Periodic Labs

Periodic Labs is an American artificial intelligence research company founded in 2024 by Liam Fedus, a former senior researcher at OpenAI, and Ekin Dogus Cubuk, a former research scientist at Google Brain. The company is headquartered in San Francisco. Periodic Labs's stated mission is to automate scientific discovery by building "AI scientists" that combine advanced AI models with autonomous physical laboratories, where robots conduct experiments, collect data, and iterate without human intervention. The company emerged from stealth in September 2025 with a $300 million seed round, and as of March 2026 is reported to be in deal talks at a $7 billion valuation.

At a glance

• Founded: 2024 in San Francisco. Public launch September 2025.
• Status: Private. Pre-product in the conventional consumer-AI sense; physical-lab and AI-scientist platforms in active development.
• Funding: $300 million seed at a $1.3 billion valuation, announced September 2025. March 2026 reports indicate deal talks at a $7 billion valuation, a five-fold step from the seed.
• CEO: Liam Fedus (co-founder; former senior researcher at OpenAI)
• Other notable leadership: Ekin Dogus Cubuk (co-founder; former research scientist at Google Brain)
• Open weights: None. The Periodic First Release model is closed; broader product strategy has not been disclosed.
• Flagship products: Periodic First Release (January 2025; closed-weights model). Autonomous-lab platforms in development with first results targeted at superconducting-materials discovery.

Origins

Periodic Labs was founded in 2024 by Liam Fedus and Ekin Dogus Cubuk. Fedus was a senior researcher at OpenAI from 2020 through 2024, contributing to several frontier-model research lines. Cubuk was a research scientist at Google Brain working on AI for materials science, including the GNoME (Graph Networks for Materials Exploration) line of work that produced computational materials predictions.

The founding thesis combines two research traditions. The frontier-AI tradition contributes the modeling capability necessary to reason about scientific hypotheses, experimental designs, and data interpretation at the scale of contemporary AI systems. The materials-science and physical-experiment tradition contributes the architecture of robotic laboratories where AI-driven hypotheses can be tested without human intervention. The combination is positioned as a path to closing the loop between hypothesis generation and physical validation, a step the prior generation of AI-for-science research did not take.

The company emerged from stealth in September 2025 with a $300 million seed round, described in industry coverage as one of the largest seed rounds in venture-capital history at the time. The round was backed by Andreessen Horowitz, DST Global, NVIDIA, Accel, Elad Gil, Jeff Dean (Google Senior Fellow), Eric Schmidt (former Google CEO), and Jeff Bezos (Amazon founder). The investor list reflects unusual concentration of senior AI-industry technical and strategic backing.

In October 2025, additional reporting characterized the seed round as the result of "VC frenzy" around the founders' research credentials. In March 2026, Bloomberg reported that Periodic Labs was in deal talks for a follow-on round at approximately a $7 billion valuation, a five-fold increase from the seed-round valuation.

Mission and strategy

Periodic Labs's stated mission is to "automate scientific discovery" by building autonomous AI scientists. Fedus and Cubuk have framed the strategy as a deliberate alternative to AI labs focused on language and consumer-product capabilities, with the underlying premise that the highest-value applications of frontier AI are in scientific and engineering domains where capability gains translate into materials, energy, and pharmaceutical breakthroughs.

The strategy combines three threads. First, advanced AI models trained on scientific literature, experimental data, and reasoning tasks specific to scientific method. Second, autonomous physical laboratories where robots execute experiments according to AI-generated hypotheses, collect results, and provide closed-loop feedback to the AI systems. Third, focused initial application areas including superconductors, semiconductors, energy storage, pharmaceuticals, nuclear fusion, and defense materials, where commercial value of AI-accelerated discovery is concentrated.

The competitive premise is that AI for scientific discovery requires both modeling capability (which frontier labs are building) and experimental infrastructure (which frontier labs are not building), and that a vertically integrated company combining both layers can produce capability advantages that pure AI labs and pure experimental labs cannot. The autonomous-lab architecture is the structural innovation; the AI-scientist framing is the commercial wrapper.

The company's first publicly stated target is the discovery of new superconducting materials, an area where AI-accelerated experimentation could produce commercially valuable results that conventional research cycles have not delivered. The framing of "AI scientist" extends to subsequent industries on the application list.

Models and products

• Periodic First Release. Initial closed-weights model released in January 2025. Public details on architecture, parameter count, and capabilities have been limited.
• Autonomous laboratory platforms. In active development. The company has not publicly demonstrated a full closed-loop AI-and-robot lab system at scale, but published statements indicate that physical experimental capability is being built alongside the model research.
• Application focus. Superconducting-materials discovery is the publicly stated first commercial target. Subsequent applications include semiconductors, energy storage, and pharmaceuticals.

The company's product roadmap beyond the autonomous-lab platforms has not been disclosed. Whether Periodic Labs licenses its AI scientist capability, develops materials internally for direct commercialization, or operates as a research-services partner for existing industrial customers is an open question.

Benchmarks and standing

There are no published benchmark evaluations of Periodic Labs models or research as of April 2026. Standardized benchmarks for AI-driven scientific discovery have not converged in the industry, and the company has not released enough public artifacts to enable comparative evaluation.

