- Khosla Ventures is backing Discovery Loop, an AI startup aimed at accelerating scientific experimentation.
- Bloomberg reports the company was founded by former Google leaders, including Jeff Dean.
- Khosla Managing Director Samir Kaul says the opportunity could be “as profound as frontier AI models.”
- Kaul argues soaring valuations are forcing investors to redefine what “seed stage” means.
What Happened
Khosla Ventures is backing Discovery Loop, a new AI startup that Bloomberg reports was founded by former Google leaders, including Jeff Dean, with the goal of dramatically accelerating scientific experimentation. Speaking on Bloomberg Tech on August 21, 2026, Khosla Managing Director Samir Kaul explained why the firm moved quickly to invest and why he sees the opportunity as comparable in scale to building frontier models.
Why It Matters
“AI for science” has become one of the most-funded theses in the industry, following work such as DeepMind’s AlphaFold — which earned a share of the 2024 Nobel Prize in Chemistry — and a wave of startups applying models to biology, materials, and chemistry. A firm-level bet from Khosla, an early backer of OpenAI, signals that investors increasingly see automated experimentation, not just chatbots, as a route to durable value. Kaul’s comparison to frontier labs frames Discovery Loop as an attempt to industrialize the scientific method rather than to sell a single model.
Technical Details
Discovery Loop has not published a technical paper, and specific benchmarks or methods were not disclosed in the segment. The stated aim — closing the loop between hypothesis, experiment, and analysis — describes systems that use models to propose experiments, interpret results, and refine the next round, a pattern already visible in “self-driving lab” prototypes at academic and corporate research centers. The hard part is not generating hypotheses, which models do readily, but connecting them to real instruments and validating results, which is where most automated-discovery efforts have stalled. Kaul also used the interview to argue that current valuations are compressing the traditional financing ladder, blurring the line between seed and later stages.
Who’s Affected
Research-heavy sectors — pharmaceuticals, materials science, and chemistry — are the intended customers for automated-discovery platforms. Rival venture firms chasing the same thesis, and incumbent lab-automation vendors, are the most directly affected competitively. For founders, Kaul’s comments underscore how larger, earlier checks are reshaping expectations at formation.
What’s Next
Neither Khosla nor Discovery Loop disclosed the round size or valuation in the segment. Concrete signals to watch are a formal funding announcement, named research partners, and any published results demonstrating that the platform shortens real experimental cycles rather than only generating hypotheses.