple rebuilds Siri on a 1.2-trillion-parameter Gemini model and opens iOS 27 to third-party AI defaults
Plus: NVIDIA and SK hynix commit to co-develop memory for Vera Rubin, RTX Spark, and Jetson Thor · OpenAI says 'chat is dead' as it rebuilds ChatGPT into an agent superapp
7 stories
Sat
6
Jun
Google pays SpaceX $920M a month for GPU bridge capacity, citing unexpected Gemini Enterprise demand
Plus: OpenAI's Dreaming V3 makes ChatGPT memory fully automatic, extending it to free-tier users · Apple rebuilds Siri on a custom 1.2T-parameter Gemini model at $1B per year
9 stories
Thu
4
Jun
Alphabet's record $84.75B equity raise closes oversubscribed, with Berkshire buying $10B
Plus: Microsoft's MAI-Thinking-1 scores 97% on AIME 2025 with no OpenAI distillation · Cloudflare acquires VoidZero — Vite, Rolldown, and Oxc stay MIT-licensed and vendor-agnostic
10 stories
Wed
3
Jun
SoftBank pledges €75B to build Europe's largest AI data center cluster in France
Plus: Japan commits $500M to the US Genesis Mission as its first international partner · MiniMax M3 scores 59% on SWE-Bench Pro, surpassing GPT-5.5, as the first open-weight model with 1M-token context
14 stories
Tue
2
Jun
Anthropic files confidential S-1 at a $965B valuation, topping OpenAI's private mark
Plus: Trump's Frontier AI Order: Voluntary 30-Day Review After Labs Killed the Mandatory 90-Day Draft · Microsoft's MAI-Thinking-1 matches Claude Opus 4.6 on coding — trained without OpenAI data
9 stories