OpenAI spends nearly $3 for every $1 subscribers pay, its ad business is 90% behind forecast, and Scott Galloway thinks the AI bubble is already starting to unravel — just like 1999.
The Prof G Pod with Scott Galloway
OpenAI spends nearly $3 for every $1 subscribers pay, its ad business is 90% behind forecast, and Scott Galloway thinks the AI bubble is already starting to unravel — just like 1999.
TL;DR
Scott Galloway's "No Mercy No Malice" newsletter, read by George Hahn, draws sharp parallels between the 1999 dot-com bubble and today's AI investment frenzy. OpenAI lost $21 billion in 2025, spends nearly $3 for every $1 of ChatGPT revenue, and its projected $100 billion ad business is tracking 90% below forecast [1] — George Hahn "OpenAI lost $21 billion in 2025: OpenAI's leaked financials reveal the company lost $21 billion in 2025, a scale of losses Galloway compare…" 09:20 . Uber burned its entire 2026 AI budget in four months, and enterprise sobriety is spreading [2] — George Hahn "Uber burned entire 2026 AI budget in 4 months: Uber blew through its entire AI budget for 2026 in just four months, prompting a broader ent…" 12:53 . The key takeaway: AI will create enormous value, but most of it will flow to users, not shareholders [3] — George Hahn "AI is likely to be like electricity, or jets, or the PC: transformative for society, ruinous for many early investors. The value won't conc…" 14:40 .
Scott Galloway's No Mercy No Malice newsletter essay, read by George Hahn, draws detailed parallels between the 1999 dot-com bubble and today's AI investment landscape, using OpenAI's financials and enterprise spending pullback as evidence of an unraveling bubble.
The episode opens with a sponsored segment for Thumbtack, a home services marketplace. The ad makes the case that personal recommendations — 'I know a guy' — are insufficient when the stakes are a flooded kitchen rather than a bad TV show. Thumbtack positions itself as the data-driven alternative, letting users browse verified ratings, photos of past work, and professional credentials before committing.
The second sponsor segment is a German-language advertisement for Flaschenpost, a grocery and drinks delivery platform. The ad promotes a €30 saving spread across three orders using the promo code HALLO30, emphasising convenience and reduced household effort. The segment is entirely in German and unrelated to the episode's editorial content.
Soccer legend Megan Rapinoe delivers a cross-promotional segment for her podcast 'A Touch More: The Beautiful Game.' The episode being teased features an interview with US Women's National Team and Denver Summit captain Lindsey Heaps, discussing her journey from Denver to Lyon and back, and her hopes for 2027 as national team captain. Rapinoe also promises her own take on moments and controversies from the World Cup quarterfinals.
Scott Galloway briefly frames the essay before handing off to narrator George Hahn. The thesis is stated plainly from the opening: the contagion pattern of the dot-com crash — B2C first, then B2B, then infrastructure — is beginning to form again in AI, with cracks appearing at OpenAI. Hahn then sets the philosophical scene, invoking Jamie Dimon's observation that financial crises come every 5 to 7 years, noting it's been 18 years since the last one. With age comes pattern recognition, and the pattern Galloway sees — slowly, then suddenly — is one he lived through personally in 1999.
George Hahn narrates Galloway's account of the dot-com boom's defining philosophy: get big fast. By 1999, 39% of all venture capital was flowing into internet companies, and 80% of US IPOs were web-related. The atmosphere was one of profound optimism — a once-in-a-generation land grab for margin and market share. Galloway makes this personal: his own company, Red Envelope, raised $30 million at a $120 million valuation on $30 million in revenues while losing $20 million annually. It was absurd, and the market eventually said so. The detail grounds what could be abstract history in lived, scarring experience.
