Jul 23, 2025 · 1h 33m · big-technology
Are AI's Economics Unsustainable? — With Ed Zitron
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Technology critic Ed Zitron joins Alex Kantrowitz on the Big Technology Podcast to conduct a rigorous financial audit of the generative AI industry, arguing that astronomical capital expenditures, negligible software revenues, and fundamental reliability gaps have created an unsustainable economic bubble.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Alex holds 22.5% of the talking time here. How this is scored →
Alex as informed peer 5.6 Guest teaching 5.9 Guest disagreement 6.7 Alex pushing back 5.4
speaking balance: gold is Alex, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 30:21 Dismissal of theoretical AI breakthroughs
Zitron rejects the premise of inevitable model breakthroughs with sharp sarcasm, stating 'If my grandmother had wheels, she'd be a bicycle' before tearing into executive claims.
Hardest push from Alex ▶ 18:09 Challenging ad monetization assumptions using Apple ATT dataKantrowitz challenges Zitron's assertion that advertisers will abandon AI search by demonstrating how marketers continued high spend on Meta despite degraded tracking signals post-ATT.
Biggest teaching moment ▶ 5:38 Deconstruction of Google's search advertising moatZitron educates Kantrowitz on why matching Google's search UX is irrelevant without owning the underlying ad exchanges, double-sided bidding networks, and global physical infrastructure.
Alex holds their own ▶ 1:04:06 Citing Sergey Brin on the limits of brute-force GPU scalingKantrowitz demonstrates first-hand industry reporting by citing his discussion with Google co-founder Sergey Brin regarding the shift away from raw compute scaling toward architectural innovation.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Alex as informed peer | Guest teaching | Guest disagreement | Alex pushing back | Why |
|---|---|---|---|---|---|---|
| Assessing AI Industry Capital Expenditures and Lack of Profitability | 4 | 5 | 7 | 4 | Ed Zitron opens with a barrage of capital expenditure figures comparing Microsoft, Amazon, and OpenAI's spend against actual generative AI revenues. Kantrowitz introduces his role as a fair pressure-tester of Zitron's arguments. | |
| Search as a Product Versus Search as an Advertising Business | 6 | 7 | 6 | 6 | Kantrowitz argues that even capturing a fraction of Google's market cap makes AI search viable. Zitron clearly separates search as a product from search as an advertising business, explaining Google's massive ad-tech rails and infrastructure moat. | |
| Generative AI Search Reliability and Advertising Monetization Obstacles | 7 | 5 | 6 | 7 | Kantrowitz counters Zitron by referencing how advertisers continued spending on Meta even after Apple's App Tracking Transparency reduced attribution signal. Zitron counters that OpenAI lacks ad networks, sales teams, and established scale. | |
| Coding Copilots, Market Size Realities, and Venture Capital Hype | 5 | 6 | 7 | 5 | Zitron argues coding copilots serve a relatively modest TAM and won't replace human developers due to hallucination risks and catastrophic bugs. Kantrowitz probes the VC model of funding probabilistic breakthroughs. | |
| Scrutinizing Model Improvement Claims, Benchmark Limitations, and Agent Failures | 6 | 6 | 8 | 7 | Kantrowitz pushes back using qualitative model improvements ('vibes') and multi-step reasoning capabilities. Zitron forcefully rejects benchmark hype, mocking Sundar Pichai's agent demos and citing failure rates in multi-step task execution. | |
| Deconstructing Executive Narratives Around White-Collar Labor Replacement | 5 | 7 | 8 | 5 | Kantrowitz asks if slow corporate bureaucracy explains the lack of AI deployment rather than tech failure. Zitron counters by highlighting that no major enterprise has produced a working labor-replacement agent despite breathless CEO promises. | |
| AI Companionship Markets, Infrastructure Valuations, and Circular Deals | 6 | 7 | 7 | 6 | Kantrowitz brings up AI companionship and uses CoreWeave's surging valuation to show market appetite for narrative. Zitron unpacks CoreWeave and OpenAI's asset-light structures and circular debt-backed GPU financing. | |
