# This Tech Expert Thinks It's 'Time to Call Bullshit on the AI Industry' > Source: > Published: 2026-08-11 16:42:03+00:00 LAS VEGAS—Most of the panels at [the Ai4 conference here](https://ai4.io/) last week were overrun with enthusiasm for the possibilities AI is opening. Ed Zitron’s talk was not one of them. In a 20-minute session, the [tech publicist-](https://www.pcmag.com/news/the-best-work-from-home-cities-for-2022/spring-valley-nv)turned[-tech critic](https://www.wheresyoured.at/) tore into the business models of OpenAI and Anthropic and warned of financial doom for the generative-AI industry and for everyday Americans whose retirement holdings could be within the blast radius. His thesis, also unpacked in a [blog post](https://www.wheresyoured.at/the-ai-demand-bubble/) and[ podcast](https://podcasts.apple.com/us/podcast/monologue-the-pale-horsemen-arrive/id1730587238?i=1000780384187) episode: Now that AI providers have switched to usage-based billing and [sent some customers' costs skyrocketing](/ai/165301/github-copilot-costs-skyrocket-as-users-are-pushed-to-per-token-billing), they can’t make enough money to cover hundreds of billions of dollars of data-center expenses that they’ve already locked in. “They know regular consumers are not going to pay,” Zitron said. He rejected the idea that companies still have time to figure out basic business models: “With four years and over a trillion dollars into this, we’re anything but early days.” But the likes of OpenAI CEO Sam Altman (“clammy Sam Altman” in Zitron’s first reference) keep pushing companies to get on board anyway. “There is an industrial-sized pressure on each and every person in this room, and most of the people in the economy, to adopt AI services with no regard for what they can do, what their outcomes are, or the actual costs,” he declared. “It is a con, it is a grift, and it is a result of a tech industry obsessed with growth.” ## A Dystopian Duo OpenAI served as the chief villain for Zitron, who this summer obtained copies of that firm’s financial documents and focused [his June 15 write-up](https://www.wheresyoured.at/exclusive-openai-financials/) on the company's $38.5 billion net loss. Zitron also shared those reports [with the ](https://www.ft.com/content/e15b0d7e-ff6b-4f16-ba7a-4068feddb828?syn-25a6b1a6=1)[Financial Times](https://www.ft.com/content/e15b0d7e-ff6b-4f16-ba7a-4068feddb828?syn-25a6b1a6=1), and that London-based broadsheet did not have such a dire read. It [calculated that OpenAI really lost about $8 billion](https://arstechnica.com/ai/2026/06/leaked-financial-docs-show-openai-is-losing-billions-of-dollars-a-year/) on $13.07 billion in revenue, after a roughly $30 billion write-down required by OpenAI’s [2025 conversion to “public-benefit corporation” status](https://openai.com/our-structure/). Evoking the speculative bubble that [led to the 2008 financial crisis](https://www.economicsobservatory.com/why-did-lehman-brothers-fail), Zitron told Ai4 attendees that OpenAI was “the Lehman Brothers of AI.” Many observers of the industry have pointed to Anthropic as plotting a different, [more careful course](/ai/165160/pope-leo-xiv-unchecked-ai-development-risks-building-a-new-tower-of-babel) than OpenAI, but Zitron didn’t buy that or the notion that Anthropic CEO Dario Amodei (“Wario Amodei” in his diss) brings a more serious perspective than Altman. For example, while Anthropic hasn’t lurched into costly consumer ventures like the [now-defunct Sora video generator](/ai/164120/sora-video-generator-reportedly-costs-openai-1-million-a-day) and its [purchase of former Apple design chief Jony Ive’s AI-device startup](/ai/158164/openai-wants-to-eventually-sell-you-a-jony-ive-designed-ai-gadget), Zitron sees no further sustainability in Anthropic’s overall AI subscription business model. He also swatted aside the *Wall Street Journal* [reporting in May](https://www.wsj.com/tech/ai/mind-blowing-growth-is-about-to-propel-anthropic-into-its-first-profitable-quarter-7edbf2f4?st=1w4bQk&reflink=desktopwebshare_permalink) that Anthropic [eked out a narrow profit](/ai/165295/anthropic-moves-to-go-public-ahead-of-openai) in Q1, calling that an accounting gimmick enabled by xAI giving Anthropic a break on [renting its