On May 6, 2025, a new social media star named Tilly Norwood made her debut on Instagram, introducing herself as an aspiring actress. With her breezy captions—"Hey besties"—and polished photos, Tilly quickly amassed hundreds of thousands of followers on both Instagram and TikTok. Followers found her relatable, witty, and seemingly just another influencer chasing dreams in the big city. There was just one catch: Tilly Norwood never existed. She was entirely the creation of artificial intelligence, her online persona a carefully crafted digital illusion.
Tilly’s rapid ascent is more than just a quirky internet story. As reported on October 5, 2025, by The Economist, her viral popularity exposes a growing skepticism about the AI hype that has swept through the tech world and, increasingly, the broader economy. The Tilly Norwood phenomenon isn’t just a social media oddity—it’s a mirror reflecting the anxieties and uncertainties swirling around the trillion-dollar bets being placed on artificial intelligence.
This skepticism is far from unfounded. For almost as long as the AI boom has been in full swing, analysts and insiders have warned of a speculative bubble reminiscent of the dotcom craze of the late 1990s—a frenzy that ended in spectacular collapse and a wave of bankruptcies. Today, tech firms are pouring hundreds of billions of dollars into advanced chips and sprawling data centers. The goal? To keep up with the surging demand for AI tools like ChatGPT, Gemini, and Claude, and to position themselves for what many believe will be a fundamental shift in economic activity from humans to machines.
The numbers are staggering. In January 2025, OpenAI’s CEO Sam Altman unveiled a $500 billion AI infrastructure plan called Stargate at the White House, a price tag that stunned even seasoned industry watchers. Not to be outdone, Meta’s Mark Zuckerberg pledged to invest hundreds of billions more in data centers. In September, Nvidia announced an agreement to invest up to $100 billion in OpenAI’s data center buildout—a deal so large that some analysts wondered if Nvidia was propping up its own customers to keep demand for its expensive chips high.
OpenAI, meanwhile, expects to burn through $115 billion in cash through 2029, according to reporting by The Information. The company is exploring debt financing rather than relying solely on partners like Microsoft and Oracle, both of which have long-established, highly profitable businesses. Meta turned to lenders for $26 billion to finance a data center complex in Louisiana, while JPMorgan Chase and Mitsubishi UFJ Financial are leading a $22 billion loan for Vantage Data Centers’ massive new campus, as Bloomberg News has reported.
But as the spending spree accelerates, so too do the warning signs. Bain, a consulting firm, projected in September 2025 that by 2030, AI companies will need $2 trillion in combined annual revenue just to fund the computing power required to meet projected demand. Yet Bain predicts that actual revenue will likely fall $800 billion short of that mark. "The numbers that are being thrown around are so extreme that it’s really, really hard to understand them," David Einhorn, founder of Greenlight Capital, told The Economist. "I’m sure it’s not zero, but there’s a reasonable chance that a tremendous amount of capital destruction is going to come through this cycle."
It’s not just the financials that are drawing scrutiny. In August 2025, researchers at the Massachusetts Institute of Technology found that a whopping 95% of organizations saw zero return on their AI investments. This dismal finding was echoed by Harvard and Stanford researchers, who coined the term "workslop" to describe AI-generated content that looks impressive but fails to move the needle on productivity or meaningful progress. Their study concluded that the prevalence of such "workslop" could cost large organizations millions in lost productivity each year.
Despite these sobering assessments, AI’s biggest boosters remain undeterred. Sam Altman has publicly acknowledged the possibility of a bubble, but insists the technology’s long-term promise is real. "Are we in a phase where investors as a whole are overexcited about AI? In my opinion, yes," Altman said in August. "Is AI the most important thing to happen in a very long time? My opinion is also yes." Mark Zuckerberg echoed this sentiment in a podcast interview, noting, "If we end up misspending a couple of hundred billion dollars, I think that that is going to be very unfortunate, obviously. But what I’d say is I actually think the risk is higher on the other side." Zuckerberg’s bigger worry, he said, is not spending enough to seize the opportunity.
AI developers argue that the massive infrastructure buildout is necessary to support the rapid adoption of their services. OpenAI and Anthropic have released research suggesting that their AI systems are already making a meaningful impact on work tasks. In September, Anthropic reported that about three-quarters of companies were using its Claude AI model to automate work. OpenAI rolled out a new evaluation system, GDPval, to measure AI performance across dozens of occupations. "We found that today’s best frontier models are already approaching the quality of work produced by industry experts," OpenAI wrote in a blog post. "Especially on the subset of tasks where models are particularly strong, we expect that giving a task to a model before trying it with a human would save time and money."
Still, the question remains: how much are customers willing to pay for these services? OpenAI’s chief financial officer, Sarah Friar, captured the industry’s optimism when she remarked in late 2024, "If it’s helping me move about the world with literally a PhD-level assistant for anything that I’m doing, there are certainly cases where that would make all the sense in the world." OpenAI has reportedly discussed a $2,000 monthly subscription for its AI products, banking on the idea that improved models will persuade businesses and individuals to pay premium prices.
Yet the risk of a bubble is never far from investors’ minds. The AI boom has already experienced its own mini-crash: in January, China’s DeepSeek upended the market with a competitive AI model built at a fraction of the cost incurred by top U.S. developers. The announcement triggered a trillion-dollar selloff in technology shares, with Nvidia’s stock plunging 17% in a single day. But the setback was short-lived. By April, tech companies had doubled down on their AI spending, and Nvidia’s shares rebounded to fresh records, making it the world’s most valuable company by the end of September at over $4 trillion.
For those who remember the dotcom bubble, the parallels are hard to ignore. The late 1990s saw companies attract vast sums of investor capital, often using questionable metrics and unsustainable business models. When the bubble burst in 2001, many firms vanished or were absorbed by healthier rivals, and the market reset. Today, AI startups are being courted by venture capitalists wielding private jets and big checks, and some are completing multiple massive funding rounds in a single year. As Bret Taylor, OpenAI’s chairman and CEO of Sierra, put it, "I think there’s a lot of parallels to the internet bubble... I think we’re also in a bubble, and a lot of people will lose a lot of money."
Yet there are differences. The biggest AI players—like OpenAI, Meta, and Nvidia—are established giants with stable revenue streams and deep cash reserves. And AI adoption, despite the doubts, continues at a historic pace. ChatGPT boasts about 700 million weekly users, making it one of the fastest-growing consumer products ever. OpenAI expects its revenue to more than triple in 2025 to $12.7 billion, and its latest deal implied a valuation of $500 billion—despite never having turned a profit.
Ultimately, the story of Tilly Norwood and the trillion-dollar AI bubble is one of both promise and peril. As AI reshapes industries and captures the imagination of investors and the public alike, the question isn’t just whether the bubble will burst, but who will be left standing when it does.
