In the rapidly evolving landscape of artificial intelligence, a unique dynamic is unfolding among companies building their own foundation models. The sector is witnessing an influx of seasoned industry veterans striking out on their own, alongside legendary researchers whose commercial aspirations remain somewhat opaque. While some of these nascent AI labs could undoubtedly grow into giants akin to OpenAI, others might simply pursue intriguing research without significant pressure for immediate commercialization.

This divergence makes it increasingly difficult to discern which AI labs are genuinely striving for commercial success. To bring clarity to this complex environment, a new five-level "ambition scale" is proposed, designed to measure a company's commercial intent rather than its current revenue or success.

Introducing the AI Lab Ambition Scale

This sliding scale categorizes foundation model developers based on their stated or implied commercial ambition:

  • Level 5: Actively generating millions in daily revenue.
  • Level 4: Possessing a detailed, multi-stage plan aimed at achieving significant market dominance and wealth.
  • Level 3: Having multiple promising product ideas, with specific details to be unveiled in due course.
  • Level 2: Operating with a conceptual outline of a business plan.
  • Level 1: Primarily focused on research or non-commercial pursuits, where "true wealth" is defined by self-fulfillment rather than financial gain.

Established industry leaders like OpenAI, Anthropic, and Gemini firmly occupy Level 5. However, the scale becomes particularly insightful when applied to the new generation of AI labs, many of which harbor grand visions but whose commercial strategies are less transparent.

Navigating Ambition in a Cash-Rich AI Landscape

A crucial factor enabling this ambiguity is the sheer volume of capital currently flowing into the AI sector. Founders and researchers often have the luxury of choosing their desired level of commercial engagement. Investors, eager to be part of the next big thing, are frequently content to back even purely research-focused projects without demanding rigorous business plans. This environment allows individuals to prioritize a happier, less commercially driven life at Level 2, for instance, over the intense pressures of a Level 5 enterprise.

However, this lack of clarity can also fuel significant industry drama. The controversy surrounding OpenAI's transition from a non-profit to a profit-driven entity, for example, stemmed from its rapid leap from Level 1 to Level 5. Similarly, one could argue that Meta's early AI research, despite the company's ultimate commercial goals, often operated at a more exploratory Level 2.

Assessing Key Contemporary AI Labs

With this framework in mind, let's examine four prominent contemporary AI labs and how they measure up on the ambition scale.

Humans&

Humans& recently made headlines with its substantial seed round and a compelling vision for next-generation AI models, shifting focus from scaling laws to communication and coordination tools. Despite the positive press, the startup has been notably reticent about how this vision will translate into concrete, monetizable products. While the team clearly intends to build products, they have yet to commit to specifics, vaguely mentioning an "AI workplace tool" that would replace and redefine existing solutions like Slack, Jira, and Google Docs.

The precise implications of a "workplace software for a post-software workplace" remain somewhat unclear, even to industry observers. Nevertheless, the stated intent to build transformative workplace tools is specific enough to place Humans& at Level 3 on the ambition scale.

Thinking Machines Lab (TML)

Rating Thinking Machines Lab, co-founded by former ChatGPT CTO Mira Murati, presents a challenge. Given Murati's background and the reported $2 billion seed round, one would initially assume a highly specific commercial roadmap, positioning TML firmly at Level 4 for 2026.

However, recent events have introduced uncertainty. The departure of CTO and co-founder Barret Zoph, along with at least five other employees citing concerns about the company's direction, suggests internal discord. Within just one year, nearly half of TML's founding executives have left. This could be interpreted as a realization that their initial Level 4 commercial plan was less robust than anticipated, potentially indicating a current standing closer to Level 2 or 3. While a definitive downgrade isn't yet warranted, the evidence is mounting.

World Labs

Fei-Fei Li, a highly respected figure in AI research known for establishing the ImageNet challenge, launched World Labs in 2024 with $230 million in funding for a spatial AI company. Given her academic background and numerous accolades, one might have initially placed World Labs at Level 2 or even lower, expecting a primary focus on research.

Yet, the AI world moves quickly. In just over a year, World Labs has not only shipped a full world-generating model but also a commercialized product built upon it. During this period, significant demand for world-modeling technology has emerged from the video game and special effects industries, with no major labs offering comparable solutions. This rapid progress and market traction strongly suggest World Labs is operating as a Level 4 company, with potential to soon reach Level 5.

Safe Superintelligence (SSI)

Founded by former OpenAI chief scientist Ilya Sutskever, Safe Superintelligence (SSI) appears to be a quintessential Level 1 startup. Sutskever has deliberately insulated SSI from commercial pressures, reportedly even rejecting an acquisition attempt from Meta. The company emphasizes a lack of product cycles and, beyond its foundational superintelligent model still in development, seems to have no immediate commercial products. Despite this research-first pitch, SSI successfully raised $3 billion, underscoring Sutskever's long-standing interest in the scientific aspects of AI over business.

However, the dynamic nature of AI means it would be premature to entirely discount SSI from the commercial realm. Sutskever himself has indicated two scenarios for a potential pivot on his recent Dwarkesh appearance: if development timelines prove unexpectedly long, or if the immense value of powerful AI necessitates its broader impact on the world. In essence, significant shifts in research outcomes, whether highly successful or challenging, could prompt SSI to rapidly ascend several levels on the ambition scale.