Scientists are marveling at a new AI from DeepMind's drug discovery company that rivals AlphaFold's breakthrough protein folding capabilities. But unlike AlphaFold, which was open to researchers worldwide and transformed biology overnight, this one's staying private.The era of open AI research in biology may be over. And we should all be paying attention.AlphaFold changed biology by being open. When DeepMind released it, they didn't just publish a paper—they released the code, the models, and a database of predicted protein structures that any researcher could access. It was a genuine gift to science, one that accelerated drug discovery and structural biology in ways we're still measuring.Now the same team has built what scientists are calling "an AlphaFold 4"—a model with comparable or even superior capabilities for understanding molecular structures. But this time, it's coming from Isomorphic Labs, DeepMind's for-profit drug discovery spin-off. And it's staying behind corporate walls.I get it. Drug discovery is expensive. Building these models costs millions. Companies need competitive advantages. Isomorphic Labs wasn't created as a charity—it was created to make money from AI-driven drug development.But here's what we're losing. When AlphaFold was open, thousands of researchers could build on it. Academic labs could use it without licensing fees. Scientists in countries without massive research budgets could access the same tools as well-funded Western institutions. The pace of discovery accelerated because knowledge was shared.Now we're back to the old model. Proprietary tools. Competitive advantages. Knowledge as intellectual property rather than public good. The very model that AlphaFold disrupted.This is the inflection point where AI research shifts from public good to competitive advantage. DeepMind proved that open research could coexist with corporate success. Isomorphic Labs is proving that the temptation to keep things private is too strong to resist.The pharmaceutical industry will argue this is necessary. That drug development requires protected IP. That open-sourcing everything would eliminate the business model. And they're not entirely wrong—if you accept that drug discovery should be a for-profit enterprise.But we just saw what happens when you don't accept that premise. was free, and it accelerated science faster than any proprietary tool could have. Because when knowledge compounds, everyone benefits.The question now is whether this becomes the new normal. Will every breakthrough AI in biology go straight to a spin-off company? Will the next -level advance be locked behind enterprise licenses and NDAs?Scientists are already mourning what could have been. Another open tool. Another leap forward for biology. Another moment where AI served science instead of shareholders.The technology is impressive. The business model is profitable. But we're trading the pace of open science for the promise of proprietary returns. showed us a different path. just closed it.