The Pets.com story is one of the most instructive in dot-com history, and Galloway retells it with precision. The thesis — that consumers would buy pet food online — was entirely correct. It was simply a decade early. Pets.com's mascot was a Macy's Thanksgiving Day Parade balloon, and the company was one of 17 dot-coms to buy Super Bowl ads in 1999. Then it IPO'd for $82.5 million and went bankrupt in under a year. Webvan, eToys.com, and hundreds of others followed. The lesson is not that the idea was wrong but that timing is everything — and that capital can sustain a wrong-timed business for only so long before reality reasserts itself.
The B2C collapse didn't stay contained. Sun Microsystems, whose tagline was 'We're the dot in dot-com,' had built its business on powering the internet companies now going bankrupt. At its 2000 peak, Sun was valued at $205 billion — nearly as much as General Electric. Then its clients vanished. Net income of $1.8 billion in 2000 turned into a $2.4 billion loss by 2003, and the company shed 96% of its market cap before Oracle acquired what remained for $7.4 billion in 2009. DoubleClick, the era's dominant ad tech company, fell from a $12 billion valuation to $800 million before being taken private, only to be acquired by Google in 2007 for $3 billion — a reminder that not all dot-com technology was worthless, just the business models built on top of it.
The most catastrophic phase of the dot-com crash hit infrastructure last. Nortel Networks, which at its peak carried 75% of North America's internet traffic and was valued at $230 billion, saw over 90% of that value erased within a year. The mechanism was devastating: Nortel, Global Crossing, and Lucent had all extended vendor financing to dot-com clients who were now filing for bankruptcy. None of the three survived. Galloway notes that at the dot-com peak, 74% of stocks carried analyst buy recommendations — up from 60% just four years earlier — illustrating how incentive structures in financial markets drive consensual hallucination all the way to the edge of the cliff.
This is the essay's sharpest section. Galloway assembles the OpenAI evidence with the precision of a prosecutor. Leaked financials show a $21 billion loss in 2025. For every dollar subscribers pay for ChatGPT, the company spends nearly three. OpenAI is projecting $100 billion in advertising revenue by 2030, but per eMarketer, the ad business is on pace to miss that target by 90%. C-suite departures, a lawsuit from Apple, and reports of a delayed IPO complete the picture. Galloway's verdict: the business model resembles an LLM hallucination — plausible-sounding output with no grounding in reality.
Sam Altman's pitch to give US taxpayers a 5% stake in OpenAI is, in Galloway's framing, an SOS signal dressed up in the language of public benefit. The proposition — that every citizen should share in AI's profits — collapses immediately on contact with reality: AI has no profits. Galloway quotes his Markets co-host Ed Elson to drive the point home. Meanwhile, Senator Bernie Sanders is floating a sovereign wealth fund financed by a one-time 50% tax on AI equities. When the far left and far right converge on an idea, Galloway observes, it is almost always a terrible one. He compares the company's 2025 advertising spend to the cost of buying every Super Bowl ad slot for the past seven years — a marketing machine trying to paper over an existential financial hole.
The numbers were staggering — corporate AI spending rose 13x from 2025 to 2026, per The Economist. But the hangover is arriving. Axios reported that an anonymous company spent $500 million in a single month after failing to set usage limits on Claude licenses. Uber blew through its entire 2026 AI budget in four months. Now DoorDash, Meta, Microsoft, and Salesforce are pivoting from token maximalism to proven ROI. Meta CTO Andrew Bosworth's April memo is blunt: nobody should be using AI tools just for the sake of using them, and token usage alone is not a measure of impact. The pivot to sobriety is good in principle, Galloway notes, but it benefits the cheapest alternatives — including Chinese open-source models that deliver 80% of frontier performance at 20% of the cost.
