AI Compute Control Concentrates in Few Hands
In brief
- Two companies capture most revenue from AI-native companies and three American cloud companies control 63% of the infrastructure on which most AI runs.
- NVIDIA holds 85% of the data center GPU market.
- The United States controls approximately 75% of global high-performance AI compute capacity.
- Large language models are trained on English-language data, which means non-English users pay more for lower-quality results.
- The world will soon rely on whoever controls the compute layer to participate in the AI economy.
Terms in this brief
- AI-native
- Refers to companies that are built specifically around AI technologies and infrastructure, rather than being traditional businesses that happen to use AI.
Read full story at Fortune →
More briefs
New AI Agent Hits 140,000 GitHub Stars
Hermes Agent is a new open source agentic framework that has gained popularity quickly. It crossed 140,000 GitHub stars in under three months. This matters because Hermes is designed for reliability and self-improvement. It can run 24/7 and integrate with messaging apps. It also has standout capabilities like self-evolving skills and contained sub-agents. The agent is ideal for running on NVIDIA RTX PCs and workstations. Hermes Agent is now the most used agent in the world and will continue to improve with new updates and models like Qwen 3.6. Next week more users will try Hermes Agent.
AI Toe Scan May Improve PAD Screening
A new technology can detect peripheral artery disease with a seconds-long toe scan. The scan uses a light to measure blood volume in the toe. This matters because 8 to 12 million Americans have peripheral artery disease. The disease can cause limb amputation if not treated early. The new technology may lead to earlier diagnosis and treatment for these patients. It could widen access to screening and help save lives.
ChatGPT Gains Access to Bank Accounts for Personalized Financial Advice
ChatGPT is now allowing users in the U.S. to connect their bank accounts, offering tailored financial insights through Plaid integration. This feature leverages OpenAI's advanced GPT-5.5 Thinking model, providing personalized analysis based on actual transaction data. While it aims to help manage spending and savings habits-like advising you to cut back on takeout-it’s important to note that ChatGPT isn’t a licensed financial advisor. This development could make financial planning more accessible for everyday users by offering quick, AI-driven insights. However, OpenAI emphasizes that the tool is still in its early stages and should not replace professional financial advice. The feature is currently available only to Pro users but will expand to all users in the future. As AI continues to integrate with financial services, expect more tools like this to emerge. Users should remain cautious about sharing sensitive financial data and ensure they understand the limitations of these technologies.
AI Startup Runway Expands Beyond Video Generation
Runway launched its first world model in December. This company makes video generation tools for filmmakers and ad agencies. Runway's technology is used by major media players like Lionsgate and AMC Networks. The company is valued at $5.3 billion and added $40 million in annual recurring revenue in the second quarter of 2026. Runway's plan to build world models could change how we make films and discover new drugs.
Google Gemini Leads in AI Model Usage
Google's Gemini model is in high demand amongst AI customers. It led in the number of tokens handled by Vercel's AI Gateway in April. Google's Gemini model is popular due to its speed and affordability. In April, Google led with 21% of dollars spent on AI models, while Anthropic led with 61%. OpenAI's spend share tripled from March to April. The AI market is constantly changing, with new models and features being released. Google will likely unveil new AI models and tools at its upcoming conference.