
Aug 7, 2026 · 24 min
Ed Zitron challenges the economics behind AI’s trillion-dollar buildout
Monologue: The Pale Horsemen Arrive
The episode questions whether AI revenues reflect durable customer demand or an industry propped up by subsidies, financial engineering, and social pressure.
- 1AI usage may appear stronger than it is because companies subsidize services and customers feel pressure to adopt them.
- 2Microsoft, Google, and Amazon increasingly depend on OpenAI and Anthropic to generate AI revenue and validate infrastructure spending.
- 3Reported growth may not justify the roughly trillion dollars committed to AI infrastructure, leaving the market exposed to a demand shock.
Don't miss
Ed Zitron contrasts roughly a trillion dollars in AI infrastructure commitments with what he sees as insufficient genuine customer demand.
The brief
Ed Zitron opens by arguing that the AI market is experiencing a demand bubble, with financial estimates obscuring whether customers genuinely want services at the scale implied by spending.
His central claim is that social pressure and below-cost pricing inflate apparent demand, allowing AI companies to present subsidized usage as evidence of a durable market.
Microsoft, Google, and Amazon appear tied to AI revenue generated through relationships with OpenAI and Anthropic, raising questions about how much growth is truly independent.
The episode’s sharpest tension is between roughly a trillion dollars in committed AI infrastructure spending and the weaker customer demand Zitron believes that spending actually demonstrates.
Zitron’s conclusion is a warning about an unstable market: if subsidies and financial engineering recede, the industry may have to confront demand far below its current expectations.
What was said on this episode
36 statements · 2 positive · 32 negative · 2 neutral
Microsoft has spent over $200 billion creating infrastructure for OpenAI-related demand.
“Microsoft has spent over $200 billion”
Listen at 4:09
OpenAI represents 7% of Microsoft revenue and depends on continual venture funding.
“a customer that is now worth 7% of its annual revenue that cannot sustain its existence without near-constant flows of venture capital funding”
Listen at 4:21
Microsoft’s fiscal-year 2026 year-over-year growth was approximately 17.7%
“Microsoft's year-over-year growth for fiscal year 2026 was about 17.7%”
Listen at 4:30
Excluding OpenAI revenue, Microsoft’s fiscal-year growth was approximately 9.24%
“when you remove OpenAI's $24.1 billion in revenue, which comes from OpenAI's compute spend and revenue share, Microsoft's year-over-year growth was more like 9.24%”
Listen at 4:35
Excluding OpenAI revenue, Microsoft growth was approximately 9.24%.
“Microsoft's year-over-year growth was more like 9.24%”
Listen at 4:42
Microsoft spent $35.8 billion on capital expenditures in fiscal 2026’s fourth quarter.
“Microsoft spent $35.8 billion in the 4th quarter of the fiscal year of 2026”
Listen at 4:59
OpenAI will run out of money and accounts for nearly half of Microsoft’s recent growth.
“OpenAI, arguably Microsoft's largest customer and responsible for nearly half of its growth in the last fiscal year, is gonna run out of money”
Listen at 5:17
Revenue shortfalls caused by insolvency, weak demand, or capacity constraints will anger Wall Street.
“Any failure for these revenues to appear out of insolvency, lack of demand, or unbuilt data center capacity, meaning they can't get paid, will piss off a Wall Street”
Listen at 6:21
Microsoft depends on OpenAI for future revenue growth.
“Microsoft is now dependent on OpenAI for future revenue growth”
Listen at 6:59
Microsoft’s OpenAI investment created dependence rather than an independent growth engine
“Microsoft believed it would create an independent ultra-growth engine from its investments in OpenAI. Instead, it has created a parasite.”
Listen at 7:20
Microsoft spent $260 billion on infrastructure without demand beyond OpenAI and Anthropic.
“Microsoft has spent $260 billion on infrastructure for demand that does not exist outside of 2 different companies”
Listen at 7:33
OpenAI and Anthropic must aggressively increase compute spending annually.
