The artificial intelligence landscape is rapidly evolving, with a surge of new companies building foundation models. However, determining the true ambitions of these labs – beyond simply securing funding – is becoming increasingly difficult. A new framework categorizes AI companies not by their current success, but by their stated intent to generate revenue, revealing a spectrum of goals from pure research to aggressive commercialization.
This ambiguity stems from a unique moment in the industry, where experienced professionals and researchers are launching independent ventures fueled by substantial investment. While some, like OpenAI and Anthropic, are clearly focused on monetization, others operate with less defined business plans, leading to questions about their long-term strategies and potential impact. The distinction between labs aiming for massive profits and those prioritizing research is crucial for understanding the future of AI development.
Understanding the Ambition Behind Foundation Models
To clarify this landscape, a five-level scale has been proposed to measure the commercial ambition of AI labs. Level 5 represents companies already generating significant revenue, while Level 1 prioritizes values beyond financial gain. The levels are as follows: Level 5 – making millions daily; Level 4 – a detailed plan for immense wealth; Level 3 – promising product ideas in development; Level 2 – outlining a concept of a plan; and Level 1 – prioritizing personal fulfillment over profit. This scale isn’t about judging success, but rather assessing the drive for commercial dominance.
The current influx of capital into AI allows many labs to operate comfortably even without immediate revenue streams. Investors are often content to be involved in cutting-edge research, even if it doesn’t translate into profits quickly. This creates a situation where a company’s stated level of ambition may not accurately reflect its potential or future trajectory. The shift from non-profit to for-profit, as seen with OpenAI, highlights the potential for rapid change in these ambitions.
A Look at Contemporary AI Labs
Here’s an assessment of four prominent AI labs based on this scale:
Humans&
Humans& recently garnered attention with its focus on the next generation of AI models, emphasizing communication and coordination over sheer scaling. The company’s founders propose a shift in AI development, but have remained relatively vague about specific, monetizable products. They envision an AI workplace tool that could potentially replace or redefine existing platforms like Slack and Google Docs.
While the vision is compelling, the lack of concrete details suggests Humans& is currently operating at Level 3. They clearly intend to build products, but their plans are still evolving and lack the specificity of more commercially-driven labs. The company’s approach prioritizes innovation and exploration over immediate market application.
Thinking Machines Lab
Founded by former OpenAI CTO Mira Murati, Thinking Machines Lab (TML) initially appeared poised for rapid growth, securing a substantial $2 billion seed round. Murati’s experience and leadership suggested a well-defined roadmap for building a world-class AI lab. This led to an initial assessment of Level 4, indicating a serious plan for significant financial success.
However, recent departures of key personnel, including co-founder Barret Zoph, have raised questions about the company’s direction and stability. These exits, coupled with concerns expressed by departing employees, suggest that the initial plan may have encountered unforeseen challenges. While not yet a downgrade, TML is approaching a point where a reassessment of its ambition level may be necessary. The situation highlights the difficulties in maintaining momentum and executing a complex AI strategy.
World Labs
Led by Fei-Fei Li, a highly respected figure in AI research, World Labs initially raised $230 million with a focus on spatial AI. Given Li’s academic background and commitment to research, an initial assessment of Level 2 or lower seemed reasonable.
However, World Labs has made significant progress in the past year, shipping both a full world-generating model and a commercialized product. This rapid development, combined with emerging demand for world-modeling in industries like gaming and special effects, suggests a more ambitious trajectory. Currently, World Labs appears to be operating at Level 4, and could potentially reach Level 5 as its technology gains wider adoption. This demonstrates the potential for research-focused labs to quickly transition to commercial success.
Safe Superintelligence (SSI)
Founded by Ilya Sutskever, formerly of OpenAI, Safe Superintelligence (SSI) is explicitly focused on the safe development of artificial general intelligence (AGI). Sutskever has actively shielded SSI from commercial pressures, even declining acquisition offers from major companies. This commitment to pure research places SSI firmly at Level 1.
Despite this focus, the possibility of a future pivot remains. Sutskever has acknowledged that SSI might pursue commercialization if research timelines extend or if the potential benefits of widespread AI access become compelling. While currently prioritizing safety and scientific advancement, SSI’s long-term trajectory remains uncertain. The company’s approach reflects a growing concern within the AI community about the responsible development of powerful technologies.
The AI landscape will continue to shift as these labs mature and refine their strategies. Monitoring the evolution of their stated ambitions, alongside their technological advancements and market activities, will be crucial for understanding the future of artificial intelligence and the development of generative AI. The next year will likely reveal clearer paths for these companies, particularly as they navigate the challenges of scaling AI models and securing long-term funding. The industry will be watching closely to see which labs prioritize profit and which remain committed to the pursuit of knowledge.
