If a future film is ever made about the bursting of the artificial intelligence bubble, one British commentator is positioning himself as its prophetic star. Ed Zitron, the podcaster and writer behind the newsletter Where's Your Ed At, has become a cult figure for his blunt, brash scepticism towards the multi-trillion-dollar AI hype cycle. From his base in Las Vegas, the 39-year-old Londoner is shouting that the emperor has no clothes, and a growing audience is starting to listen.
The Anatomy of a Skeptic's Argument
Zitron's critique is twofold, targeting both the technological capabilities of generative AI and the precarious financial architecture supporting its boom. He began scrutinising the sector in 2023, a year after ChatGPT's seismic launch, and came away deeply unimpressed. "The more I looked, the more confused I became," Zitron explains. "Large language models very clearly did not do the things people were excited about, and they didn't have any path to doing them either."
He dismisses the notion that current LLMs possess true intelligence, comparing them to a pair of dice. He argues they are fundamentally transformer-based systems using probability to generate the next token, reliant on vast data corpora. "We would not credit an Excel formula with intelligence, and we should not credit generative AI as intelligent," he states. Despite industry warnings, like that from Anthropic's CEO Dario Amodei about AI wiping out half of entry-level white-collar jobs, Zitron is confident the current generation of models is incapable of such disruption, pointing to their static efficacy and persistent issues with hallucinations.
The Trillion-Dollar House of Cards
The second, and perhaps more damning, part of Zitron's thesis concerns the economics. He describes an ecosystem propped up by eye-watering investment with little evidence of sustainable demand or profitability. The "Magnificent Seven" tech giants—Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla—now constitute 34% of the S&P 500. While Nvidia, as the dominant GPU maker, is "printing money," Zitron contends everyone else is spending billions they may never recoup.
The costs are astronomical. Building AI datacentres requires tens of thousands of GPUs, each costing over $50,000, plus immense software, real estate, energy, and water. The price for 1GW of AI datacentre capacity is estimated at a staggering $35 billion. Yet, on the demand side, the picture is hazy. Zitron forensically details a carousel of circular deals where companies like Nvidia invest in OpenAI, which then uses the cash to buy Nvidia chips. He claims that outside the hyperscalers (Google, Meta, Amazon, Microsoft), there is "less than a billion dollars total in AI compute revenue in 2025."
Profitability remains elusive. With most of ChatGPT's 800 million users paying nothing, and each query incurring a variable compute cost, the model is economically inverted. "The more someone is a power user... the more they're going to cost you," Zitron notes. There are no economies of scale, and unsatisfactory answers that require re-generation simply burn more cash.
From Outsider to Leading Voice of Doubt
Zitron is no longer a lone voice. He aligns with critics like author Cory Doctorow and finds himself part of a building backlash. Local groups oppose environmentally destructive datacentres, creators sue over copyright, and public frustration grows with poorly integrated AI features. Even figures like Michael "Big Short" Burry are betting against Nvidia, and speculation mounts, with a New York Times op-ed suggesting OpenAI could run out of money within 18 months.
Zitron, who has dyspraxia and ADHD, credits his detail-oriented nature and background in tech PR for his analytical approach. He insists he doesn't hate technology but despises what the industry has become. "I love technology, but I hate what the tech industry is doing," he says, framing AI as the logical, damaging conclusion of growth-at-all-costs neoliberalism.
As major tech companies prepare to report annual earnings, Zitron watches closely. If targets are missed, it could trigger a sector-wide reckoning. He jokes we might be witnessing "the largest laser-tag arena construction of all time." For now, the self-described asshole over the details continues his mission: not to write fan fiction about AI's potential, but to stubbornly, vocally pursue the truth about its present.
