The industry is spending like crazy. Most people haven't noticed — or don't care.
Amazon, Alphabet, Microsoft, Meta and Oracle have collectively committed to spending up to $690 billion on AI infrastructure this year alone, nearly double what they shelled out in 2025. OpenAI recently told investors it's targeting around $600 billion in total compute by 2030. The theory: build it, and they will come.
They haven't come yet.
"Looking at what's possible, it does feel sort of surprisingly slow," OpenAI CEO Sam Altman told the New York Times. Nvidia CEO Jensen Huang was even more direct, telling the No Priors podcast that "the battle of narratives is being won by the critics."
The critics have receipts. According to Pew Research Center, half of Americans say they're more concerned than excited about AI — and that number keeps climbing. A CNET survey found only 44% of people feel confident they can spot AI-generated content online. For a lot of folks, AI isn't a tool they chose to adopt. It's something happening to them — in their social feeds, their search results, their kids' homework. And they're not thrilled about it.
The enterprise side isn't making up the difference. Jared Spataro, who leads Microsoft's AI at Work efforts, recently admitted he was struggling to show return on investment for Copilot "because a lot of knowledge work doesn't translate directly into top-line or bottom-line figures." A BCG report found that while companies are buying AI tools, proving those tools generate revenue remains the exception, not the rule.
That hasn't cooled the rhetoric. Microsoft AI chief Mustafa Suleyman told the Financial Times that "most tasks that involve 'sitting down at a computer' will be fully automated by AI within the next year or 18 months" — listing accounting, legal, marketing and project management. Promises like these might be doing more harm than good. When the pitch keeps inflating and results stay incremental, even would-be believers tune out.
The New York Times draws a comparison worth sitting with: Americans were far more enthusiastic about the internet during the dot-com era than they are about AI today. In the '90s, the value was tangible — email, shopping, information on demand. AI, for all its raw capability, hasn't had its "email moment" for most people. And the flood of AI-generated slop filling social feeds is actively eroding trust rather than building it.
Meanwhile, the spending only accelerates. AI firms including OpenAI and Anthropic are now among the largest corporate lobbyists in Washington. The shift to agentic AI — systems that don't just answer questions but take actions on your behalf — is demanding even more computing power than the chatbot era. "Tens of thousands of CPUs are now needed to process and manage the petabytes of data generated by the GPUs," Semianalysis analysts wrote, "a use case that wouldn't have otherwise been required without AI."
Valley View
The AI industry is building highways before people have bought cars. The bet is that transformative applications will emerge and make today's chatbots look like dial-up internet. Maybe. But unlike the dot-com boom, AI can't count on public enthusiasm to carry it through the messy middle — and every overwrought promise about automating your accountant within 18 months pushes skeptics further away. For an industry that just committed the better part of a trillion dollars to the cause, closing the gap between what's being spent and what people actually use isn't a marketing problem. It's an existential one.