Galloway ends on a note that is simultaneously bullish on AI and bearish on AI stocks. AI could be like electricity — a foundational technology that distributes value broadly to end users and new companies rather than to incumbents or early investors. The fact that productivity gains are modest at legacy software firms but explosive at AI-native companies like Anthropic and OpenAI may be early evidence of the same dynamic. The danger is concentration: the 10 most valuable S&P 500 companies account for 43% of the index, so an AI correction is not a sector story but an economic contagion event. The ultimate irony, Galloway concludes, is that AI may have already delivered a de facto wealth redistribution — just not the kind AOC envisioned. It will pass value to the people who use the technology, not to the shareholders who funded it.
Chapter 4 · 01:45
Scott Galloway briefly frames the essay before handing off to narrator George Hahn. The thesis is stated plainly from the opening: the contagion pattern of the dot-com crash — B2C first, then B2B, then infrastructure — is beginning to form again in AI, with cracks appearing at OpenAI. Hahn then sets the philosophical scene, invoking Jamie Dimon's observation that financial crises come every 5 to 7 years, noting it's been 18 years since the last one. With age comes pattern recognition, and the pattern Galloway sees — slowly, then suddenly — is one he lived through personally in 1999.
The dot-com crash followed a predictable sequence: B2C companies fell first, then B2B, then infrastructure. That same domino pattern is now forming in AI. The echoes are unmistakable — and for anyone who lived through 1999, the moment is recognizable.
Chapter 5 · 02:40
George Hahn narrates Galloway's account of the dot-com boom's defining philosophy: get big fast. By 1999, 39% of all venture capital was flowing into internet companies, and 80% of US IPOs were web-related. The atmosphere was one of profound optimism — a once-in-a-generation land grab for margin and market share. Galloway makes this personal: his own company, Red Envelope, raised $30 million at a $120 million valuation on $30 million in revenues while losing $20 million annually. It was absurd, and the market eventually said so. The detail grounds what could be abstract history in lived, scarring experience.
By 1999, 39% of all venture capital investments were being deployed into internet companies, a concentration that foreshadowed the eventual crash.
Galloway isn't just an observer of the dot-com bubble — he lived it. His firm Red Envelope raised $30 million at a $120 million valuation on revenues of $30 million while losing $20 million. The market eventually showed up and corrected the absurdity. He's drawing on those scars now.
In 1999, 80% of US IPOs were related to internet companies, reflecting the extreme exuberance of the dot-com era.
Pets.com had the right idea — consumers would buy pet food online. It was just a decade early. Chewy proved the thesis in 2011. OpenAI may be the same story: the correct thesis, but the infrastructure, economics, and market readiness aren't there yet.
Pets.com raised $82.5 million in its IPO and declared bankruptcy less than a year later — the poster child of dot-com excess.
Chapter 6 · 04:25
The Pets.com story is one of the most instructive in dot-com history, and Galloway retells it with precision. The thesis — that consumers would buy pet food online — was entirely correct. It was simply a decade early. Pets.com's mascot was a Macy's Thanksgiving Day Parade balloon, and the company was one of 17 dot-coms to buy Super Bowl ads in 1999. Then it IPO'd for $82.5 million and went bankrupt in under a year. Webvan, eToys.com, and hundreds of others followed. The lesson is not that the idea was wrong but that timing is everything — and that capital can sustain a wrong-timed business for only so long before reality reasserts itself.
Sun Microsystems was the backbone of the dot-com economy with a $205 billion peak valuation. When its internet clients went bankrupt, Sun imploded — losing 96% of its market cap. The company that powered the dot in dot-com couldn't survive the web's collapse.
Chapter 7 · 06:00
The B2C collapse didn't stay contained. Sun Microsystems, whose tagline was 'We're the dot in dot-com,' had built its business on powering the internet companies now going bankrupt. At its 2000 peak, Sun was valued at $205 billion — nearly as much as General Electric. Then its clients vanished. Net income of $1.8 billion in 2000 turned into a $2.4 billion loss by 2003, and the company shed 96% of its market cap before Oracle acquired what remained for $7.4 billion in 2009. DoubleClick, the era's dominant ad tech company, fell from a $12 billion valuation to $800 million before being taken private, only to be acquired by Google in 2007 for $3 billion — a reminder that not all dot-com technology was worthless, just the business models built on top of it.
