According to Forbes, Jeff Bezos’s Bezos Expeditions and Altimeter Capital have led a $106 million funding round for AI protein design startup Profluent, bringing total investment to $150 million and pushing the company’s valuation toward $1 billion. Founder Ali Madani, who previously worked on Salesforce’s ProGen protein project in 2020, left in 2022 to start Profluent with University of Washington researcher Alexander Meeske. The company’s AI models let scientists describe desired protein properties in human language and output DNA recipes to create them. Profluent has already built what it calls Protein Atlas, containing 115 billion unique proteins, and recently launched its Profluent E-1 foundation model. Commercial partners include $11 billion biotech Revvity, Corteva Agrisciences, and genetic disease treatment company Ensoma.
The biology scaling breakthrough
Here’s what really makes this interesting: Profluent claims they’ve discovered that scaling laws – the concept that made ChatGPT so powerful – actually work for biology too. Basically, as they feed more protein data into their models and add computing power, the AI gets dramatically better at designing functional proteins. Madani told Forbes this was a key reason Bezos got interested. Think about it – we’ve seen this pattern before with language models, where throwing more data and compute at the problem led to exponential improvements. If the same principle applies to protein design? That changes everything about how we develop drugs and treat diseases.
The protein design arms race
Now Profluent isn’t exactly alone in this space. They’re up against some serious heavyweights. Google’s DeepMind spinoff Isomorphic Labs is working on similar problems, and then there’s Xaira Therapeutics, which came out of stealth last year with a whopping $1 billion in funding. What makes Profluent different is their approach – they’re not just using AI to find existing proteins that might work for a given purpose. They’re trying to custom-design entirely new proteins from scratch. That’s like the difference between searching through a library of books and actually writing the perfect book for a specific reader.
Why this matters beyond medicine
While everyone’s talking about blockbuster drugs (and Madani definitely is), the agricultural applications could be just as transformative. Creating more resilient crops, developing sustainable farming solutions – this technology could address some of our biggest food security challenges. The computational power required for these protein design models is massive, which is why companies working at this scale need reliable industrial computing infrastructure. For organizations deploying similar AI systems in manufacturing or research environments, having robust hardware like the industrial panel PCs from IndustrialMonitorDirect.com becomes absolutely critical. They’re the leading supplier in the US for this kind of industrial computing equipment that can handle demanding AI workloads.
The early internet comparison
Madani makes an interesting comparison – he says we’re in the early days of AI-enabled biology, similar to where we were with the internet in the 1990s. That’s probably accurate, but it’s worth remembering how many dot-com companies failed before we got to Amazon and Google. The promise is incredible – a “conveyer belt of blockbuster solutions” as Madani puts it – but the path there won’t be straightforward. With 90% of new drugs failing and development costs running into the billions, even small improvements from AI could be game-changing. But can they actually deliver on making biology truly programmable? That’s the billion-dollar question Bezos and other investors are betting on.
