The Weekly Waypoint, Issue #15
For two years, the AI world had a simple map. OpenAI ran on Microsoft. Anthropic stayed out of the Pentagon. Google kept its models in-house. The UK watched from the sidelines.
This week, every one of those assumptions broke. OpenAI gutted its exclusive deal with Microsoft and showed up on AWS. Google signed a classified Pentagon contract the day after Anthropic got blacklisted. The UK dropped £500M on a sovereign AI fund. And DeepMind's David Silver raised $1.1 billion on the premise that AI doesn't need human data at all.
The map just got redrawn. Let's get into it.
OpenAI Breaks Free: Microsoft Exclusivity Is Over
The biggest infrastructure story of the year, and it barely made a ripple compared to the model launches. OpenAI and Microsoft restructured their partnership, ending Microsoft's exclusive cloud reselling rights. Within 24 hours, OpenAI models were available on AWS. Google Cloud is likely next.
This isn't a breakup. Microsoft still owns a chunk of OpenAI and gets preferred terms. But the exclusive lock is dead. For anyone building on OpenAI APIs, this means more cloud options, better pricing leverage, and less single-vendor risk.
GPT-5.5: The Super App Takes Shape
OpenAI released GPT-5.5 on April 23rd, and the pitch has shifted. It's not "the smartest model" anymore -- it's the model built for doing work. Coding, research, data analysis, computer use, document creation. OpenAI is positioning ChatGPT as an operating system, not a chatbot.
The model is better at complex multi-step tasks, but the real story is the direction: OpenAI wants you living inside ChatGPT, not visiting it.
Google + Pentagon: The AI Military Alliance Reshuffles
After Anthropic refused to work with the Pentagon (and got blacklisted for its trouble), Google stepped in fast. Classified AI contract signed. The Pentagon's AI chief explicitly said they're diversifying away from single-vendor dependency.
This is the beginning of a split in the AI industry: companies that work with the military and companies that don't. Your procurement decisions just got a political dimension.
Ineffable Intelligence: $1.1B to Ditch Human Data
David Silver -- the DeepMind researcher behind AlphaGo, AlphaZero, and AlphaStar -- emerged from stealth with Ineffable Intelligence and a $1.1 billion seed round at a $5.1B valuation. Backed by Sequoia, Lightspeed, Nvidia, and Google.
The bet: AI that learns without human data. No scraping, no licensing, no data wall. If it works, it changes the economics of every AI company on earth.
Harvard Study: AI Beats Doctors at ER Diagnosis
A landmark Harvard study found that an AI model outperformed two human physicians by over 11% on clinical reasoning tasks using real emergency department data. Not a simulation. Real patients, real cases.
This isn't "AI replaces doctors" -- it's "AI becomes the most powerful diagnostic second opinion in history."
UK Drops £500M on Sovereign AI
The UK government launched a £500 million Sovereign AI fund, investing directly in domestic startups. First investments: Callosum (infrastructure) and Ineffable Intelligence. The message: the UK isn't content being a customer of American AI. It wants to build its own.
Quick Hits
- NVIDIA Nemotron 3 Nano Omni -- 30B parameter open multimodal model that handles vision, audio, and text in one model. 9x more efficient for AI agents. Free and open.
- GUARD Act passes Senate committee -- Bipartisan bill to restrict AI chatbot access for kids. Unanimous committee vote. Likely headed to the Senate floor.
- California AI procurement order -- Newsom directs state agencies to develop AI procurement standards with mandatory harm assessments. States are building their own regulation while DC talks.
- House bipartisan AI bill package -- Reps. Lieu and Obernolte introduce consolidated AI legislation from the 2024 task force. Voluntary standards, testing frameworks, incident reporting. A starting point, not a finish line.
The Pattern: Pick Your Alliances
The AI industry isn't one market anymore. It's splitting into factions: US vs. UK vs. EU, military-aligned vs. military-refusing, cloud-locked vs. multi-cloud, human-data-dependent vs. self-learning. Every tool you pick, every API you call, every vendor you sign with -- it's not just a technical decision. It's a strategic one.
Deep Dive This Week
The Great Reshuffle: How AI's Alliances Broke, Reformed, and What It Means for Your Stack
The full breakdown: what the OpenAI/Microsoft split means for cloud strategy, why the Pentagon deal changes vendor selection, how Ineffable's "no human data" thesis could upend the industry, what the UK sovereign fund means for European founders, and a decision framework for navigating the new factional landscape. Pro members get the full deep dive.
Not Pro yet? Upgrade for $6/month
Stay sharp,
WaypointsAI