Hi AI Futurists,
Mira Murati spent seven years at OpenAI, most of them as CTO. Today her new AI lab signed its first major cloud deal, with Google. What does this mean for the state of AI? Let's take a look.
Our agenda.
Top AI news
Mira Murati chooses Google
3 AI tools to boost your workflow
AI in the wild
AI Investment Report
Best,
Lex Sokolin
P.S. Enjoying the newsletter? Have a suggestion? Hit reply and let us know!
Manage your settings: Share | Unsubscribe | Upgrade
Ecosystem: AI Venture Fund | Fintech Research | Lex Linkedin / Twitter

The IT strategy every team needs for 2026
2026 will redefine IT as a strategic driver of global growth. Automation, AI-driven support, unified platforms, and zero-trust security are becoming standard, especially for distributed teams. This toolkit helps IT and HR leaders assess readiness, define goals, and build a scalable, audit-ready IT strategy for the year ahead. Learn what’s changing and how to prepare.

Top AI news
🧠 Google Debuts Eighth-Gen TPUs Sunfish and Zebrafish at Cloud Next (9to5Google). Sunfish trains, Zebrafish serves, and a new Workspace Intelligence layer reads every Gmail thread and Doc you own. The silicon and the software finally share a sales team.
👷 OpenAI Rolls Out Codex-Powered Workspace Agents for Teams (9to5Mac). Shared agents that ship code, write reports, and answer Slack inside your company's permissions. The agent wars moved from consumer to enterprise overnight.
🌐 Chrome Lands Gemini Auto Browse for the Enterprise (TechCrunch). Gemini reads every open tab and does the work you were about to do. You still click "are you sure."
💰 OpenAI Commits $1.5B to a Private-Equity Enterprise Venture (PYMNTS). $500M now, $1B later, and a 17.5% guaranteed return for TPG, Bain, and Brookfield. OpenAI is paying private equity to install ChatGPT in every portfolio company they own.
🚀 SpaceX Gets $60B Option to Acquire Cursor (TechCrunch). Either $10 billion for the partnership or $60 billion for the company. Cursor gets Colossus compute. Musk gets a coding model ahead of the SpaceX IPO. Somebody drew this up on a napkin.
🤖 Perplexity's Computer Agent Arrives for Enterprise (VentureBeat). 20 orchestrated models, Slack and Snowflake connectors, SOC 2 stamped. The company that was supposed to disrupt Google search is now going after Microsoft's lunch.
⚖️ Nebraska Supreme Court Suspends Lawyer Over Fabricated AI Citations (Nebraska Public Media). 20 fake cases, 4 that don't exist in any court anywhere, one starring "Kennedy v. Kennedy." The first AI hallucination to actually end a career.
📊 Stanford Drops the 2026 AI Index (IEEE Spectrum). Global AI compute capacity has grown 3.3x every year since 2022. Grok 4 alone emits 72,000 tons of CO2 per training run. The stack is growing faster than the planet can cool.
💸 AI Captures 80% of Q1 2026 Venture Funding (Crunchbase News). $242 billion of the $300 billion in global VC went to AI last quarter. Four companies took $188 billion of it. The word "concentration" is doing heavy lifting.
🦾 NVIDIA and Adobe Bolt Agents Onto Photoshop (Diginomica). Jensen Huang on the Adobe Summit stage: "Agentic AI isn't coming. It's here." The Creative Cloud is getting a coworker. Nobody asked the designers.

