Radical Ventures leads $100M Series B in Twelve Labs, bets on Anthropic becoming a pharma company by 2030

Twelve Labs' $100M Series B

Radical Ventures led a $100 million Series B in Twelve Labs, a company focused on video intelligence and understanding rather than video generation. Radical first backed Twelve Labs at seed in 2021, and Rob Toews frames the distinction as central to the investment thesis: where Sora and Veo generate video, Twelve Labs enables semantic search and comprehension of existing footage — identifying concepts, people, and dialogue within video at scale.

Anthropic's Life Sciences Pivot

Toews' sharpest prediction is that Anthropic will become one of the largest life sciences companies in the world by 2030 — not just a horizontal technology provider serving pharma as a customer, but an active drug developer in its own right.

The argument rests on pattern recognition from Anthropic's coding push. The company focused intensively on coding, built the best models in that domain, and grew quickly off the back of it. Toews believes life sciences is the next deliberate concentration of effort, driven partly by Dario Amodei's personal background as a trained neuroscientist.

Two recent data points sharpen the case. John Jumper, who led the AlphaFold project at DeepMind and shared the Nobel Prize with Demis Hassabis, left DeepMind to join Anthropic. Separately, Anthropic announced Claude for Life Sciences and a new internal drug discovery program targeting orphan and neglected diseases — framed publicly as a learning exercise, but Toews reads it as the first step toward Anthropic developing its own drugs.

“Yesterday, we announced Twelve Labs' $100M Series B. Twelve Labs is not focused on video generation the way Sora or Vio are — focused on video intelligence and video understanding. On Anthropic: Dario is personally very passionate about biology. John Jumper, who led the AlphaFold work at DeepMind and won the Nobel Prize, just left DeepMind to join Anthropic, which was an incredible get for them.”

The Other 2030 Predictions

Toews publishes annual AI predictions in Forbes, grading himself publicly each December. His five predictions for 2030 span several themes:

• TSMC and ASML's dominance fractures. Both companies hold near-monopoly positions in chip fabrication and lithography respectively. Toews sees the seeds of disruption already germinating, with Elon Musk's TeraFab as the single most significant challenger announced so far. Musk's reasoning, as Toews describes it, is simple first-principles math: his companies alone will need roughly 50 times more chips than current global production can supply.

• AI energy intensity collapses per task. Toews is not predicting lower total AI energy consumption — he explicitly acknowledges Jevons' paradox and argues demand for intelligence is effectively infinite. The prediction is that today's frontier models will look "comically energy inefficient" in hindsight, with per-task energy falling by potentially millions of times.

• Telepathy becomes an established communication method. The most speculative entry. Toews defines telepathy as translating thoughts directly into words others can read or hear — something he argues has already been demonstrated in small-scale research by UCSF neurosurgeon Eddie Chang, whose lab has implanted BCI chips and successfully decoded intended speech. Commercial availability by 2030 remains the stretch.

• AI legal rights enter mainstream political debate. Toews expects that as AI models grow more sophisticated, gain physical form through humanoid robots, and become embedded in people's daily lives as doctors, friends, and companions, the question of model sentience will move from fringe to mainstream. He notes that Anthropic and Google DeepMind are already hiring researchers to work on model consciousness and welfare.

The through-line across all five is that the most consequential AI developments by 2030 won't be incremental capability improvements — they'll be structural reconfigurations of industries, supply chains, and social norms that most people aren't yet pricing in.