| Algorithmic Efficiency Realities, Conversion Rates, and OpenAI Financial Losses | 6 | 6 | 7 | 6 | Kantrowitz notes declining token costs and mixture-of-experts architectures as proof of algorithmic efficiency. Zitron counters that lowering sale prices while burning billions does not equate to a sustainable profit model. | |
| Stock Market Concentration, NVIDIA Dependency, and Data Center Limits | 7 | 5 | 7 | 6 | Zitron warns of systemic market risks given S&P 500 concentration in Nvidia. Kantrowitz cites his direct reporting with Sergey Brin on how tech leaders recognize diminishing returns from brute-force scale and are pivoting to algorithmic improvements. | |
| Escalating Tensions Between OpenAI and Microsoft Over Equity and IP | 5 | 8 | 7 | 4 | Zitron details the high-stakes contract disputes between OpenAI and Microsoft over IP access, enterprise undercutting, and the complex debt requirements of SoftBank's investment tranche. | |
| Maintaining PR Firewalls and Reflecting on the Updog Industry Prank | 4 | 3 | 2 | 3 | Kantrowitz shifts to Zitron's dual role in PR and journalism, leading to an amusing discussion of Zitron's viral Updog industry prank at CES. | |
| Public Anxiety Over Tech Hype, AGI Fallacies, and Episode Conclusion | 6 | 6 | 8 | 6 | Kantrowitz defends the inherent utility of generative AI over past crypto hype and defends journalistic coverage. Zitron lambasts credulous AGI narratives and media figures who celebrate replacing human workers. |
Statements from this episode (34)
OpenAI will burn $12B to $13B this year, Zitron projects
“They will burn probably 12, thirteen billion dollars after revenue this year. That's based on projections.”
OpenAI has no viable path to profitability, Zitron argues
“They also have no path to profitability. They don't have one.”
Microsoft's $13B AI revenue largely comes from OpenAI cloud spend
“The information also reported that Microsoft will only make about thirteen billion dollars, not profit, just revenue, on AI this year. Ten billion of that is OpenAI's Azure cloud spend. Three billion is them selling Copala.”
Amazon is spending $105B in capex for $5B AI revenue
“There's an analyst, I was quoted by Laura Brattner at Yahoo Finance, who said that he only thinks that Amazon is going to make five billion dollars in revenue, again, not profit, this year on AI. They are spending one hundred and five billion dollars in capita…”
Total generative AI industry revenue is only $35B to $40B
“What the current combined revenue of all the generative AI companies is like And that's including the big tech companies. It's about 35, forty billion dollars.”
Building a Google-scale search business will cost OpenAI tens of billions
“Making that into a search size business is an entirely different thing and will cost them tens of billions of dollars.”
The Stargate data center corporate entity hasn't even been formed
“The Stargate entity does not exist. They haven't formed it yet.”
OpenAI hasn't signed a compute contract for the Abilene data center
“OpenAI, according to the information, hasn't even signed a contract for the compute in, in Abilene.”
OpenAI effectively operates as a Microsoft subsidiary, Zitron argues
“OpenAI is a Microsoft subsidiary. That's what everyone needs to just accept right now, and what's happening in the news, which I imagine we'll talk in a bit.”
Ad dollars will follow users to LLM search despite weaker attribution
“If people move from Google search to OpenAI or to some other LLM search, my anticipation is that the money won't go away. I think marketers have gotten, and advertisers have gotten used, so used to spending online that they will, they'll be willing to spend ev…”
Perplexity sought a massive $50 CPM for its advertising product
“Perplexity hasn't. They wanted 50 dollars a CPM!”
The AI coding assistant total addressable market is $30B to $50B
“I think this is a 30 to fifty billion dollar TAM, total addressable market business. I think that, what's that, the IDE market, development environment, I think that's like a thirteen billion dollar one.”
You.com received a $1B valuation on under $20M in revenue
“You go and you see, like, a company like u.com. Everyone's saying, wow, they got a valuation of a billion dollars. Annualized revenue of, like, oh, I'm gonna misquote this. It was, like, 12,000,020 million dollars. That's insanely small, man.”
