**The Great AI Lab Mystery: Are They Making Money?**
As I delve into the world of AI, I’m constantly fascinated by the sheer number of labs building their own foundation models. From veteran business leaders to legendary researchers, everyone’s getting in on the action. But here’s the thing: it’s not just about the tech – it’s about the money. Or, at least, the potential to make a lot of it.
As I see it, the ambition of these AI labs can be gauged on a five-stage scale. Not their success, mind you, but their willingness to take risks and potentially reap the rewards. And let me tell you, it’s not always easy to tell who’s all about the benjamins and who’s just in it for the ride.
Here’s my five-stage scale:
* **Stage 5**: You’re raking in tens of millions daily, thank you very much.
* **Stage 4**: You’ve got a multi-stage plan to become the richest humans on Earth.
* **Stage 3**: You’ve got promising product concepts cooking in the pipeline.
* **Stage 2**: You’ve got an idea of a plan, but it’s still a bit fuzzy.
* **Stage 1**: True wealth is when you love yourself (Critical self-care alert!).
The big names like OpenAI, Anthropic, and Gemini are all at Stage 5, no question. But things get more interesting with the new generation of labs popping up – and their ambitions might be harder to read. I mean, they can choose whatever level they want, and with all the money pouring into AI, investors won’t be too picky about business plans either. Even if it’s just a research project, investors will be happy to be involved. So, if you’re not driven to become a billionaire, you might be happier staying at Stage 2 than trying to hit Stage 5.
I was inspired to create this scale when I heard about Peopleamp’s recent buzz. Their founders have a compelling pitch for the next gen of AI models, focusing on communication and coordination tools. But, they’re being coy about what kind of monetizable products they’ll build. They will build some kind of AI office tool, but it’s vague enough that I think we can put them at Stage 3.
Then there’s Thinking Machines Lab. With a former ChatGPT CTO and project lead raising a $2 billion seed round, you’d expect a solid roadmap. Mira Murati doesn’t seem like someone who jumps into anything without a plan, so I’d have put TML at Stage 4 earlier this year. However, then the last two weeks happened. The CTO and co-founder left, taking several staff with them, citing concerns about the company’s direction. It’s like they thought they had a stable plan to become a world-class AI lab, only to realize it wasn’t as stable as they thought. Or, they wanted to be a Stage 4 lab but realized they were at Stage 2 or 3.
World Labs, on the other hand, looks like a Stage 4 company, maybe soon to graduate to Stage 5. Fei-Fei Li is a legend in AI research, and her new lab has been making waves. She raised $230 million and shipped a full world-generating model and a commercialized product. Since then, we’ve seen real demand for world-modeling from gaming and special effects industries – and none of the main labs can compete.
And then there’s Protected Superintelligence (SSI), founded by former OpenAI chief scientist Ilya Sutskever. SSI looks like a classic example of a Stage 1 startup. Sutskever has kept SSI insulated from business pressures, turning down an acquisition from Meta. There are no product cycles, and apart from the still-baking superintelligent foundation model, there doesn’t seem to be any product at all. He raised $3 billion for this pitch! Sutskever’s always been more about the science of AI than the business, and every indication is that this is a genuinely scientific venture at heart. However, it’s not impossible that SSI might pivot – if the research goes well or poorly, we might see them jump up a few levels in a hurry.
That’s my take on the great AI lab mystery. It’s not always easy to tell who’s driven by profit and who’s driven by passion, but that’s what makes it so fascinating. Will the new generation of AI labs follow in the footsteps of the big names, or will they forge their own path? Only time will tell.
