Twenty-Eight AI Startups and One Central Question: Where Is the Real Business?
Where among twenty-eight AI startups is the real business, and where is there only an expensive demonstration?
Separate the investment signal and the impressive demonstration from the real business in the case of NVIDIA; the assessment must check who pays, which indispensable part of the chain the product controls, and whether the economics survive scale.
What to watch for
Key takeaways
In the context of “The most cozy sub-categories on modern technology, investment, start-ups and AI,” this criterion applies: the relevant signal is not one number or one round: runway, access to the next round, and the ability to retain a customer reveal whether the business is durable.
In the context of “28 AI-strates with investments over $100 million: how so much?,” this criterion applies: the relevant signal is not one number or one round: runway, access to the next round, and the ability to retain a customer reveal whether the business is durable.
For the “The most promising start-ups are dark horses?” scene, the decisive point is this: the important signal is not one funding number: the next round, available runway, and closure rate show whether a company can survive the new cost of capital.
The working conclusion from “Hebbia's company started before heip?” is that this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.
The “Skild AI - will build robots Detroit Become Human?” scene leads to a working conclusion: the important signal is not one funding number: the next round, available runway, and closure rate show whether a company can survive the new cost of capital.
The “Full automated smartphone factory from XIAOMI” topic becomes clearer once this point is included: this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.
The boundary of the “Etched.ai - creates chips for AI, which is more than GPU” case is defined by this point: the important signal is not one funding number: the next round, available runway, and closure rate show whether a company can survive the new cost of capital.
For the “AlphaSense - AI, predicting business trends” scene, the decisive point is this: the important signal is not one funding number: the next round, available runway, and closure rate show whether a company can survive the new cost of capital.
The working conclusion from “Cyera - Cybersecurity stuns” is that this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.
The “10 other AI companies” topic becomes clearer once this point is included: this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.
What this episode is about
Hundreds of millions of dollars are going into much more than chatbots. Investors are betting on robots, chips, drug discovery, medical billing, cybersecurity, data, and developer tools. The most interesting companies are often not in AI's shop window, but in infrastructure and difficult industry workflows.
A list of twenty-eight startups that have each raised more than one hundred million dollars shows that the AI market is far broader than OpenAI and NVIDIA. Large sums are flowing into companies that the mass audience has barely heard of.
That makes sense: the most durable business often appears not in a universal chat interface, but where there is an expensive, narrow problem and a customer already accustomed to paying for it.
Hebbia works with financial documents—an industry with strong purchasing power and outdated tools. AKASA automates medical billing, where errors and manual labor are expensive. AlphaSense gathers business intelligence. In products like these, the model is not sold separately. It is embedded in a process that existed before AI.
Another layer is the physical world. Skild AI is building a foundation for robots, Bright Machines is automating manufacturing, and Xiaomi is demonstrating factories with minimal human involvement. If robots begin collecting data and learning from real actions at scale, that will create a new compute market and expand demand for infrastructure even further.
Etched.ai and Celestial AI are trying to change the computing foundation itself. Scale AI, data companies, and cybersecurity providers sell picks and shovels to those building models. This is often the more defensible position: the supplier is needed by many players and does not have to predict a single winner.
Suno and Udio show the other side of rapid growth: user attention arrives together with lawsuits from rights holders. Xaira Therapeutics and EvolutionaryScale promise to use models for medicine and biology, but results there take years to verify. The central lesson from all twenty-eight companies is that an AI startup should be evaluated not by the spectacle of its demo, but by its data, industry expertise, implementation cost, and the identity of the customer actually signing the check.
An AI startup should be judged by its data, industry expertise, implementation cost, and the customer actually signing the check—not by the spectacle of its demonstration.
Episode transcript
The episode is in Russian; below is an English reading guide to the transcript (the full EN transcript is a machine translation). Voice matching applied to 74 segments: 48 identified, 10 mixed, 12 marked with ✓, and 4 unresolved.
Loading…