Project Prometheus

Project Prometheus is an artificial intelligence research company founded in November 2025 by Jeff Bezos, the founder and former chief executive of Amazon, and Vik Bajaj, a Google X veteran and biotechnology entrepreneur. The company is headquartered in San Francisco and builds artificial intelligence systems specifically targeted at understanding and acting on the physical world, with stated application areas spanning engineering, manufacturing, aerospace, robotics, and drug discovery. As of April 2026, Project Prometheus has raised approximately $16.2 billion in cumulative funding across two rounds and is valued at roughly $38 billion, placing it among the highest-capitalized recently founded AI laboratories.

At a glance

• Founded: November 2025 in San Francisco by Jeff Bezos and Vik Bajaj.
• Status: Private. Series A closed April 2026.
• Funding: Approximately $16.2 billion cumulative funding. The November 2025 founding round raised $6.2 billion. The April 2026 round added $10 billion at a $38 billion post-money valuation. JPMorgan Chase and BlackRock are among the disclosed Series A participants. No single lead investor was disclosed for the Series A.
• Co-CEOs: Jeff Bezos (co-founder; founder of Amazon and Blue Origin) and Vik Bajaj (co-founder; co-founder of Verily and Xaira Therapeutics, formerly of Google X).
• Open weights: Not declared. The company has not stated a posture on the open-versus-closed distribution question.
• Flagship outputs: None publicly disclosed as of April 2026. Stated research direction is artificial intelligence systems that understand and reason about the physical world.

Origins

Project Prometheus was founded in November 2025 with $6.2 billion in initial capitalization. The founding occurred in a period when several senior figures from Big Tech and earlier-generation Silicon Valley research organizations had been moving into independent AI ventures, and when the broader "physical-AI" or "world-model" research direction had emerged as a recognized fourth track within the AI laboratory landscape alongside large-language-model frontier development, sovereign-AI national programs, and specialized-modality labs.

Jeff Bezos's involvement in artificial intelligence pre-dates Project Prometheus. Bezos was an early investor in Anthropic and in several other AI companies through personal and Amazon corporate vehicles, and Amazon's own corporate AI investments under his prior leadership had been substantial. The decision to found Project Prometheus represented a shift from investor or executive-sponsor roles toward direct co-founder and co-chief-executive engagement, with Bezos taking an operating role for the first time since stepping down as Amazon CEO in July 2021.

Vik Bajaj brings a research-organization-building track record to the founding. Bajaj was a senior leader at Google X, the moonshot research organization within Alphabet, where he helped lead the early development of Wing (the autonomous-delivery drone program) and contributed to Waymo (the autonomous-driving subsidiary). After Google X, Bajaj co-founded Verily, Alphabet's life-sciences research organization, and subsequently Xaira Therapeutics, an AI-driven drug-discovery startup. Bajaj holds a doctorate in physical chemistry from the Massachusetts Institute of Technology.

The company operated in stealth from November 2025 through early 2026, with first significant press coverage emerging in April 2026 around reports of the second financing round. Public disclosure of the company name, the founder pair, and the research direction occurred via the Series A fundraising announcements rather than through a coordinated launch event.

Mission and strategy

Project Prometheus has stated that its mission is to build artificial intelligence systems that understand the physical world. The company's research direction targets domains where artificial intelligence has historically been less mature than in language and image domains: engineering simulation, materials and manufacturing design, aerospace systems, robotics, and drug discovery. Each of these application areas requires AI systems that can reason about the dynamics, constraints, and causal structure of physical environments rather than only about patterns in human-generated data.

Strategically, this places Project Prometheus in a deliberately differentiated position relative to language-model-centric frontier labs. Where OpenAI, Anthropic, and Google DeepMind compete on general-purpose chat-and-reasoning capability with models trained on internet-scale text corpora, Project Prometheus is targeting application areas where the value-proposition is process automation, design optimization, and discovery within physical systems. The competitive question is whether physical-world AI systems are best built by purpose-built laboratories like Project Prometheus, by specialized arms of frontier labs (DeepMind's Isomorphic Labs in drug discovery, OpenAI's robotics research line), or by integration of general-purpose foundation models with domain-specific verifiers.

The application focus areas overlap directly with several of Bezos's existing investments and operational interests. Blue Origin (his aerospace company) has direct engineering and manufacturing applications. Amazon Web Services and Amazon's robotics fulfillment infrastructure represent both customer demand and deployment surface area. The drug-discovery focus aligns with Bajaj's prior experience at Verily and Xaira Therapeutics. Industry coverage has noted that this overlap creates plausible commercial pathways for early Project Prometheus output, though no commercial relationships with any of these organizations have been publicly disclosed.

