In partnership with
Anthropic, Amazon, and Google redraw the compute map
TL;DR
Anthropic committed over $100 billion to AWS over 10 years in exchange for up to 5 gigawatts of compute, while Amazon invests $5 billion now with an option for $20 billion more. Days later, Google announced plans to invest up to $40 billion in Anthropic at a $350 billion valuation, with $10 billion upfront and $30 billion tied to performance milestones.
How Jennifer Aniston’s LolaVie brand grew sales 40% with CTV ads
The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.
Key Takeaways
Anthropic's AWS deal specifically calls out Trainium2 through Trainium4 chips, with the option to buy future generations, positioning Amazon's custom silicon as a real alternative to NVIDIA.
Google's investment follows the release of Anthropic's Mythos model, a limited-release system with cybersecurity applications that has already been accessed by unauthorized users.
Anthropic has faced widespread complaints about Claude usage limits, pushing the company to stack infrastructure deals, including a recent CoreWeave agreement.
OpenAI is running a parallel playbook, locking in multi-hundred-billion-dollar deals with cloud providers, chip suppliers, and energy providers, including an expanded Cerebras deal worth over $20 billion.
Why It Matters
Frontier AI is now shaped as much by access to infrastructure as by model quality. Anthropic's willingness to commit nine figures in capital over a decade signals that long-term access to compute has become the single most important strategic asset in the AI race. The companies at the frontier are the ones that can secure chips, power, networking, and cloud capacity at a global scale.
For entrepreneurs, the second-order effect is what matters. As hyperscalers lock up multi-gigawatt capacity with frontier labs, pricing power shifts. Expect API costs, model access tiers, and enterprise pricing to reflect these capital commitments in the coming quarters.
📰 In the News
Headlines & Launches 📣
OpenAI released GPT-5.5, calling it the "smartest and most intuitive" model yet. Co-founder Greg Brockman framed it as "a faster, sharper thinker for fewer tokens" and another step toward the company's long-rumored super app combining ChatGPT, Codex, and an AI browser. The update improves multi-step task reliability, reducing the back-and-forth prompting that earlier versions required.
Ramp platform data across 50,000+ U.S. businesses shows 35.2% pay for OpenAI versus 30.6% for Anthropic as of March 2026, a spread of just 4.5 points. In January 2025, that gap was nearly 13 points. Anthropic grew more than sevenfold in 14 months, driven by the adoption of Claude Code and Cowork across enterprises. Google's AI products remain stuck between 3% and 4.5% despite heavy investment.
Former DeepMind reinforcement learning lead David Silver raised $1.1 billion at a $5.1 billion valuation for Ineffable Intelligence, backed by Sequoia and NVIDIA. The lab aims to build a "superlearner" that discovers knowledge through trial and error without human data, extending the approach Silver used on AlphaZero. The company's stated ambition compares its eventual breakthrough to Darwin's theory of evolution.
At Google Cloud Next 2026, Google announced Workspace Intelligence, a new layer that gives Gemini access to organizational context across documents, emails, meetings, and third-party connectors. Users can build reports and forecasts, draft presentations from multiple data sources, and execute multi-step, multi-app workflows. It's Google's clearest pitch yet for enterprises considering a move away from Microsoft 365.
10x the context. Half the time.
Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.
Hot New Tools 🧰
Skye, built by Signull Labs, is reimagining AI on smartphones by replacing the chatbot interface with iOS widgets that deliver ambient intelligence. The app drafts email replies, preps meetings, sends reminders, flags suspicious bank charges, and surfaces location-specific recommendations based on user-authorized data connections. Signull Labs raised $3.58 million in pre-seed funding at a $19.5 million post-money valuation, with tens of thousands of users already on the waitlist before public launch.
OpenAI activated cost-per-click advertising in ChatGPT, with bids ranging from $3 to $5 per click. The minimum spend dropped from $250,000 to $50,000, and a self-serve ads manager is now in testing with a small group of advertisers. The shift came after CPMs collapsed from $60 at launch to $25 within 10 weeks, leaving pilot advertisers unable to spend their budgets. The lower barrier opens ChatGPT ads to a much wider pool of businesses.
Copy.ai was acquired by Fullcast in October 2025 and is no longer an AI copywriting tool. The product is now a workflow automation platform for revenue operations teams, with use-case pages aimed at Siemens, Lenovo, and ServiceNow. Founders and solopreneurs who relied on Copy.ai's fast, cheap copywriting workflows will need to migrate. The teardown walks through the new pricing, the product's actual capabilities, and an alternative called Wovly, positioned for early-stage builders.
Research & Innovation 🧪
A new IAB report finds that between 60% and 75% of marketers believe current measurement frameworks can't keep up with AI-driven ad platforms. The findings come as ChatGPT shifts to CPC, Microsoft builds ad infrastructure for AI agents, and attribution becomes increasingly opaque across agentic commerce surfaces. For performance marketers, the gap between spend and verifiable outcomes is widening.
Miscellaneous 🎁
The Consumer Federation of America filed a $16 billion lawsuit against Meta, alleging the company profited from fraudulent ads on its platforms. The case lands as ad platforms restructure around AI-generated creative and agentic targeting, raising fresh questions about platform liability for content generated or amplified by AI systems.
Microsoft is developing an ad infrastructure designed for AI agents, not humans, as part of its push into agentic commerce. The bet is that as consumers delegate purchasing decisions to AI assistants, advertisers will need new formats, targeting signals, and measurement tools built for agent-to-agent interactions. It's an early signal of how deeply AI agents may reshape the $700B+ digital advertising market.
See my curated list of all recommended newsletters here 👇
Thanks for reading,
— Cagri Sarigoz
P.P.S. Don't Miss Out on These Resources:
🤝 Follow me on X and LinkedIn for regular updates and insights.
Automate Your Newsletter using HeyNews (50% off early bird discount inside).
