The Entire Tech Industry Talks About ChatGPT, Yet Outside the Bubble Surprisingly Few People Know What to Do With It
Why does the entire technology market discuss ChatGPT while people outside the bubble still do not understand why they need it?
Separate attention inside the technology bubble from mass adoption and identify the concrete everyday value that could move ChatGPT beyond early users.
What to watch for
Key takeaways
The boundary of the “The ToTheMoon channel is your guide to the world of modern technology” case is defined by this point: NVIDIA becomes the world's most valuable company, Ilya Sutskever leaves to build a new lab, hundreds of billions flow into AI — while the average user often still does not understand why they need ChatGPT or what to do with it every day.
The decision in “NVIDIA has become the world's most valuable company” depends on one criterion: NVIDIA rises on expectations of enormous demand for compute, and from inside the industry it looks like a race of historic scale — yet a chip supplier's market value measures expectations, not how many people already get daily value from the models.
The practical meaning of “Is Ilya Sutskever off to work at Sber? Or Tinkoff?” is that the joke "insider news" about Sutskever joining Sberbank highlights a real feature of the bubble: the technology world discusses people only the technology world knows — most viewers have no idea who he is.
The decision in “Who is Ilya Sutskever? Was he part of the plot against Sam Altman?” depends on one criterion: Sutskever was behind AlexNet — the breakthrough computer-vision model of the early 2010s — yet the wider public heard of him only after ChatGPT took off and the Altman firing saga: research reputation and public fame follow different laws.
The “Emoji in iPhones — does it matter?” topic becomes clearer once this point is included: generative emoji stand out in a presentation, but the value of the integration will be decided elsewhere: whether the system can help with messages, photos, search, calendars, and personal context without constant mistakes.
The discussion of “TikTok unveiled video generation. On the fakes that will flood the internet” yields a practical test: the more mass-market the video-generation tool, the cheaper convincing content becomes and the harder it is to tell a real event from a generated one — trust stops being a technical issue and becomes an everyday one.
The boundary of the “$600 billion of investment in AI” case is defined by this point: by the logic of Sequoia's article, with NVIDIA's revenue near $150B the software companies on top should be earning $600B — in fact OpenAI makes around three billion, Anthropic several times less, and not a single generative-AI startup clears $100M in revenue; meanwhile each new GPU generation sharply devalues the previous one.
The discussion of “NVIDIA — shovel sellers in a gold rush?” yields a practical test: the hosts find the parallel with Cisco — the market-cap leader of 2000 on the internet boom — incomplete: NVIDIA has the CUDA layer that ties neural-network workloads to it; still, the gap between the shovel seller's revenue and what the companies on top must earn is heading toward a trillion.
The decision in “Why isn't ChatGPT growing in value? Real-life ChatGPT use cases” depends on one criterion: many people have heard of ChatGPT but barely use it beyond one or two trial prompts: the product has not explained to a mass audience which recurring task it solves — so the model can be technologically strong while the user habit stays weak.
The decision in “SpaceX brought the rocket back” depends on one criterion: if infrastructure investment does not turn into durable products and revenue, expectations will prove too high. The true scale of AI will therefore be measured not by a chip supplier's market value, but by how many people begin using models every day and can name a specific benefit.
What this episode is about
NVIDIA became the world's most valuable company, Ilya Sutskever left to build a new lab, and hundreds of billions are flowing into AI. Yet the average user often still does not understand why they need ChatGPT or what they could use it for every day.
The technology market behaves as though AI has already become the basic infrastructure of the world. NVIDIA rises on expectations of enormous demand for compute, companies announce hundreds of billions in investment, and leading researchers such as Ilya Sutskever leave OpenAI to create new teams. From inside the industry, it looks like a race of historic scale.
Outside the industry, the picture is much calmer. A great many people have heard of ChatGPT but barely use it beyond one or two experimental prompts. The product has not explained to a mass audience which recurring task it solves. The model can therefore be technologically powerful while the user habit remains weak.
Apple can change that through distribution. When AI appears inside the phone, people no longer have to choose a separate service or learn a new interface. But even here, a presentation is not the same thing as a real product.
Generative emoji are easy to notice; the value will be determined by whether the system can help with messages, photos, search, calendars, and personal context without making constant mistakes.
At the same time, TikTok shows how quickly video generation can fill the internet with plausible fakes. The more widely available the tool becomes, the cheaper convincing content is to create and the harder it is to distinguish a real event from a generated one. Trust stops being a technical issue and becomes an everyday problem.
NVIDIA is currently selling the picks and shovels of the gold rush, but the market still remembers Cisco. If infrastructure investment does not turn into durable products and revenue, expectations will prove too high.
The true scale of AI will therefore be measured not by a chip supplier's market value, but by how many people begin using models every day and can name a specific benefit.
The real scale of AI will be determined not by supplier valuations, but by the number of people who use models every day and can name a concrete benefit.
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 52 segments: 34 identified, 0 mixed, 7 probable, and 11 unresolved.
Loading…