data-center capacity](/ai/164797/anthropic-wants-in-on-elon-musks-space-data-centers-colossus-supercompute). SpaceX’s [IPO filing](/networking/165099/spacex-ipo-filing-offers-first-glimpse-at-subscriber-numbers-financials) described this discount as “a reduced fee” for May and June. OpenAI and Anthropic’s fates don’t only concern their employees, though. They also figure heavily in the AI plans of more established tech firms: Zitron ticked off estimates from analysts at Barclays, Wells Fargo, and UBS that Amazon, Google, and Microsoft see around 70% of their AI revenue coming from just those two AI providers. Anthropic and OpenAI have each raised massive amounts of venture capital funding just this year ([$65 billion](https://www.anthropic.com/news/series-h) for the former, [$122 billion](https://openai.com/index/accelerating-the-next-phase-ai/) for the latter), which they will need to repay at some point. “These are all the hallmarks of a quick visit to Enronville,” Zitron fumed. “I think that this is why so many people around hate AI. That and the stealing and the poisoning of Black neighborhoods," a reference to the [polluting gas turbines at xAI’s Memphis data center](/ai/166483/facing-protests-xai-to-phase-out-69-gas-turbines-at-memphis-data-center). "And the egregious data center capex is the fact that these people can raise as much money as they want based on nothing.” ## Growth at Growing Costs The number-go-up mindset that Zitron sees in the AI business—something I earlier saw him excoriate in a [talk at the 2024 Web Summit conference in Lisbon](https://www.youtube.com/watch?v=7Slib2bbMs4&feature=youtu.be)—looks most obvious in the feeding frenzy of AI data center construction. Or as he put it: “this entire industry that's propped up by venture capital, hype and debt.” Citing an estimate of 225 gigawatts of planned compute capacity and an operating cost of $12 million per megawatt, Zitron said data center operators will need to hit $2.7 trillion in annual compute revenue. *The Economist* did its own math in July and [calculated](https://www.economist.com/finance-and-economics/2026/07/28/ai-revenues-are-growing-fast-but-not-fast-enough) that the AI data-center industry would need to generate $2.4 trillion in revenue to cover its capital expenses. Not all of Zitron’s jabs may have landed with the audience, particularly his suggestion that large language models can’t yield reliable returns on those expenses. “You cannot measure the ROI of large language models,” he opined, citing Uber COO Andrew McDonald’s [recent admission](https://www.theverge.com/transportation/937116/uber-ai-investment-hard-to-justify?view_token=eyJhbGciOiJIUzI1NiJ9.eyJpZCI6IjFaRW1ZZmpraHMiLCJwIjoiL3RyYW5zcG9ydGF0aW9uLzkzNzExNi91YmVyLWFpLWludmVzdG1lbnQtaGFyZC10by1qdXN0aWZ5IiwiZXhwIjoxNzg2NTAwMDIzLCJpYXQiOjE3ODYwNjgwMjN9.igzRYwqzRo8tgG4uNO2nxIBj9-_nNoTg9aYZYveMauA&utm_medium=gift-link) that he was struggling to see how the company’s AI usage was leading to “useful consumer features.” Other Ai4 talks and panels offered evidence to the contrary in fields ranging from pharmaceutical research to autonomous vehicles. And over the rest of the week, speakers at the [Black Hat](/black-hat) security conference down the Strip provided their own testimonials about [AI’s effectiveness in finding software vulnerabilities](/ai/165147/companies-using-anthropics-mythos-ai-uncover-10k-serious-software-bugs). In general, [narrower, more focused models](/security/159474/this-ai-skeptic-thinks-ai-is-bringing-human-brains-down-to-its-level) trained on a particular topic—[hurricane forecasting](https://www.wired.com/story/deepmind-ai-model-can-predict-hurricanes-earlier/), for example—keep racking up achievements that seem worth more attention than vague promises of [AGI](/ai/165137/deepmind-ceo-agi-is-coming-soon-but-heres-the-test-it-must-pass-first) (“artificial general intelligence”) or [“superintelligence.”](/ai/159120/not-one-but-several-zuckerberg-eyes-more-mega-data-centers-for-ai) ## Data-Center Doubts But Zitron has plenty of company in voicing concerns over how the money keeps rolling out for those enormous data centers. On Monday, Nvidia [announced](https://nvidianews.nvidia.com/news/nvidia-partners-with-apollo-blackrock-blackstone-brookfield-goldman-sachs-and-kkr-to-establish-ai-compute-infrastructure-financing-platforms-to-mobilize-over-500-billion-of-third-party-capital) a series of [deals with six asset managers](https://www.cnbc.com/2026/08/10/nvidia-wall-street-asset-managers-500-billion-ai-push.html)—Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR—to “mobilize over $500 billion of third-party capital for the buildout of AI infrastructure over time.” The same day, the *Financial Times* [reported](https://www.ft.com/content/a0a07cce-6d19-4b1e-a73b-9855a06ba7b3) that Goldman analysts had pored over footnotes in earnings reports and identified a trillion dollars’ worth of commitments to lease data centers that did not appear on balance sheets, because accounting rules only factor in those obligations when the leases start. The sector-wide trend of [circular financing](https://am.jpmorgan.com/us/en/asset-management/adv/insights/market-insights/market-updates/on-the-minds-of-investors/does-circularity-in-ai-deals-warn-of-a-bubble/), in which Nvidia funds the construction of a data center around its processors, led CNBC’s famously bullish on-air personality, Jim Cramer, to [worry on air at the end of June](https://www.cnbc.com/2026/07/27/jim-cramer-warns-ai-circular-financing-echoes-dot-com-bubble.html) that he was looking at a sequel to the [dot-com bust](/opinions/92873/did-we-learn-nothing-from-the-dot-com-bust) of 2000. (Unlike the dark fiber-optic connectivity left under and above streets after that crash, the processors in AI data centers would face a [fast depreciation](https://www.cnbc.com/2025/11/14/ai-gpu-depreciation-coreweave-nvidia-michael-burry.html).) And Anthropic and OpenAI might even count as conservative compared with other actors in this space. In a lengthy feature published Sunday, the *New York Times* unpacked how Oracle executive chairman and CTO Larry Ellison’s aggressive spending on AI has left his company so heavily leveraged [that its bonds are nearing junk ratings](https://www.nytimes.com/2026/07/31/magazine/larry-ellison-ai-oracle.html). Meta’s rush to [build out gigantic data centers](/ai/166129/meta-plans-huge-data-center-expansion-how-will-it-use-that-excess-compute) at the direction of Mark Zuckerberg, who previously showed his feel for consumer appetites by [vaporizing more than $70 billion](/vr-1/164299/the-70-billion-delusion-why-zuckerbergs-metaverse-was-dead-on-arrival) on his “metaverse” ambitions, looks delusional as well. SpaceX’s ambitions of deploying a million [data-center satellites](/ai/163932/musk-offers-sneak-peek-at-orbiting-data-centers-theyre-bigger-than-the-iss), which in a best-case scenario calculated by Ars Technica space reporter Eric Berger would [require 10 launches of its Starship rocket ](https://arstechnica.com/space/2026/07/how-hard-is-it-to-build-orbital-data-centers-actually/)[every day](https://arstechnica.com/space/2026/07/how-hard-is-it-to-build-orbital-data-centers-actually/), seem even more of a technological fever dream. Zitron wrapped up by warning who might wind up paying the dues for those dreams. “Private credit is sinking hundreds of billions of dollars into AI data centers for customers that do not exist, and they're funding it by buying retirement and insurance funds and working with public pension funds,” he said, suggesting that attendees look up [Pitchbook’s coverage of the topic](https://pitchbook.com/news/articles/data-center-overbuilding-poses-risk-to-pension-dollars). “We are going to see the losses socialized.” His conclusion drew applause in the room, but it was unclear how many other conference attendees were ready to hear it. “It is time to call bullshit on the AI industry,” Zitron vented. “I want the world to be a better place, and it starts with building a software industry on meaningful returns on investment and actual useful tools.” *Disclosures: Ziff Davis, PCMag's parent company, filed a lawsuit against OpenAI in April 2025, alleging it infringed Ziff Davis copyrights in training and operating its AI systems. I moderated two panels at Ai4, with the conference paying me a speaking fee, and Zitron has had me as a guest on his podcast *[twice](https://www.iheart.com/podcast/1119-better-offline-150284547/search?q=pegoraro)*.*