Industry coverage has framed the company's strategic positioning as the most directly capitalized attempt by a credible founding team to build a "physical-world foundation model" laboratory. Whether the competitive position holds against Google DeepMind's long-running robotics and protein-folding research programs and against newer entrants in the space will depend on the company's near-term technical output.

Models and products

No public products as of April 2026. The company has not disclosed a model name, a target capability, or a public release date. Press coverage of the Series A indicates that the round will fund a multi-year capital-intensive research program focused on infrastructure, simulation environments, and recruitment, with first public artifacts not expected within the immediate post-funding window.

Distribution channels have not been disclosed. The closed-versus-open-weights posture has not been declared. Pricing structure, if applicable, has not been described. Whether eventual artifacts will be commercialized through licensing, through deployment within Bezos-related operating companies (Amazon, Blue Origin), or through standalone product offerings has not been addressed.

Benchmarks and standing

Project Prometheus has not yet shipped a public model and accordingly has no benchmark positions. The stated research direction (physical-world AI for engineering, manufacturing, aerospace, robotics, and drug discovery) suggests that early public artifacts, when they appear, are likely to be evaluated less on general-language-model benchmarks than on domain-specific physical-AI benchmarks. Conventional benchmarks for this category of system include simulation-environment policy benchmarks for robotics, structural-prediction accuracy benchmarks for drug discovery (analogous to the AlphaFold-era Critical Assessment of protein Structure Prediction evaluations), and manufacturing-process-optimization benchmarks. Standardization of physical-world AI benchmarks remains less mature than in the language-model space.

Leadership

• Jeff Bezos. Co-founder and co-chief executive officer. Founder of Amazon (1994) and chief executive officer through July 2021. Founder of Blue Origin (2000), the private aerospace company, where he serves as executive chair. Long-running personal investor in artificial intelligence companies including Anthropic, Perplexity, and Adept. Project Prometheus represents Bezos's first operating-executive role since stepping down from Amazon.
• Vik Bajaj. Co-founder and co-chief executive officer. Doctorate in physical chemistry from the Massachusetts Institute of Technology. Senior leader at Google X for several years, including early-stage work on Wing and contributions to Waymo. Co-founder of Verily (Alphabet's life-sciences subsidiary). Co-founder of Xaira Therapeutics, an AI-driven drug-discovery company. Brings research-organization-building and physical-systems-engineering experience to the partnership.

The senior research and engineering team has not been disclosed in detail. Press coverage of the Series A described an active recruitment effort across both AI research and applied-physical-systems engineering disciplines, with a stated emphasis on hiring researchers with backgrounds in robotics simulation, protein structure prediction, materials simulation, and large-scale reinforcement learning for physical environments. The dual co-CEO structure is unusual among AI laboratories at this scale and signals a deliberate division of executive responsibilities between Bezos's strategic and capital-allocation focus and Bajaj's research-organization-building focus.

Funding and backers

Project Prometheus has raised approximately $16.2 billion in cumulative funding across two rounds. The November 2025 founding round raised $6.2 billion in initial capitalization, with disclosure of the participating investors limited at the time of the round. The April 2026 Series A round added approximately $10 billion at a $38 billion post-money valuation.

JPMorgan Chase & Co. and BlackRock Inc. are among the disclosed participants in the Series A round. No single lead investor was identified. Press coverage has described the round as a syndicate of large traditional financial institutions and select strategic partners rather than a venture-capital-led financing in the conventional sense. The participation of major institutional asset managers (rather than venture funds) reflects both the round size and the pre-product nature of the company at the time of close.

The combined $16.2 billion capitalization places Project Prometheus among the most highly capitalized recently founded AI laboratories. Comparable scale exists at Safe Superintelligence ($32 billion valuation, $3 billion-plus cumulative funding as of late 2025), Thinking Machines Lab ($12 billion valuation), and at older established labs at higher valuations. The Project Prometheus valuation reflects investor confidence in the founding team's track records rather than any disclosed technical demonstration.

The capital intensity of physical-world AI research is generally higher than that of language-model research at equivalent scale, given the requirement for simulation infrastructure, robotics hardware, biological wet-lab partnerships, and engineering-design tooling. The funding scale is consistent with this capital-intensity expectation and is sufficient to fund several years of operation at frontier-research scale.

Industry position

Project Prometheus sits in a small but growing cluster of laboratories explicitly positioned around physical-world AI rather than around general-purpose language models. The cluster includes Physical Intelligence, the robotics-foundation-model startup founded in 2024; World Labs, Fei-Fei Li's spatial-AI laboratory; Skild AI and 1X on the humanoid robotics side; Isomorphic Labs (Alphabet's drug-discovery arm) on the biology side; and several smaller specialty labs targeting individual physical-AI verticals.

