Vivodyne opens the world's largest 'human data center' to fix AI drug discovery
AI drug discovery keeps promising cancer cures — but the models are training on the wrong data. Vivodyne, a University of Pennsylvania spinout, says the industry's fundamental problem isn't compute or algorithms; it's that most AI models learn from animal tests, single cells, or static snapshots of biology, not living human tissue responding to stimuli in real time. Last week, the company opened what it calls the world's largest "human data center" just outside San Francisco, and CEO Andrei Georgescu says his team is already achieving twice the throughput of all animal trials being held in the US.
Vivodyne's HIVE machines — modular robotic labs — can grow 20 kinds of human tissue, then autonomously dose and monitor them, generating the kind of causal biological data that today's AI models are missing. The company claims its liver cells have 94% predictive accuracy compared to human toxicity trials, its airway tissue matches real human tissue behavior 96% of the time, and its bone marrow has achieved 100% concordance across tests of 20 different chemotherapy drugs. "Absent human testing, what are these models going to do?" Georgescu told TechCrunch. "They're going to cure cancer in mice." The critique lands as even Dario Amodei acknowledged over the weekend that claims about AI curing cancer have become "more cliche than credible," and Isomorphic Labs — Google DeepMind's drug-discovery arm — is only now expecting its first trials by year's end, originally planned for 2025.
The broader thesis is that today's generative AI models for biology don't learn causation. A Nature Methods study published last month found no clear data scaling laws when training on existing cellular data. Vivodyne's pitch is that its hundreds of thousands of ongoing experiments — tracking diseased tissue exposed to various stimuli — could provide the reinforcement learning signal needed to build models that actually understand how cells get from state A to state B. The company has raised just under $80 million across two rounds led by Khosla Ventures, and is working with multiple major pharma companies, though it won't name them publicly.
What to watch: whether pharma partners actually publish results using Vivodyne's data — the "human data center" is an ambitious claim, and the proof will be in whether AI models trained on this causal data outperform existing approaches in clinical trials. The 90% clinical trial failure rate for animal-tested drugs is the number Vivodyne is trying to move.
Amazon's Prime Air drone delivery expands to nearly 500 US cities
Amazon says its Prime Air drone delivery service will reach nearly 500 US cities and towns by the end of 2026, a sixfold expansion from its current 11-site footprint. Deliveries are launching in Chicago, Syracuse, Cleveland, Atlanta, and Boise metro areas, with each Prime Air site covering roughly 175 square miles. Amazon says it has already delivered "hundreds of thousands of packages" this year, with most orders arriving around 60 minutes after checkout.
The expansion comes with heightened scrutiny — Amazon's drones have previously crashed into cranes, internet cables, gardens, and an apartment building. The company is emphasizing its "Detect-and-Avoid system" as it scales. For AI readers, the interesting angle is that autonomous drone delivery at this scale is a real-world stress test for computer vision and path-planning systems operating in uncontrolled urban environments with regulatory oversight.
Is the "human data center" concept the missing piece for AI in biology, or just anotherVC pitch? Tell us in the comments.
Sources: TechCrunch · The Verge · Vivodyne · Nature Methods · About Amazon
