Sam Altman says OpenAI has $20B ARR and about $1.4 trillion in data center commitments

Key Takeaway: OpenAI says it doesn’t want government guarantees for private datacenters, backs government-owned AI infrastructure and U.S. chip fabs, and defends a massive compute buildout to meet surging demand and unlock scientific breakthroughs.

More Insights:

• Rejects “too big to fail”: if OpenAI falters, the market should replace it—no taxpayer backstops.

• Supports a government-owned “strategic reserve” of compute where public benefit (and upside) flows to the state.

• Open to loan guarantees only for U.S. semiconductor fabs to secure an American supply chain—not for private datacenters.

• Funding plan: >$20B ARR run rate by year-end; aiming for hundreds of billions by 2030; exploring AI-cloud sales plus potential equity/debt raises to support ~$1.4T over 8 years.

• Urgency over caution: compute scarcity is throttling features today; investing now prepares for near-term AI advances, including scientific discovery, and a future of abundant, cheaper AI.

Why it matters: This frames AI infrastructure as national strategy, not corporate entitlement—pushing governments to own core compute while leaving companies to market risk; the sheer scale hints AI will be as foundational as power grids or the internet, and policy choices made now will determine who controls capability, safety, and access in the next economy.

Discover How to Save 10+ Hours Per Week Using AI as Your Personal Assistant

Are you drowning in tasks while dreaming of a way to clone yourself? Your AI productivity breakthrough has arrived.

• Get the exact templates used by top performers to delegate 80% of their routine work to AI

• Access our "AI Assistant Command Center" - a ready-to-use system for managing all your AI tools

• Master AI delegation in under 60 minutes with our step-by-step implementation guide

• Track your time savings with our built-in ROI calculator

• Leverage advanced prompts that turn ChatGPT into your 24/7 productivity partner

Imagine completing a full day's work by lunch, knowing your AI assistant handles the rest. Over 10,000 professionals have already transformed their productivity using these templates.

Microsoft forms SuperIntelligence team under AI chief Suleyman ‘to serve humanity’

Key Takeaway: Microsoft has launched the MAI Superintelligence Team led by Mustafa Suleyman to push advanced, practical AI—focused on digital companions, medical diagnostics, and renewable energy—while stressing that the work will be constrained, safe, and explicitly aimed at serving humanity.

More Insights:

• Mustafa Suleyman — former DeepMind co-founder and Inflection CEO — will lead the new MAI Superintelligence Team.

• The group will target practical breakthroughs: better digital companions (Bing/Copilot), expert-level medical diagnostics, and optimizations for renewable energy.

• Microsoft frames the effort as “grounded and controllable,” rejecting vague, unrestricted notions of superintelligence.

• The move follows hires from Inflection and sits alongside Microsoft’s multi-supplier strategy (OpenAI, Google, Anthropic) and its large Azure/OpenAI investment ties.

• Suleyman emphasized caution amid investor worries about heavy AI spending — asserting Microsoft won’t pursue superintelligence “at any cost.”

Why it matters: This is a pivotal moment where a major tech giant publicly ties high-ambition AI research to concrete social applications and governance rhetoric—if Microsoft delivers, it could accelerate powerful improvements in healthcare, education, and energy, but it will also concentrate technical, ethical, and commercial power in ways that will shape regulation, competition, and who ultimately decides how far “helping humanity” goes.

China's open-source AI closes the gap

Key Takeaway: Moonshot AI’s open-source Kimi K2 Thinking claims near-frontier reasoning and tool-use performance—sometimes topping GPT-5 and Claude 4.5 Sonnet—while being dramatically cheaper to train and run.

More Insights:

• Outperforms GPT-5 and Sonnet 4.5 on several agentic benchmarks; sets a 44.9% record on Humanity’s Last Exam.

• Major coding gains vs. its 4-month-old predecessor, though still a hair behind the very top models.

• Can autonomously chain 200–300 tool calls to complete complex tasks; also strong in creative writing.

• Reported training cost under $5M; pricing positioned well below current frontier models.

• Underscores claims that China is “nanoseconds” behind—marking the closest open-source and Chinese labs have come to the frontier.

Why it matters: If near-frontier capability becomes open-source and inexpensive, the moat shifts from raw model quality to data, distribution, safety, and integration—accelerating agent adoption, squeezing closed-model premiums, and raising urgent questions about governance and misuse at scale.

Other stuff

• Why Sam Altman Won’t Be On The Hook For OpenAI’s Massive Spending Spree

• Oddest ChatGPT leaks yet: Chat logs found in Google Analytics tool

• Hard drives on backorder for two years as AI data centers trigger HDD shortage

• China bans foreign AI chips from state-funded data centres

• Nested Learning: A new ML paradigm for continual learning

• The Company Quietly Funneling Paywalled Articles to AI Developers

• Google Maps releases new AI tools that let you create interactive projects

• Elon Musk uses Grok to imagine the possibility of love

Hope you enjoyed today's newsletter

⚡️ Join over 300,000 people using the Superpower ChatGPT extension on Chrome and Firefox.