Within this cluster, Project Prometheus is differentiated by capital scale, by the breadth of stated application areas (most peers focus on a single physical domain), and by the operating involvement of Jeff Bezos as co-CEO. The breadth-of-application-areas positioning is a strategic bet on whether physical-world AI is best approached as a general-purpose foundation-model problem or as a set of vertical-specific problems. The dominant pattern in physical-AI research to date has been vertical specialization (Isomorphic Labs in drug discovery, Physical Intelligence in robotics manipulation, World Labs in 3D-scene generation); Project Prometheus's broader scope is a deliberate counter-bet on horizontal foundation-model applicability across physical domains.

The competitive relationship to Google DeepMind is particularly worth tracking. DeepMind has historically held the leading position in several of Project Prometheus's stated application areas: Isomorphic Labs in drug discovery, the robotics line including the RT-X and RoboCat work, and the long-running protein-folding research program around AlphaFold. Whether Project Prometheus competes head-to-head against DeepMind's research lines or carves out distinct application territory will be one of the principal competitive dynamics over the coming 24 months.

Competitive landscape

Direct competitors to Project Prometheus's stated research direction:

• Google DeepMind (and Isomorphic Labs). DeepMind's robotics research line and the Isomorphic Labs drug-discovery subsidiary represent the most directly competitive existing research programs. DeepMind has structural advantages in talent, in compute, and in research output to date, but Project Prometheus is positioned as a focused single-purpose laboratory rather than a research line within a larger company.
• Physical Intelligence. A robotics-focused foundation-model laboratory founded in 2024 with a $400-million-class Series A from Jeff Bezos's personal investment vehicle (among others). The competitive question is whether the Bezos-funded Physical Intelligence becomes complementary to, or directly competitive with, the Bezos-co-founded Project Prometheus.
• World Labs. Fei-Fei Li's spatial-AI laboratory, focused on AI systems that reason about three-dimensional space. The competitive overlap is partial: spatial reasoning is a component of Project Prometheus's stated direction but not the primary commercial framing.
• OpenAI and the broader robotics-research efforts at frontier labs. OpenAI has revived its robotics research line in 2025 and 2026 after a period of dormancy. Whether frontier general-purpose labs add physical-AI capability to their main flagship product lines, rather than carving out separate physical-AI organizations, is one of the structural strategic questions in the segment.
• Skild AI and 1X. Humanoid-robotics-focused laboratories. Direct competitors on the robotics application axis but with narrower scope than Project Prometheus's broader physical-AI framing.

Outlook

Open questions for Project Prometheus over the next 6 to 18 months:

• First public artifact. Whether the company ships any public-facing demonstration of capability in any of its stated application areas, and whether such an artifact takes the form of a foundation model, a domain-specific result (drug-discovery candidate, robotics policy, manufacturing-design output), or research publications.
• Application-area prioritization. Which of the five stated application areas (engineering, manufacturing, aerospace, robotics, drug discovery) the company prioritizes as its first public output. The choice will signal both research-team composition and strategic-commercial direction.
• Operating relationships with Bezos-controlled entities. Whether Project Prometheus develops formal commercial or research relationships with Blue Origin, Amazon, or other Bezos-affiliated organizations, and how such relationships are structured given the conflict-of-interest considerations involved.
• Talent acquisition trajectory. Which senior researchers Project Prometheus hires from DeepMind's robotics and Isomorphic Labs teams, from OpenAI's robotics line, and from academic robotics and protein-folding research groups. The hiring pattern will signal which application areas are receiving early research investment.
• The horizontal-versus-vertical strategic question. Whether the breadth of stated application areas resolves into a single horizontal physical-world foundation model or into a portfolio of vertical-specific systems. This is the principal research and product-strategy question the company will need to answer over the coming 24 months.
• Series B timing and structure. The timing and lead investor of the next financing event, particularly given the unusual Series A composition of large traditional financial institutions rather than venture funds.

Sources

• Bloomberg: Bezos's Project Prometheus AI Lab Hits $38 Billion Valuation After Funding Round. April 2026 Series A close coverage with valuation, lead investor structure, and stated application areas.
• Bloomberg: Jeff Bezos Nears $10 Billion Funding Round for Project Prometheus. Round structure and pre-close coverage.
• Tech Funding News: BlackRock and JPMorgan back Bezos' AI lab in a $10B raise at $38B valuation. Series A participating-investor disclosure.
• The Next Web: Jeff Bezos' physical AI lab is close to raising $10 billion. Stated application-area description and physical-AI strategic positioning.
• Wikipedia: Project Prometheus (company). Founding date, founder pair, and initial capitalization.
• Inc.: Jeff Bezos's Secretive AI Startup Just Hit a $38 Billion Valuation. Co-CEO structure and operational disclosure.
• Wikipedia: Jeff Bezos. Career history including Amazon founding, Blue Origin, and prior AI investments.
• Wikipedia: Verily. Bajaj's co-founder role and Alphabet life-sciences context.