We\u2019re\u00a0in a unique moment for AI companies building their own foundation model. <\/p>\n
First, there is a whole generation of industry veterans who made their name at major tech companies and are now going solo. You also have legendary researchers with immense experience but ambiguous commercial aspirations.\u00a0There\u2019s\u00a0a clear chance that at least some of these new labs will become OpenAI-sized behemoths, but there\u2019s also room for them to putter around doing interesting research\u00a0without worrying too much about commercialization.<\/p>\n
The end result?\u00a0It\u2019s\u00a0getting hard to tell who is\u00a0actually trying\u00a0to make money.\u00a0<\/p>\n
To\u00a0make things simpler,\u00a0I\u2019m\u00a0proposing a kind of sliding scale for any company making a foundation model.\u00a0It\u2019s\u00a0a five-level scale where it\u00a0doesn\u2019t\u00a0matter if\u00a0you\u2019re\u00a0actually making\u00a0money \u2013 only if\u00a0you\u2019re\u00a0trying to.\u00a0The idea here is to measure ambition, not\u00a0success.\u00a0<\/p>\n
Think of\u00a0it\u00a0in these terms:\u00a0<\/p>\n
Level 5: We are already making millions of dollars every day, thank you very much.\u00a0<\/p>\n
Level 4:\u00a0We have a detailed multi-stage plan to become the richest human beings on Earth.\u00a0<\/p>\n
Level 3:\u00a0We have many promising product ideas, which will be revealed in\u00a0the fullness of time.\u00a0<\/p>\n
Level 2:\u00a0We have the outlines of a concept of a plan.\u00a0<\/p>\n
Level 1:\u00a0True wealth is when you love yourself.\u00a0<\/p>\n
The big names are all at Level 5: OpenAI, Anthropic,\u00a0Gemini, and so on. The scale gets more interesting with the new generation of labs launching now, with big dreams but ambitions that can be harder to read.\u00a0<\/p>\n
Crucially, the people involved in these labs can\u00a0generally choose\u00a0whatever level they want.\u00a0There\u2019s\u00a0so much money in AI right now that no one is going to interrogate them for a business plan. Even if the lab is just a research project, investors will count themselves happy to be involved. If you\u00a0aren\u2019t\u00a0particularly motivated to become a billionaire, you might well live a happier life at Level 2 than\u00a0at\u00a0Level 5.\u00a0<\/p>\n
Techcrunch event<\/p>\n
\n\t\t\t\t\t\t\t\t\tSan Francisco
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 13-15, 2026\n\t\t\t\t\t\t\t<\/p>\n
The problems arise because it\u00a0isn\u2019t\u00a0always clear where an AI lab lands on the scale \u2014 and a lot of the AI industry\u2019s current drama comes from that confusion.\u00a0Much\u00a0of the anxiety over OpenAI\u2019s conversion from a non-profit came because the lab spent years at Level 1, then jumped to Level 5\u00a0almost overnight. On the other\u00a0side, you might argue that Meta\u2019s early AI research was firmly at Level 2, when what the company really wanted was Level 4.\u00a0<\/p>\n
With that in mind,\u00a0here\u2019s\u00a0a quick rundown of four of the biggest contemporary AI labs, and how they\u00a0measure up\u00a0on the scale.\u00a0<\/p>\n
Humans&\u00a0<\/p>\n
Humans& was the big AI news this week<\/a>, and part of the inspiration for\u00a0coming up with\u00a0this whole scale. The founders have a compelling pitch for the next generation of AI models, with scaling laws giving way to an emphasis on communication and coordination tools.\u00a0\u00a0<\/p>\n But for all the glowing press, Humans& has been coy about how that would translate into actual monetizable products.\u00a0It seems\u00a0it\u00a0does\u00a0want to build products; the\u00a0team\u00a0just\u00a0won\u2019t\u00a0commit to anything specific.