Phyraxis AI lands early funding as $3B pours into physical AI for drug discovery
What's the deal? Phyraxis AIDealroom has a profile for this one. Try Dealroom →, a Shanghai-based startup building physical artificial intelligence for science, has raised a seed round, announced in September 2026. Founded by Wang Tong, the company — which goes by the English name Phyraxis AI — was barely a month old when it closed its first strategic investment, in the millions of yuan.
What is physical AI? The term describes AI that merges with the physical world, long associated with uses such as autonomous driving. In life sciences, it means letting AI perceive real lab environments and operate instruments or robots once controlled by humans.
Why now? As biological models churn out more molecular designs, validation and data feedback lag far behind, creating huge demand for wet-lab experiments. "This makes data from real experiments, especially failed ones, critically important," one industry figure told Arterial Network (translated from Chinese).
What's the endgame? Phyraxis AI pitches itself squarely at physical AI for science. Founder Wang Tong argues that embodied intelligence in life sciences "cannot stop at the demo level of a robot completing a single action."
The bigger picture: Phyraxis is one of a wave of startups attaching the physical AI label. New York-based Medra raised a $52 million Series A in late 2025, bringing its total to $63 million. Shenzhen's Jindu Biotech closed a near-100-million-yuan Series A led by GTJA InvestmentDealroom has a profile for this one. Try Dealroom →, and US peer Transfyr raised a $25 million seed in 2026.
Pharma giants are leaning in too. Nvidia and Eli Lilly said in January 2026 they would spend up to $1 billion over five years on a joint AI lab, while Simcere PharmaceuticalDealroom has a profile for this one. Try Dealroom → struck a global drug discovery deal with Schrödinger in August 2026.
The signal: Arterial Network estimates more than $3 billion has flowed directly or indirectly into physical AI for science. Two camps are emerging: companies treating the physical world as a venue — building robots to boost experiment throughput — and those treating it as knowledge, encoding molecular physics into models to cut experiment needs. Phyraxis, which logs commands, visual evidence, and outcomes in a platform ledger, sits in the lighter-touch group embedding into existing labs.
Read more: 36kr.com
Image credit: Generated with Gemini