Sun Microsystems, valued at $205 billion at its 2000 peak, shed 96% of its market cap as its internet clients went bankrupt.
Nortel carried 75% of North America's internet traffic and was valued at $230 billion at its peak. Then the dominoes hit infrastructure. Within a year, 90% of its value was gone. Along with Global Crossing and Lucent, Nortel had lent money to the same dot-coms now going bankrupt — and none of the three survived.
Chapter 8 · 07:45
The most catastrophic phase of the dot-com crash hit infrastructure last. Nortel Networks, which at its peak carried 75% of North America's internet traffic and was valued at $230 billion, saw over 90% of that value erased within a year. The mechanism was devastating: Nortel, Global Crossing, and Lucent had all extended vendor financing to dot-com clients who were now filing for bankruptcy. None of the three survived. Galloway notes that at the dot-com peak, 74% of stocks carried analyst buy recommendations — up from 60% just four years earlier — illustrating how incentive structures in financial markets drive consensual hallucination all the way to the edge of the cliff.
At its peak, Nortel Networks carried 75% of North America's internet traffic and was valued at $230 billion before losing over 90% of its value.
At the dot-com market peak, 74% of stocks had buy recommendations from analysts, up from 60% four years earlier — a sign of consensual hallucination.
OpenAI lost $21 billion in 2025. For every dollar subscribers pay for ChatGPT, the company spends nearly three. Its projected $100 billion ad business is 90% below its own forecast. This isn't a startup burning toward profitability — it's a hallucination with a balance sheet.
Chapter 9 · 09:20
This is the essay's sharpest section. Galloway assembles the OpenAI evidence with the precision of a prosecutor. Leaked financials show a $21 billion loss in 2025. For every dollar subscribers pay for ChatGPT, the company spends nearly three. OpenAI is projecting $100 billion in advertising revenue by 2030, but per eMarketer, the ad business is on pace to miss that target by 90%. C-suite departures, a lawsuit from Apple, and reports of a delayed IPO complete the picture. Galloway's verdict: the business model resembles an LLM hallucination — plausible-sounding output with no grounding in reality.
OpenAI's leaked financials reveal the company lost $21 billion in 2025, a scale of losses Galloway compares to the unsustainable burn rates of 1999 dot-com companies.
For every dollar subscribers spend on ChatGPT, OpenAI spends nearly three — a business model Galloway likens to an LLM hallucination.
OpenAI projected $100 billion in advertising revenue by 2030, but its ad business is on pace to fall short of that forecast by 90%, according to eMarketer.
Sam Altman requested a government bailout framed as an investment opportunity, offering US taxpayers a 5% stake in OpenAI — what Galloway calls cronyism, not capitalism.
Chapter 10 · 10:45
Sam Altman's pitch to give US taxpayers a 5% stake in OpenAI is, in Galloway's framing, an SOS signal dressed up in the language of public benefit. The proposition — that every citizen should share in AI's profits — collapses immediately on contact with reality: AI has no profits. Galloway quotes his Markets co-host Ed Elson to drive the point home. Meanwhile, Senator Bernie Sanders is floating a sovereign wealth fund financed by a one-time 50% tax on AI equities. When the far left and far right converge on an idea, Galloway observes, it is almost always a terrible one. He compares the company's 2025 advertising spend to the cost of buying every Super Bowl ad slot for the past seven years — a marketing machine trying to paper over an existential financial hole.
Sam Altman wants US taxpayers to take a 5% stake in OpenAI, framing it as giving citizens a share of AI's profits. But AI has no profits. Galloway calls it cronyism — a well-connected private firm extracting a government lifeline and marketing it as patriotism.
OpenAI's advertising spend in 2025 alone would have been enough to buy every Super Bowl ad spot for the past seven years, a signal of marketing-driven hype.
