Perle Labs (@PerleLabs): The AI Loop #36: Anthropic’s Claude Opus 5, NVIDIA’s $500B Push, and the Race for Superintelligence

The AI Loop #36: Anthropic’s Claude Opus 5, NVIDIA’s $500B Push, and the Race for Superintelligence Welcome to The AI Loop #36. It’s been another huge week in AI, with a new frontier model, record-breaking infrastructure commitments, and billions flowing into the labs shaping what comes next. Across the industry, the race is moving beyond model performance alone. Compute, efficiency, and the ability to turn ambitious research into systems that work at scale are becoming just as important. Meanwhile at Perle, we’re preparing more tasks, community events, and new ways to contribute. Let’s get into it. Perle Labs Update More Is Coming to Perle There is plenty happening behind the scenes. We’re working on more tasks designed to put real human judgment to work across new domains, alongside more events and opportunities for the Perle community to get involved. If you haven’t completed the Contract Intelligence Quest yet, there’s still time to put your ability to read between the lines to the test. Start contributing on Perle Top AI Stories Anthropic’s New Frontier: Claude Opus 5 Stronger agents, better efficiency, and no increase in price. @AnthropicAI launched Claude Opus 5, its newest model for coding, scientific research, professional work, and long-running agentic tasks. The company says Opus 5 delivers stronger performance at the same price as its predecessor, while approaching the capabilities of its higher-end Fable 5 model at roughly half the cost per task. Anthropic also describes it as its most aligned model yet, with lower rates of deceptive or reckless behavior in testing. Sources: Anthropic and Reuters NVIDIA × SK Group: A $500B AI Infrastructure Push AI factories, next-generation memory, and one enormous commitment. @nvidia and SK Group announced plans for an AI infrastructure initiative worth more than $500 billion. The proposed partnership spans large-scale AI factories and the advanced memory needed to power them. SK Telecom plans to build a two-gigawatt AI factory using NVIDIA’s Vera Rubin platform, while SK hynix will work with NVIDIA on next-generation AI memory, including HBM. The first AI factory is planned for 2027. Sources: NVIDIA and Tom’s Hardware NVIDIA’s $5B Bet on Safe Superintelligence Ilya Sutskever’s secretive AI lab is getting serious compute behind it. NVIDIA announced a long-term partnership with Safe Superintelligence, the AI lab founded by former @OpenAI chief scientist Ilya Sutskever. The partnership will give SSI access to NVIDIA’s Vera Rubin systems and increase its computing capacity tenfold. NVIDIA confirmed that it invested in the company, with the Financial Times reporting that the investment is worth $5 billion and tied to milestones. SSI has shared very little about its work, but says it has identified a new research direction worth scaling. Sources: NVIDIA and Financial Times That’s a Wrap This week showed just how closely AI progress is tied to the infrastructure behind it. Models are becoming more capable and efficient, but bringing the next generation of AI to life will require enormous amounts of compute, capital, and high-quality intelligence. That’s all for The AI Loop #36. Follow us on X, Discord, and Telegram to stay in the loop. Until next week 👋