Twenty-Eight AI Startups and One Central Question: Where Is the Real Business?

In the context of “The coziest podcast about modern technology, investment, startups, and AI,” this criterion applies: hundreds of millions go far beyond chatbots — investors bet 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 industry workflows.

In the context of “28 AI startups with over $100 million raised: where is it all coming from?,” this criterion applies: the list of twenty-eight startups with rounds above one hundred million shows the AI market is far broader than OpenAI and NVIDIA: big money flows into companies the mass audience has barely heard of.

For the “The most promising start-ups are dark horses?” scene, the decisive point is this: the most durable business often appears not in a universal chat but where there is an expensive, narrow problem and a customer already used to paying for it — which is why the dark horses on the list look sturdier than the loud names.

The working conclusion from “Hebbia — started before the hype?” is that Hebbia works with financial documents — an industry with strong purchasing power and outdated tools: the model here is not sold separately but embedded in a process that existed long before AI.

The “Skild AI — will it build robots like in Detroit: Become Human?” scene leads to a working conclusion: Skild AI is building a general foundation for robots, and if robots start collecting data and learning from real actions at scale, a new compute market emerges — demand for infrastructure widens even further.

The “A fully automated smartphone factory from XIAOMI” topic becomes clearer once this point is included: Xiaomi demonstrates factories with minimal human involvement while Bright Machines automates manufacturing — the physical world becomes AI's next layer, where value is proven by a working assembly line, not a demo.

The boundary of the “Etched.ai — building AI chips cooler than GPUs” case is defined by this point: Etched.ai and Celestial AI try to change the very foundation of computing: a specialized chip versus a universal GPU — a bet that model architectures will stabilize and purpose-built hardware will prove several times more efficient.

For the “AlphaSense — AI that predicts business trends” scene, the decisive point is this: AlphaSense gathers business intelligence while AKASA automates medical billing, where errors and manual work are costly: in both cases AI sells not as a model but as part of an industry process with an existing budget.

The working conclusion from “Cyera — audacious cybersecurity specialists” is that Scale AI, data companies, and cybersecurity players like Cyera sell shovels to those building models — often the more defensible position: the supplier is needed by many players and does not have to pick a single winner.

The “10 other AI companies” topic becomes clearer once this point is included: fast growth has a flip side: attention to Suno and Udio arrived together with rights-holder lawsuits, while Xaira Therapeutics and EvolutionaryScale face results that take years to verify — all twenty-eight companies must be judged by data, expertise, implementation cost, and who signs the check.