Patronus AI Raises $50M to Stress-Test AI Agents in Simulated Environments
Patronus AI, a startup founded by former Meta AI researchers, has secured $50 million in funding to develop what the company describes as "digital worlds" designed to evaluate and stress-test AI agents, according to TechCrunch. The investment reflects what the company's backers characterize as nearly insatiable demand for robust agent-testing infrastructure. As AI agents are deployed across increasingly complex workflows, the need for controlled, simulated environments to assess their reliability and behavior before real-world deployment has become a growing focus for enterprise buyers.
Claude Gains Ground Among Paid AI Consumers
Anthropics's Claude is capturing a growing share of the paid consumer AI market, a segment that has historically been dominated by OpenAI's ChatGPT, TechCrunch reports. According to data cited in the report, consumers who pay for AI subscriptions have been increasingly opting for Claude over competing offerings. ChatGPT continues to hold a commanding overall market lead, but the trend among paying users marks a notable shift in competitive dynamics within the premium tier of the consumer AI space.
General Intuition Raises $320M to Train AI Agents on Gameplay Data
General Intuition has closed a $320 million funding round as part of what the company describes as a $2.3 billion bet on using video game data to train AI agents, TechCrunch reports. The company is scaling AI systems trained on millions of hours of gameplay, with the thesis that action-dense data from games can help AI develop capabilities more closely resembling human intuition. The approach represents a distinct data strategy compared to text- or image-centric training pipelines, positioning gameplay as a source of structured, goal-oriented behavioral data that could generalize to real-world agent tasks.
Former Databricks AI Chief Claims 1,000x Reduction in AI Power Consumption
A new venture led by Databricks' former AI chief is claiming it can reduce AI's energy consumption by a factor of 1,000, according to TechCrunch. The company has developed a system called Un-0, an image-generation tool that the company says demonstrates for the first time how its underlying technology can replicate the outputs of conventional AI systems at dramatically lower energy costs. Power consumption has become a significant operational and infrastructure concern for AI providers, and the claim, if validated at scale, would represent a meaningful development in the economics of AI deployment.
Netris Raises $15M Series A from a16z to Accelerate Neocloud Deployments
Netris has raised a $15 million Series A led by Andreessen Horowitz (a16z) to expand its network infrastructure software for AI neoclouds, TechCrunch reports. The company provides software that runs directly on network switches and offers a platform aimed at helping neocloud operators reduce the time required to bring new infrastructure online. As demand for AI compute capacity continues to drive the rapid build-out of specialized cloud providers, tooling that compresses deployment timelines has become an area of active investment.
These developments collectively reflect continued momentum across AI agent infrastructure, model competition, training data strategies, energy efficiency research, and cloud networking — spanning both the software and hardware layers of the broader AI stack.