The company's standing rests on the founders' research credentials at OpenAI and Google Brain, the unusual depth of the seed-round investor list, and the strategic framing of autonomous scientific discovery as a high-value application of frontier AI. Industry coverage has placed Periodic Labs among the most-watched 2025 to 2026 AI startups despite limited public capability evidence.

Leadership

As of April 2026, Periodic Labs's senior leadership includes:

• Liam Fedus, Chief Executive Officer and co-founder. Former senior researcher at OpenAI (2020 through 2024). Public face for the company on the AI-scientist research thesis and commercial strategy.
• Ekin Dogus Cubuk, co-founder. Former research scientist at Google Brain. Worked on AI for materials science prior to founding Periodic Labs, including the GNoME materials-prediction research line at Google.

The company has hired aggressively from AI-research and adjacent scientific fields. Specific senior-leadership additions beyond the two co-founders have not been broadly profiled in industry coverage.

Funding and backers

Periodic Labs's funding history through April 2026 consists of the September 2025 $300 million seed round at a $1.3 billion valuation, with additional reports of deal talks at a $7 billion valuation as of March 2026 (terms and lead investor not yet disclosed publicly).

The seed round investor base is unusually concentrated in senior AI-industry figures. Andreessen Horowitz and DST Global led with venture-capital scale. Strategic investors included NVIDIA (which has a strategic position across many Insurgent labs through its hardware-and-investment program). Notable individual investors included Elad Gil (high-profile AI investor), Jeff Dean (Google Senior Fellow and former Chief Scientist of Google AI), Eric Schmidt (former Google CEO), and Jeff Bezos (Amazon founder).

The structure of investor participation is notable. The combination of venture-capital firms with multiple senior AI-industry insiders, plus strategic-corporate participation through NVIDIA, signals unusual technical and strategic conviction in the autonomous-lab thesis.

Industry position

Periodic Labs occupies a structurally distinctive position among Insurgent labs. The combination of the AI-for-scientific-discovery thesis, the autonomous-physical-laboratory architecture, the founders' OpenAI and Google Brain credentials, and the senior investor base produces a profile no other AI company has matched.

Strategic risks include the engineering complexity of building autonomous-laboratory infrastructure at scale, the long time-to-market of materials-science and pharmaceutical applications, the open question of whether AI-driven hypothesis generation produces materially better results than existing computational and experimental research methods, and the commercial-model uncertainty of how the company will monetize its capability.

Strategic strengths include the differentiated technical and commercial positioning, the depth of senior investor support, the founders' research credentials in the specific subfields the company addresses, and the strategic alignment with a class of applications (advanced materials, energy, pharmaceuticals) where conventional research cycles are slow and AI acceleration could produce meaningful economic and societal value.

The company's positioning combines elements of Google DeepMind's AlphaFold-style scientific-AI research with elements of an industrial research-and-development organization. The closest commercial analog is the partnership between Google DeepMind and Isomorphic Labs, which similarly applies AI to scientific discovery (drug design specifically); Periodic Labs is broader in scope and structurally independent.

Competitive landscape

Periodic Labs competes with several research and Insurgent labs:

• Google DeepMind. AlphaFold (protein structure), AlphaProteo (de novo protein design), and adjacent scientific-AI research lines overlap with Periodic Labs's mission. DeepMind's research output is in academic-publication form rather than as a commercial offering, which is a different commercial model.
• Isomorphic Labs. Alphabet's drug-discovery subsidiary. Closest commercial analog on the AI-for-science framing, but focused specifically on pharmaceuticals.
• Anthropic, OpenAI. Frontier labs that may eventually pursue scientific-discovery use cases, though their current product strategies emphasize general-purpose AI and consumer or enterprise products.
• National laboratories and academic research institutions. Indirect competition through traditional materials-science and physical-experiment research methodologies.
• Industrial research and development organizations. Materials, semiconductor, and pharmaceutical companies' internal research arms.
• Other Insurgent labs. Periodic Labs is structurally distinct from the LLM-focused 2024 to 2025 Insurgent cohort, but competes for senior AI-research talent and for compute-and-funding access.

Outlook

Several open questions affect Periodic Labs's trajectory in 2026 and 2027:

• The reported $7 billion valuation deal in negotiation. Closing terms, lead investor, and capital deployment plans will indicate the scale of operations the company plans to support.
• First publicly demonstrated results from the autonomous-laboratory platforms. A capability claim in superconductors or another application area would validate the architectural thesis.
• The commercial model. Whether Periodic Labs sells AI-scientist services, develops materials for direct commercialization, or operates as a research-services partner determines the revenue trajectory.
• Continued senior-talent recruitment from AI research, robotics, and the relevant scientific subfields.
• Strategic relationships with semiconductor, energy, pharmaceutical, or defense customers as the autonomous-lab platforms reach commercial readiness.
• Possible expansion of the application portfolio beyond superconductors and the publicly stated first targets.

Sources

• TechCrunch: Former OpenAI and DeepMind researchers raise whopping $300M seed to automate science. September 2025 seed-round announcement.
• TechCrunch: Top OpenAI, Google Brain researchers set off a $300M VC frenzy for their startup Periodic Labs. Strategic context.
• Bloomberg: AI Science Startup Periodic Labs Is in Deal Talks at About $7 Billion Valuation. March 2026 follow-on funding talks.
• Maginative: Periodic Labs launches with $300M to build an "AI scientist". Mission and strategy context.
• Crunchbase: Periodic Labs. Funding history reference.