“OpenAI and Anthropic must grow their compute spend aggressively every single year to keep up with the expectations of analysts”
Listen at 7:45
There is not meaningful real demand for AI.
“there's not really real demand for AI”
Listen at 8:06
Demand does not support over 300 gigawatts of planned data-center capacity.
“The demand is not there for the over 300 gigawatts of data center capacity in planning”
Listen at 9:12
Sam Altman has signed over $750 billion in commitments.
“Sam Altman has also signed up for over $750 billion worth of commitments”
Listen at 9:31
Anthropic has made over $300 billion in commitments.
“Dario Amadei's Anthropic has made over $300 billion worth of commitments”
Listen at 9:53
OpenAI and Anthropic sell roughly $20–$40 of services for each dollar.
“they're selling $20 or $40 for a dollar”
Listen at 13:17
Nvidia’s circular financing supports the continued survival of neo-cloud companies.
“The only reason that any neo-cloud is still alive is Nvidia's circular financing rituals”
Listen at 13:36
Hyperscaler AI revenues are small relative to their capital expenditures.
“The reason that Microsoft and Google and Amazon have been so squirrelly about their AI revenues is that they know that they're pathetic compared to their CapEx”
Listen at 14:19
“Google and Amazon have invested a combined $15 billion in Anthropic”
Listen at 14:40
“Amazon's invested a combined $50 billion in OpenAI”
Listen at 14:45
Hyperscalers overbuilt capacity and funded their largest customers.
“Hyperscalers overbuilt capacity, funded their largest customers”
Listen at 15:10
Hyperscalers made a major strategic mistake in their AI investments
“The reality is that hyperscalers made a huge mistake”
Listen at 15:53
Hyperscalers misjudged AI capabilities and OpenAI and Anthropic’s growth rates.
“hyperscalers made a huge mistake, both in what AI could do, how fast OpenAI and Anthropic could grow”
Listen at 15:54
Meaningful diverse AI revenue would be prominently disclosed by these companies.
“If these companies had diverse and meaningful AI revenues, they would be screaming about them from the rooftops”
Listen at 16:18
Funding for OpenAI and Anthropic will eventually run out.
“At some point, the money runs out to fund these 2 companies”
Listen at 17:36
The AI bubble is a financial crisis
“the AI bubble is a psyop, a melodrama, a financial crisis”
Listen at 17:45
AI will consume resources during a cultural, financial, and ecological collapse.
“AI has become a parasitic presence culturally, financially, and ecologically, and it will exhaust as much as it can in its collapse”
Listen at 18:20
A bailout cannot restart growth for major hyperscalers after growth fully slows.
“A bailout cannot make growth restart for Google, Microsoft, Meta, Amazon, or Oracle once it fully slows”
Listen at 18:35
AI growth is finite outside semiconductor companies and subsidized AI labs.
“AI's growth is finite”
Listen at 19:01
Major technology companies will eventually exhaust resources and decline
“I think these companies simply exhaust their resources at some point and drop off.”
Listen at 19:24
The era of hyperscale growth is ending.
“I think the era of hyperscale growth is ending”
Listen at 19:28
The era of hyperscale growth is ending
“the era of hyperscale growth is ending”
Listen at 19:28
Hyperscalers rely on venture capital and debt markets to sustain AI-driven growth
“the only reason they're so desperately joining the no IT loads refused AI compute cum dump is because they know that the only way to keep growth going is to hope they can drain the venture capital and debt markets infinitely by proxy.”
Listen at 19:31
Current growth metrics will eventually stop rising and decline.
“at some point they'll stop going higher, and then they'll start going lower”
Listen at 19:58
Microsoft, Google, and Amazon growth is primarily driven by OpenAI and Anthropic.
“Microsoft, Google, and Amazon numbers are going up, but that's really only happening because of OpenAI and Anthropic”
Listen at 20:10
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.