On Shark Tank if you walk in and say “I have no revenue, no viable business model, and an idea that’s easily copied in a space that’s rapidly marching towards commoditization. I’m asking for $5 billion for 25% of my company.” You’d be laughed off the set.
In the increasingly frothy world of GenAI that’s a Wednesday and you get the cheque.
If it looks like a bubble and smells like a bubble…
The tech is cool, but investors are about to lose their shirts in this space.
If it's actually that easy, why don't you do it?
Are you seriously just leaving $5 billion on the table?
I'll tell you why you won't. Because you wouldn't get it. Because it's not that easy.
Mira Murati is not some rando off the street. Ilya Sutskever is not some rando.
Zuckerberg is not paying Millions to randos of the street to come work for meta.
Nothing gets people to care so much about VC losses as does a 10 figure check!
Seriously, this stuff is great. A lot of people are getting rich, and I'm sure some value will be created out of it. LPs will be the ones holding the bag but thankfully we don't have a tradition of bailing them out like banks, so what do I care? In the meantime I benefit from VC subsidized coding assistants.
Tell me one unique skill, from any of those individuals you quoted, that is worth even 0,01% of $5 billion, for a non existent company with a non existent product.
LOL. So now it's not even their own skill but the "experience of having worked with certain people at well-funded and rapidly growing companies" that would value a nothingburger 12 Billion dollars.
I am sure there must be a better argument that just that. Because I have "worked with certain people at well-funded and rapidly growing companies" a lot and I'm worth nothing whatsoever.
LLMs are a commodity now. It is all about capital. DeepSeek and Grok proved that.
It’s not Klingon cloaking tech.
With minor variations, it is Transformers via autoregressive next-token prediction on text. Self-attention, residuals, layer norm, positional encodings (RoPE/ALiBi), optimized with AdamW or Lion. Training scales with data, model size, and batch size using LR schedules and distributed parallelism (FSDP, ZeRO, TP). For inference KV caching and probabilistic sampling (temperature, top-k/p).
Most differences are in scale, data quality, and marginal architectural tweaks.
Meta had the scale and whiffed so bad the entire Llama team disbanded and Zuck has been recruiting new people with 9 figure offers. Capital is essential but the people deciding when to launch the training run still do matter.
Looking at the team it's clear its a bunch of brilliant but indisciplined cats. They will be difficult to herd but for what the real plan is here, acquisition, it clearly does not matter.
They will do a couple of tools, then another panicked company like Apple, or Facebook or AWS, needing to justify the millions spent on AI and with nothing to show for, will acquire them for billions. When somebody will point out they don't make money or have no product, somebody here will point to the "team" as you just did :-)
But since we are talking about 12 billion, let's play the VC and do the due diligence first on the leader?
Reporting from The Optimist by Keach Hagey and other open sources, plus the famous thread here when Sam Altman got fired from OpenAI, reveals that Mira Murati had a central role in Sam Altman November 2023 firing.
For VCs, this is a masterclass in corporate survival and a massive red flag. Evidence shows Murati was the primary architect of the case against Altman:
- Wrote private memos questioning Altman's leadership
- Initiated board contact through Ilya Sutskever
- Sent "hefty PDF files of evidence" to board members via self-destructing emails
Then ...When employees revolted, fearing lost equity, immediately switched sides, signed the letter demanding Altman's return, then later claimed she "fought the actions aggressively."
She survived by reading the room perfectly:
- Became interim CEO during the crisis
- Positioned herself as essential to Altman's return
- Exited in September 2024 just before the for-profit restructuring
- Immediately raised $2B for her new startup showing it had it planned all along.
Investment Implications:
This reveals someone who will systematically undermine leadership while maintaining plausible deniability, then switch sides when politically convenient. Remember Paul Graham's famous quote about Sam Altman? That he "could be parachuted into an island full of cannibals and come back in 5 years and be king"? Murati might be even more astute.
She went from Intern → Tesla PM → Leap Motion → OpenAI VP → CTO → interim CEO → $2B startup founder, all with zero AI research background.
