Nvidia Just Bet $5 Billion on an AI Startup Most People Have Never Heard Of. Here’s Why

Nvidia, the world’s leading AI chipmaker, just announced it’s investing $5 billion dollars—a giant chunk of money by any reasonable measure—into a smallish AI startup that most people probably haven’t heard of. It’s an outfit called Safe Superintelligence, SSI for short, based in Palo Alto, California and Tel Aviv, Israel. The investment could nevertheless be hugely important for a number of reasons relating to the future of AI tech. Here’s what’s going on. What is Safe Superintelligence? SSI was founded only in mid-2024, and its co-founder and CEO is Ilya Sutskever, formerly OpenAI’s chief technology officer. Sutskever created a fuss when he left OpenAI because he’d been intimately involved in bringing ChatGPT to the world, and was leading the AI’s alignment team. This part of OpenAI was critical, especially at the time, because it had one main job: making sure that as OpenAI’s products got more sophisticated they aligned with humankind’s needs, rather than evolving into a threat. Sutskever’s departure led to some serious questions about OpenAI’s good intentions when developing future AI models—a debate that’s still bubbling away today, as OpenAI’s Sam Altman stirs the pot by announcing he feels we’re already in a powerful singularity moment. Safe Superintelligence is actually about more than just building AI chatbots like Siri or ChatGPT. The clue’s in the name: it’s not just about intelligence—it’s trying to make “superintelligence.” This, experts have long suggested, is an AI capable of out-thinking even the most agile human mind, in pretty much any area you can imagine. The company’s own website says this may be “the most important technical problem of our time” and it “is our mission, our name, and our entire product roadmap, because it is our sole focus.” Rather than developing ever-smarter AIs and then bolting on safety and security, SSI is developing AI and safety “in tandem, as technical problems to be solved through revolutionary engineering and scientific breakthroughs.” News like the recent international ban of Anthropic’s Mythos 5 model by the U.S. government for national security reasons, and the fact that an OpenAI model went rogue, escaped and then attacked a rival’s AI systems with a hack, show how important safe AI development is becoming. For superintelligent AIs, safety is clearly going to be even more important. SSI has been in and out of the headlines since its founding because although it’s raised sizable investments (topping $3 billion in total by late 2025,) it has yet to release any kind of meaningful product. The new $5 billion from Nvidia, its biggest financial injection yet, has resurfaced many questions about the company. Why does the investment matter? The mammoth investment is a sign that Nvidia, which by many accounts makes the world’s leading AI chips, is interested in a safe future for AI, and that it’s thinking beyond the current AI chatbot paradigm and has an eye on the next generation of superintelligent systems. As part of the deal, SSI will also gain access to Nvidia’s Vera ​Rubin chips. These are next-generation chips that are even more sophisticated than the company’s Blackwell chips, which stirred plenty of media attention when they were revealed in 2024. For certain key processes needed to make AI calculations, the new chips may be between two and five times faster than the Blackwell chips. This kind of raw power should speed the development of SSI’s AI systems significantly. In a press release, Nvidia also revealed the investment and technology deal goes both ways. It noted that the “two companies will also collaborate on the technical advancement of NVIDIA’s current and future compute platforms, leveraging SSI’s unique insights into the future of AI.” This implies that SSI has made powerful and useful progress that Nvidia executives have seen, and they want to make the most of SSI’s developments to keep their own company at the cutting edge of AI. The release quotes Nvidia CEO Jensen Huang, who praised Sutsever’s expertise and “fundamental breakthroughs at the foundation of modern AI.” He went on to say he was “excited to see what new breakthroughs SSI will discover powered by our Vera Rubin platform.” Why should you care about this? The news won’t directly impact the way you use AI in your own company. But it may remind you that current-generation AI tech isn’t inherently safe or trustworthy. Use these tools in the wrong way, and you could make a disastrous financial decision because of false information an AI told you, for example; or, a third party AI company may leak sensitive data your own staff have shared with an AI tool (as just happened with Anthropic’s Claude model). But the idea of superintelligent AI tools should be a part of your long term business plans. If you’re finding today’s AI tools helpful in building your business, you shouldn’t sleep on the innovative AI models that the future promises. Constantly checking to see if AI matches your needs could become a new business norm. BY KIT EATON @KITEATON