Nvidia Invests $5 Billion in Safe Superintelligence; AI Workers Petition for Development Pacing
A major investment signals continued acceleration in AI capabilities, even as over a thousand AI workers push for a deliberate slowdown and stricter governance.
The story
The AI software landscape saw significant movement this week, highlighting the ongoing tension between rapid advancement and calls for caution. Safe Superintelligence Inc., Ilya Sutskever's new research lab, secured a substantial $5 billion investment from Nvidia, signaling strong confidence in frontier AI development.
This influx of capital underscores the industry's drive to push the boundaries of AI capabilities. Simultaneously, a powerful counter-narrative emerged as over 1,100 employees from OpenAI, Anthropic, Google DeepMind, and Meta signed an open letter.
Titled "Pacing the Frontier," the petition urges the U.S. government to support an international effort to deliberately slow down and govern advanced AI development, emphasizing the need for society to adapt to emerging risks. This dual development illustrates the industry's complex trajectory, balancing the pursuit of next-generation AI with increasing concerns over its societal impact and control.
Who moved
Safe Superintelligence Inc.
What Changed: The new AI research lab, founded by Ilya Sutskever, secured a $5 billion investment from Nvidia.
Consequence: This substantial funding provides significant resources for the lab's secretive research efforts, indicating strong investor confidence in its potential contributions to frontier AI.
Moonshot AI
What Changed: The Chinese AI startup released the full open weights for its Kimi K3 model.
Consequence: This makes a 2.8 trillion-parameter, multimodal model freely available to developers, potentially impacting the open-source AI ecosystem.
OpenAI, Anthropic, Google DeepMind, Meta
What Changed: Over 1,100 employees from these companies signed a petition advocating for a deliberate pacing of frontier AI development.
Consequence: The collective action signals a growing internal push within major AI labs for stronger governance and international cooperation on AI safety.
Nvidia and Microsoft
What Changed: The companies launched the Open Secure AI Alliance.
Consequence: This new coalition aims to build open tools for AI cyber defense, bringing together various partners to address AI security challenges.
Anthropic
What Changed: The company clarified its stance on open-weights models, stating it has never advocated for a ban.
Consequence: This position differentiates Anthropic from some rivals and emphasizes its view that less dangerous open models can be a public good, while still advocating for other controls like chip export restrictions and mandatory safety testing.
New models
Kimi K3
Lab: Moonshot AI
What: Released as a 2.8 trillion-parameter open-weight model with native vision and a 1 million-token context window.
Use: Developers can now inspect, modify, and self-host the model, which has shown strong performance on coding leaderboards.
ChatGPT Work
Lab: OpenAI
What: Introduced as an agent in ChatGPT designed to handle ambitious, multi-step tasks.
Use: It can gather information across user apps and workflows to create finished materials and manage complex projects by breaking them into smaller steps.
Comet
Lab: Perplexity AI
What: Launched as a new AI-powered web browser that integrates search, browsing, and AI assistance.
Use: Perplexity Max subscribers can use it to answer questions about webpages, summarize content, compare information, and complete multi-step tasks without switching tabs.
Canva Code 2.0
Lab: Canva
What: Rolled out an updated AI-powered coding tool with expanded capabilities and visual editing controls.
Use: Users can create websites, apps, and interactive experiences from natural language prompts and then visually edit individual elements, with claims of 75% faster code generation.
Market signals
Safe Superintelligence Inc. received a $5 billion investment from Nvidia on July 27.
Implication: This massive capital injection into a new frontier AI lab highlights continued aggressive investment in advanced AI research and development.
Enigma secured a $71 million seed round led by Index Ventures and Ribbit Capital on July 27.
Implication: A significant seed round for an AI robotics company suggests strong early-stage investor appetite for AI applications in physical automation.
Genius AI closed a $44 million Series D round led by Lux Capital on July 27.
Implication: This later-stage funding indicates sustained momentum and investor confidence in AI companies demonstrating growth.
Major tech companies including Microsoft, Meta, Apple, and Amazon are scheduled to report Q2 earnings this week.
Implication: Investors will closely scrutinize these reports for insights into AI capital expenditures and their impact on revenue and profitability, setting a tone for broader market sentiment.
Builders saying
Dario Amodei, Anthropic CEO. Anthropic has not advocated for a blanket ban on open-weight AI models, believing less dangerous open models can serve as a genuine public good for businesses and researchers.
Jack Clark, Anthropic cofounder. He signed the 'Pacing the Frontier' petition, urging the U.S. government to support international efforts to deliberately pace the development of frontier AI.
What we'll be watching
- Microsoft's Q2 earnings report this Wednesday, July 30, will provide a key indicator of enterprise AI demand.
- Meta's Q2 earnings report, also this Wednesday, July 30, will offer insights into AI's impact on advertising and spending discipline.
- Apple's Q2 earnings report on Thursday, July 31, will be watched for signs of an AI upgrade cycle in consumer tech.
- Amazon's Q2 earnings report, also on Thursday, July 31, will reveal performance across cloud growth, retail, and AI spending.
- Federal agencies face an August 1 deadline to define 'covered frontier models' for a potential 30-day pre-release review process.
- OpenAI's annual developer conference, DevDay 2026, is scheduled for September 29 in San Francisco.
- The ITU's AI for Good Lab, launched July 28, will continue initiatives to help developing economies scale responsible AI solutions.
Reporting + analyst voices: grounded via Google Search at publish time.