Circular financing deals — where AI companies fund each other — are easy to ignore as long as enterprise spending continues. In the dot-com era, the same structure made dominoes invisible until they started falling. Galloway sees the same dynamic forming between B2C and B2B AI companies and the infrastructure layer beneath them.
Chapter 11 · 12:25
The numbers were staggering — corporate AI spending rose 13x from 2025 to 2026, per The Economist. But the hangover is arriving. Axios reported that an anonymous company spent $500 million in a single month after failing to set usage limits on Claude licenses. Uber blew through its entire 2026 AI budget in four months. Now DoorDash, Meta, Microsoft, and Salesforce are pivoting from token maximalism to proven ROI. Meta CTO Andrew Bosworth's April memo is blunt: nobody should be using AI tools just for the sake of using them, and token usage alone is not a measure of impact. The pivot to sobriety is good in principle, Galloway notes, but it benefits the cheapest alternatives — including Chinese open-source models that deliver 80% of frontier performance at 20% of the cost.
Corporate AI spending rose 13x from 2025 to 2026. Then reality hit. Uber blew its entire 2026 AI budget in four months. An anonymous firm accidentally spent $500 million in a single month on Claude licenses. Now DoorDash, Meta, Microsoft, and Salesforce are all pivoting from token maximalism to proven use cases. The party is ending.
According to The Economist, corporate spending on AI increased 13x from 2025 to 2026, a surge Galloway flags as a bubble warning sign.
Uber blew through its entire AI budget for 2026 in just four months, prompting a broader enterprise pivot from AI maximalism to measured, proven use cases.
Anthropic's runaway enterprise spending is driving $47 billion in annual recurring revenue and justifying a $965 billion valuation, which Galloway sees as vulnerable.
The pivot to AI productivity measurement benefits the cheapest solutions. Open-source models from China deliver 80% of frontier model performance at 20% of the cost. As enterprises sober up and seek proven ROI, the beneficiaries may not be the US hyperscalers — they may be the low-cost alternatives.
Chapter 12 · 14:40
Galloway ends on a note that is simultaneously bullish on AI and bearish on AI stocks. AI could be like electricity — a foundational technology that distributes value broadly to end users and new companies rather than to incumbents or early investors. The fact that productivity gains are modest at legacy software firms but explosive at AI-native companies like Anthropic and OpenAI may be early evidence of the same dynamic. The danger is concentration: the 10 most valuable S&P 500 companies account for 43% of the index, so an AI correction is not a sector story but an economic contagion event. The ultimate irony, Galloway concludes, is that AI may have already delivered a de facto wealth redistribution — just not the kind AOC envisioned. It will pass value to the people who use the technology, not to the shareholders who funded it.
AI is likely to be like electricity, or jets, or the PC: transformative for society, ruinous for many early investors. The value won't concentrate in shareholder returns — it will leak to the people who use the technology. The biggest winners in the AI era may never own a single share of an AI company.
The 10 most valuable companies in the S&P 500 now account for 43% of the index's total market cap. Almost all of them are deeply tied to AI. An AI correction isn't just a tech story — it's a systemic economic event. When AI sneezes, the US economy's lungs fill with fluid.
The 10 most valuable companies in the S&P 500 account for 43% of the index's total market cap, meaning an AI correction could ripple through the entire US economy.
No indexed bits in this chapter.
OpenAI lost $21 billion in 2025. For every dollar subscribers pay for ChatGPT, the company spends nearly three. Its projected $100 billion ad business is 90% below its own forecast. This isn't a startup burning toward profitability — it's a hallucination with a balance sheet.
AI is likely to be like electricity, or jets, or the PC: transformative for society, ruinous for many early investors. The value won't concentrate in shareholder returns — it will leak to the people who use the technology. The biggest winners in the AI era may never own a single share of an AI company.
This episode
Factual claims made this episode, and whether a source was named.