Mira Murati goes with Google

Mira Murati spent nearly seven years at OpenAI, most of them as Chief Technology Officer. She ran the ChatGPT launch. She served as interim CEO during the 48 hours Sam Altman was fired in November 2023. She was the face of OpenAI in front of Congress, the press, and most of its largest enterprise customers. She left in September 2024.
Five months later she came back with Thinking Machines Lab. It raised $2 billion at a $12 billion valuation, one of the largest seed rounds in startup history, and filled its roster with former OpenAI researchers. TML's first product, Tinker, shipped in October. It is a tool for spinning up custom frontier models using reinforcement learning. Today TML signed its first major cloud deal, and it is with Google, of all places. The founder who spent a decade telling the world OpenAI would beat Google to AGI now pays Google for compute.
Earlier this month, Anthropic signed a 3.5-gigawatt TPU deal with Google and Broadcom. Anthropic's annual run rate tripled to $30 billion in six months. Every dollar of that growth meters through a Google chip. Put those deals next to today's Cloud Next reveals: new TPUs, a Workspace Intelligence layer reading across Gmail and Docs, and a $750 million fund pointed at 120,000 partners. The shape of the market is obvious. Google is not racing the model labs anymore. It is the landlord.
The model race and the compute race split a year ago, and most people still haven't noticed. Nvidia has GPUs. Google has TPUs. That is almost the entire market for chips a frontier lab actually needs. If Claude wins, Google gets paid. If Tinker wins, Google gets paid. If enterprises skip the labs entirely and use Gemini, Google still gets paid. The one serious holdout is Meta, which is spending $135 billion this year trying to build internally what everyone else rents. The math is not flattering.
Enterprises used to pick a cloud. Now they pick a lab. The labs all pick Google. The model logo on your agent is branding. The chip underneath is the real relationship. Pay attention to the chip.
Takeaways at a Glance:
Thinking Machines Lab signed a multi-billion-dollar deal with Google Cloud today, its first cloud partnership since launch
TML will train Tinker, its reinforcement learning product, on Google's latest infrastructure
Anthropic's $30B annualized run rate runs on Google TPU capacity, rising to 3.5 gigawatts in 2027
Google announced TPU v8 (Sunfish for training, Zebrafish for inference) and a $750M partner fund at Cloud Next today
Google's TPU supply deal with Broadcom runs through 2031
What We Think About It:
Benchmarks are the storefront. Chips are the factory. Google is quietly running the toll road. Every serious AI company needs more compute than money alone can buy on the open market. The more the labs try to out-model each other, the more they all end up buying from the same two places. The model wars are loud. The plumbing war was over a year ago.
The way this lands in your work is simple. The AI vendor you pick this year is not just a vendor. It is a bet on whose infrastructure survives 2027. If the AI wave pauses or the money reshuffles, the companies with their own chips and their own power stay standing. The rest are renters.
What You Can Do Right Now:
When an AI vendor pitches you, ask who makes their chips and where their power comes from. The answer tells you how durable your integration will be
If you are choosing an AI stack for 2026 to 2027, pick two providers, not one. The market is concentrating, and a second option is how you keep pricing honest
For marketers and content teams, Gemini now does real work inside the browser. Your 2027 SEO playbook starts with what Gemini sees when it lands on your site, not what a human does
For investors, every AI valuation you hear this year has Google compute priced in. The companies that can actually get capacity at scale are the ones worth re-underwriting
For builders, the model is a commodity this year. The workflow you wrap around it is the product. Stop selling "AI." Start selling what your customer is actually trying to finish

Get more done with these AI tools
Stanley For 𝕏. Your AI Head of Content for X.

Ingests your existing posts and writes new ones in your actual voice
Walks you through research, angle, hook, and thread, not just a prompt
Built on ghostwriter frameworks, so the output reads like strategy instead of prompt spam
Cai. A local macOS AI assistant you summon with ⌥C.

Ships with Ministral 3B bundled, so AI works offline with zero setup
Connects to Ollama, LM Studio, Apple Intelligence, or OpenRouter when you want cloud power
Open source and MIT-licensed, so your workflows stay yours
Toki. AI scheduling that turns vague ideas into a real plan.

Voice, text, or screenshots become scheduled events with no time picker required
Remembers your preferences and context, so it books what actually works for you
Runs conditional actions in the background, like pinging you when a flight price drops

AI In the Wild

AI Investment Report
This 158-page research report provides the first comprehensive taxonomy of public companies, private ventures, and tokenized protocols building the infrastructure for autonomous AI systems. Compiled by Lex Sokolin, former Chief Economist at ConsenSys, fintech strategist at Autonomous Research, and current Managing Partner at Generative Ventures, this report delivers institutional-grade analysis of 100+ companies across 14 critical infrastructure layers. Learn more here.

What type of coverage would you like to see most?

That’s all for today, folks!
Reach out to our audience by becoming a sponsor here.
If you’re enjoying the newsletter, share with a friend by sending them this link: 👉 https://www.futureblueprint.xyz/subscribe
Looking for past newsletters? You can find them all here.
Working on a cool A.I. project that you would like us to write about? Reply to this email with details, we’d love to hear from you!


