OpenAI and Google researchers drive $300M funding for Periodic Labs

OpenAI researcher launches huge startup

Liam Fedus, a key figure behind ChatGPT’s post-training team at OpenAI, has teamed up with former Google Brain scientist Ekin Cubuk to start Periodic Labs.

The startup emerged from stealth with a $300 million seed round led by Andreessen Horowitz (a16z), with Felicis among the earliest backers.

The founders aim to combine AI, robotics, and simulations to automate materials science. Their vision is to have machines predict, mix, and test new compounds, creating discoveries faster than traditional lab methods allow.

From conversation to big funding

About seven months ago, Fedus and Cubuk brainstormed how AI could change scientific discovery. Their discussion highlighted recent advances in large language models, robotics, and simulations that made automated material research feasible.

This meeting led to an idea that quickly attracted attention. Once the news of Fedus leaving OpenAI spread, VCs began courting the founders, seeing enormous potential in a startup blending AI and experimental science.

The AI science breakthrough

Robotic arms capable of handling powder synthesis and LLMs that can reason about complex simulations set the stage for Periodic Labs. These tools can model physical systems and suggest experiments, essentially letting AI do lab work.

Cubuk had previously worked on a project where robots created 41 novel compounds suggested by language models. This proof-of-concept gave the founders confidence that AI-assisted science could yield real discoveries.

AI experiments generate valuable data

Fedus and Cubuk believe that even failed experiments provide crucial training data for AI. Unlike traditional science, where success is the main goal, their approach values exploration and learning from all results.

This mindset could reshape the scientific process, creating a continuous feedback loop. Data from every experiment, success or failure, will feed the AI, refining models and increasing the odds of real breakthroughs.

$300 million round led by Andreessen Horowitz

Although OpenAI was not an investor, the founders received massive attention from VCs. Andreessen Horowitz (a16z) led the $300 million round. Felicis, via partner Peter Deng, cut the first check and helped catalyze the fundraising.

This funding allowed Periodic Labs to hire top AI and science experts and set up a state-of-the-art lab. The capital ensures the team can test experiments and push boundaries without the typical early-stage constraints.

Silicon Valley love for the project

Investors were eager to get involved, some even sending letters describing their excitement for Periodic Labs. Fedus and Cubuk were quickly overwhelmed by the interest, illustrating the startup’s strong appeal in tech circles.

Such enthusiasm reflects confidence in both the founders’ track records and the concept of AI-powered scientific discovery. Investors recognized the opportunity to back a team capable of blending cutting-edge AI with experimental labs.

Building a diverse expert team

Periodic Labs recruited over two dozen experts in AI, physics, and material science. Weekly lectures help the team stay connected, ensuring everyone understands the complex tools and methods they are using.

This cross-disciplinary approach allows the lab to run experiments efficiently and integrate AI insights quickly, giving the startup a potential edge over traditional research teams.

Focus on superconductors discovery

The lab’s early mission is to identify new superconducting materials. These discoveries could revolutionize energy systems by improving efficiency and reducing energy consumption in tech and infrastructure.

Finding high-performance superconductors would be a game-changer, potentially impacting industries from electronics to clean energy. AI simulations and robotics give Periodic Labs a chance to make breakthroughs faster than conventional methods.

Robots still in training

While the lab is running experiments and simulations, the robotic systems are not fully operational yet. Training them to handle delicate chemical processes will take time and careful calibration.

Even with robots in early stages, the startup can already analyze results and refine experiments. Once fully operational, the automation will greatly speed up testing new compounds.

High stakes, uncertain outcomes

Scientific discovery is rarely predictable. Even with AI, robotics, and top talent, there’s no guarantee that the lab will find the superconductors or other breakthrough materials it seeks.

However, every experiment contributes valuable data. Even failures provide insights that refine AI models, making the approach worthwhile regardless of immediate success.

OpenAI expands science initiatives

OpenAI is also moving into AI-driven science with a new unit aimed at creating AI tools to accelerate discovery. This shows the wider trend of combining AI with experimental research.

Periodic Labs operates independently but benefits from the same AI advances, highlighting how startups can innovate quickly outside large corporate settings while leveraging research trends.

Star studded investors lineup

Besides Felicis, notable investors include Andreessen Horowitz, Accel, NVIDIA’s NVentures, and angel backers like Jeff Bezos, Eric Schmidt, and Jeff Dean. This diverse group provides both capital and industry credibility.

Having high-profile backers also gives the startup visibility and networking opportunities. These investors bring connections that can accelerate research partnerships and potential commercial applications.

Funding fuels fast scaling

With $300 million in seed capital, Periodic Labs quickly set up a fully equipped lab and hired top-tier talent. This allows the startup to move faster than typical early-stage science ventures.

Access to resources and expertise means experiments can be run in parallel, AI models refined quickly, and discoveries pursued aggressively, making speed a key advantage for the team.

Periodic Labs mission and vision

Periodic Labs aims to push the limits of AI in experimental science. Their goal is not just to discover superconductors but to create a scalable, data-driven model for scientific exploration and discovery.

The vision combines AI, top-tier researchers, and automated labs to accelerate science and make it more efficient, demonstrating the potential of technology to transform entire research disciplines.

Even the smartest AI has its blind spots. Discover what an insider has just exposed: A former OpenAI researcher reveals ChatGPT’s errors.

Could AI change how we discover?

Will AI-driven labs like Periodic change the future of science? By merging simulations, robotics, and human expertise, they might make discoveries far faster than traditional methods allow.

These experiments could unlock new materials, insights, and innovations. Every result, success or failure, pushes AI and scientific knowledge forward, offering a glimpse into a very different research landscape.

OpenAI’s hardware vision is already taking shape. See who’s leading the charge as OpenAI hires Apple experts for its debut hardware product.

Do you think this partnership will lead to the next major AI breakthrough? Drop a like if you’re excited for what’s coming, and share your thoughts below.

Read More From This Brand:

• OpenAI reorganizes group responsible for ChatGPT’s persona design
• OpenAI announces new AI hiring platform to compete with LinkedIn
• OpenAI says no to rumors of California exit amid pressure

Don’t forget to follow us for more exclusive content on MSN.

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one.

This slideshow was made with AI assistance and human editing.