Her OpenAI trajectory shows world class political instincts. Join during post-Musk power vacuum (2018), build alliances with both sides, orchestrate a coup when convenient, flip sides when it fails, exit perfectly timed before restructuring.
For VCs:
Whether you see this as exceptional political skill or dangerous executive behavior depends on your risk tolerance.
Let's just hope the 10 millions dollars the government of Albania, the poorest nation in Europe, invested on this have been secured with proper share rights. That could make for some awkward reactions back home.
Their head of AI alignment clearly has no idea on how to go on alignment, as you can see here, during this 30 min rambling into nothing, on the subject.
Going just from published research, it seems Lilian Weng focus is LLM agents, safety, and alignment, focusing on how models are used, guided, and evaluated, not how they’re built.
It seems Horace He focuses on deep learning systems and compiler optimization, improving the performance of frameworks like PyTorch.
While both are clearly highly capable, and maybe capable of focusing on other areas, again just from published papers, neither seems to have published work on core LLM architectures or foundational model training, that could help bring about a scientific advance on the performance of current models.
Their contributions seem to be on enhancing usability and efficiency, not the underlying design or scaling of modern LLMs.
If that is the core team, I would be worried if they have the researchers capable of producing a breakthrough worth of the billions committed. But maybe that is why they are still hiring?
You mean, "I have no revenue and no business model but I was the CTO of one of the most important technical companies of this generation, we put out one of the fastest growing SaaS products of all time in a completely new technical category."- not sure how much more signal you would need than that. I don't really get why anyone would be confused about why they are throwing money at her. It's an extremely large amount of money, but it's not as if she has no track record.
It is a bubble, of course, but this person would get $100m round in a non-bubble.
Were so many people saying that when Meg Whitman left eBay? Are people saying that about Mira Murati? -I haven't heard anything to suggest she gets no credit for ChatGPT (the product/technical execution, not the model)
It's true that eBay did well during her tenure. And it's also true that she was a better grownup than the people running it before. But she also said that a monkey could drive the train. Would someone else have done better? Could a monkey have delivered the same results? We'll never know, but the experience at HP doesn't validate her as a business genius.
When proposing a model for evaluating startup investments -- or any decision-making framework -- it’s useful to run a kind of regression test.
Would Shark Tank have funded the most successful startups of the past decade? In many cases, no. They famously passed on companies like Airbnb, Uber, and Dropbox early on (sometimes indirectly via pitch decks or missed introductions), and even directly rejected DoorDash and Cameo, both of which went on to become unicorns. In other cases, they offered terms that, if accepted, would likely have made follow-on funding difficult.
That doesn't mean every GenAI deal today is wise. But pointing to Shark Tank as a benchmark for rational investing is a very selective filter. Historically, it’s missed more home runs than it’s hit.
Business plan, revenue, people: these are all valid signals to form an investment hypothesis. People often outweighs all others wrt early stage companies.
You would have been a fool to invest in apple in the late 90s on any other signal than the return of Steve Jobs. That era created many rich fools.
Will file your comment under: People who forget VCs regularly burn hundreds of millions of other people’s money, chasing hype. Also...90% of startups, and that is what Thinking Machines is, fail.
On the other hand, if for whatever reason some top tier soccer/football/basketball/etc. went to shark tank and said «I want to release my own sportswear brand», they’d probably get their funding. In fact, they’d probably have investors lined up.
Same thing here. Mira is a top tier, high-profile person in the AI space. So investors lined up.
Investors are investing just as much in the person/team as what they are building. What they build will change, but having a good team is all the difference.
Correct, but they're not profitable, rely on quickly depreciating capital assets and loads of tech employees to maintain their services, so it's hard for me to believe that there's a sustainable business here for them.
And like, if anyone's gonna win it should be Google because of their ability to push capital into this for a long time, their world class infrastructure and their cost advantage by having their chips in house. But I have faith in their ability to snatch defeat from the jaws of victory.
In the increasingly frothy world of GenAI that’s a Wednesday and you get the cheque.
If it looks like a bubble and smells like a bubble…
The tech is cool, but investors are about to lose their shirts in this space.