VCs rarely inspect accounts or codebases when investing in AI startups
“I had someone tell me, a source tell me, that there are, it's very rare that venture capitalists see the books, the actual accounts, and they almost never see the code base.”
Salesforce research shows AI agents fail most multi-step processes
“The reason that we keep hearing about agents, but never about what agents can do, is because the most common feature of agents is that they fail. There was a Salesforce paper that came out fairly recently that said, I think that they just categorically break d…”
OpenAI's o3 model failed to calculate a basic spatial measurement
“I tried O-three out the other day. Yeah. I took a photo of a thing I had hung up, And I said, how much space from the bottom of that photo, of that picture, the poster, to the floor? It took four minutes. It wrote multiple Python scripts to give me the wrong a…”
Salesforce expects zero AI revenue growth in 2025 despite agent rebrand
“The information reported a few months ago that Salesforce does not expect any growth from AI this year. That is absolutely bonkers for a company that's rebranded, and I paraphrase here, as an agent first company.”
OpenAI holds no technological advantage over competitors besides scale
“OpenAI isn't ahead of everyone else other than scale.”
AI companion bots are an easily commoditized and replaceable product category
“AI companions feels like something that ChatGPT, again, because they are all over the place with all their use cases, it's something that they're getting because they're the biggest name said to everyone at all times. It's something that can be replaced by any…”
Consumers will easily pay $20 per month for AI companions
“If you have AI that replaces a friend for you, or is your companion, you would easily, I mean, pay 20 dollars a month, I think that would be an easy subscription to charge.”
CoreWeave will raise another $10 billion in stock sales, Zitron predicts
“CoreWeave will probably raise another ten billion dollars by selling another share sale.”
OpenAI is an asset-light company relying heavily on shared Microsoft IP
“OpenAI is an asset-light business with research and IP that's owned by another company. They don't have much to trade other than their name.”
OpenAI's contractual payments to CoreWeave escalate in October 2025
“OpenAI starts paying Corweave in October, twenty-twenty-five, just as Corweave's second loan, DDT-L-II, starts requiring them to pay probably more than OpenAI will be paying them.”
Perplexity gave $30 million in discounts to reach 50 million users
“Well, first of all, they're an insanely badly run company. They did, like, thirty-five million. They lost, I forget exactly how much they lost, but they did refunds or discounts of, like, thirty million dollars. They're literally giving money away to make peop…”
OpenAI reports weekly users to mask terrible subscriber conversion rates
“They talk weekly. And the reason is, because if you compared what their real monthly active users would be, five hundred million weekly, so I'm gonna guess seven hundred million monthly? Divided by the 15.5 million customers that pay for it. That's a dog's doo…”
Anthropic lost approximately $5.2 billion in 2024 despite strong API sales
“Anthropic is doing better on API. They're selling more, a larger percentage of their business is API, but they still lost like, 5.2 billion dollars last year.”
Sergey Brin believes future AI progress will be primarily algorithmic
“Just to go back to a conversation that I had a couple of weeks ago, or now month plus with Sergey Brin, where he said he thinks that the improvement is going to largely be algorithmic. Of these models, meaning not by adding more GPUs and data by like actually …”
OpenAI will not achieve $100 billion in annual revenue, Zitron predicts
“Even if OpenAI was profitable, okay, they're the biggest AI company, cool. Are they going to a hundred billion a year? Bullshit. No.”
OpenAI is directly undercutting Microsoft's enterprise subscription deals
“OpenAI has been undercutting Microsoft in deals, selling their models, and undercutting their enterprise subscription deals.”
Google is quietly routing OpenAI's cloud compute contracts through CoreWeave
“OpenAI is contracting Google for Cloud Compute. Google is contracting CoreWeave to serve that compute.”
OpenAI must convert to for-profit by year-end to keep SoftBank funding
“By the end of the year, OpenAI must convert to a for-profit entity, or SoftBank does not have to give them more than twenty billion dollars total.”
SoftBank required a $15B bridge loan from 21 banks for OpenAI
“SoftBank, to give OpenAI that money, And to buy Ampere for, I think, six billion or something, they had to get a one-year, fifteen billion dollar convert, convertible bridge loan, even. And it, they had to go to 21 banks.”