By 1999, 39% of all venture capital investments were being deployed into internet companies.
In 1999, 80% of US IPOs were related to internet companies.
Pets.com raised $82.5 million in its IPO and declared bankruptcy in less than a year.
Sun Microsystems was valued at $205 billion at its 2000 peak and subsequently shed 96% of its market cap.
Sun Microsystems reported net income of $1.8 billion in 2000, which fell to $927 million in 2001, then losses of $628 million in 2002 and $2.4 billion in 2003.
Sun Microsystems was eventually acquired by Oracle for $7.4 billion in 2009.
Nortel Networks carried 75% of North America's internet traffic and was valued at $230 billion in summer 2000, then lost more than 90% of its value within a year.
At the dot-com market peak, 74% of stocks had buy recommendations, up from 60% four years earlier.
OpenAI lost $21 billion in 2025, according to its leaked financials.
For every dollar subscribers spend on ChatGPT, OpenAI spends nearly three dollars.
OpenAI is projecting $100 billion in advertising revenue by 2030 but is on pace to fall short of that forecast by 90%, according to eMarketer.
OpenAI's advertising spend in 2025 alone would have been enough to buy every Super Bowl ad spot for the past seven years.
Corporate spending on AI increased 13-fold from 2025 to 2026, according to The Economist.
An anonymous company spent $500 million in a single month after failing to put usage limits on Claude licenses for employees, according to Axios.
Uber burned through its entire AI budget for 2026 in just four months.
Palo Alto Networks CEO Nikesh Arora said widespread AI adoption depends on token costs coming down 20% this year and 90% next year.
Anthropic has $47 billion in annual recurring revenue and a $965 billion valuation.
The 10 most valuable companies in the S&P 500 account for 43% of the index's total market cap.
This episode
Criticised for proposing a 5% OpenAI stake for US taxpayers, which Galloway calls cronyism and a distress signal.
Meta's CTO, quoted from an April memo warning employees not to use AI tools just for the sake of it and that token usage is not a measure of impact.
Discussed as exhibiting dot-com-era warning signs: $21B losses, unsustainable unit economics, delayed IPO, and a government bailout pitch.
Poster child of the dot-com bubble; raised $82.5M in IPO and went bankrupt within a year despite a correct long-term thesis.
B2B infrastructure company that lost 96% of its $205B peak market cap after its dot-com clients went bankrupt; acquired by Oracle.
Cited as beneficiary of runaway enterprise AI spending, with $47B ARR and a $965B valuation Galloway views as vulnerable.
Carried 75% of North America's internet traffic at peak; lost over 90% of its $230B value in the telecom crash.
Used to illustrate that Pets.com had the right thesis but was a decade early — Chewy proved the model viable when founded in 2011.
Dot-com-era ad company that fell from $12B to $800M in the crash; later acquired by Google for $3B, showing some dot-com tech had lasting value.
Named as a lead underwriter for SpaceX with a buy recommendation and $205 price target, used to illustrate current analyst hype.
Cited as one of several major companies pivoting from token maximalism to proven AI use cases; CTO Andrew Bosworth's memo quoted.
Scott Galloway's own dot-com-era firm, raised $30M at a $120M valuation while losing $20M — a personal example of bubble-era excess.
Cited as an example of current Wall Street buy-recommendation hype; Goldman Sachs and Morgan Stanley have price targets of $205 and $300.
Cited as an example of enterprise AI overspending, having burned through its entire 2026 AI budget in just four months.
OpenAI's flagship product, cited as generating revenue at a fraction of the cost to serve — nearly $3 spent per $1 earned.
Stats
We use essential and analytics cookies to run Vuci. To understand how the site is used: Privacy Policy.
Install Vuci on your phone
Add it to your home screen for a faster, app-like experience.
Install Vuci on your phone
Tap the Share button, then “Add to Home Screen”.
A new version is available
Reload to get the latest Vuci.