\u00a0The most\u00a0they\u2019ve\u00a0said is\u00a0that they will be building some kind of AI workplace tool,<\/a> replacing products like Slack, Jira and Google Docs but also redefining how these other tools\u00a0work at a fundamental level. Workplace software for a post-software workplace!\u00a0<\/p>\n It\u2019s\u00a0my job to know what this stuff\u00a0means,\u00a0and\u00a0I\u2019m\u00a0still\u00a0pretty confused\u00a0about that last part.\u00a0But it is just specific enough that I think we can put them at Level 3.\u00a0<\/p>\n Thinking Machines Lab\u00a0<\/p>\n This is a very hard one to rate!\u00a0Generally, if\u00a0you have a former CTO and project lead for ChatGPT raising a\u00a0$2 billion\u00a0seed round, you\u00a0have to\u00a0assume there is a pretty specific roadmap. Mira Murati does not strike me as someone who jumps in without a plan, so coming into 2026, I would have felt good putting TML at Level 4.\u00a0<\/p>\n But then\u00a0the last two weeks happened<\/a>. The departure of CTO and co-founder Barret\u00a0Zoph\u00a0has gotten most of the headlines, due in part to\u00a0the special circumstances<\/a>\u00a0involved. But at least five other employees left with\u00a0Zoph, many citing concerns about the direction of the company. Just one year in,\u00a0nearly half\u00a0the executives on TML\u2019s founding team are no longer working there. One way to read events is that they thought they had a solid plan to become a world-class AI lab, only to find the plan\u00a0wasn\u2019t\u00a0as solid as they thought. Or in terms of the scale, they wanted a Level 4 lab but realized they were at Level 2 or 3.\u00a0<\/p>\n There still\u00a0isn\u2019t\u00a0quite enough evidence to justify a downgrade, but\u00a0it\u2019s\u00a0getting close.\u00a0<\/p>\n World Labs\u00a0<\/p>\n Fei-Fei Li is one of the most respected names in AI research, best known for\u00a0establishing\u00a0the ImageNet challenge that kickstarted contemporary deep learning techniques. She currently holds a Sequoia-endowed chair at Stanford, where she co-directs two different AI labs.\u00a0I won\u2019t bore you by going through all the different honors and academy positions, but it\u2019s enough to say that if she wanted, she could spend the rest of her life just receiving awards and being told how great she is.\u00a0Her book<\/a>\u00a0is pretty good too!\u00a0<\/p>\n So\u00a0in 2024<\/a>, when Li announced she had raised $230 million for a spatial AI company called World Labs, you might think we were\u00a0operating\u00a0at Level 2 or lower.\u00a0<\/p>\n But that was over a year ago, which\u00a0is\u00a0a long time\u00a0in\u00a0the AI world. Since then, World Labs has shipped both\u00a0a full world-generating model<\/a>\u00a0and\u00a0a commercialized product<\/a>\u00a0built on top of it. Over the same period,\u00a0we\u2019ve\u00a0seen real signs of demand for world-modeling from both video game and special effects industries \u2014 and none of the major labs have built anything that can compete. The result looks an awful lot like a Level 4 company,\u00a0perhaps soon\u00a0to graduate to Level 5.<\/p>\n Safe Superintelligence (SSI)\u00a0<\/p>\n Founded by former OpenAI chief scientist Ilya\u00a0Sutskever, Safe Superintelligence (or SSI) seems like a classic example of a Level 1 startup.\u00a0Sutskever\u00a0has gone to great lengths to keep SSI insulated\u00a0from commercial pressures,\u00a0to the point of\u00a0turning down an attempted acquisition from Meta<\/a>. There are no product cycles and, aside from the still-baking superintelligent foundation model, there\u00a0doesn\u2019t\u00a0seem to be any product at all. With this pitch, he raised\u00a0$3 billion!\u00a0Sutskever\u00a0has always been more interested in the science of AI than the business, and every\u00a0indication\u00a0is that this is a genuinely scientific project at heart.\u00a0\u00a0<\/p>\n