The AI Bubble Pressure Is Real: CEOs Are Spending Billions They Can’t Justify

The confession came quietly, almost offhandedly. Lattice CEO Sarah Franklin told Futurism that company leaders are under enormous pressure to spend on AI — even when the returns aren’t there. “There is a massive bubble,” Franklin said. Not a hypothetical one. Not a maybe-someday concern. A bubble that’s inflating right now, fueled by boardroom anxiety and investor expectations that have decoupled from operational reality.

Franklin’s candor is striking because most executives won’t say it out loud. The pressure to integrate AI into every product, every workflow, every earnings call talking point has become so intense that CEOs are pouring capital into initiatives they privately doubt. It’s a familiar pattern for anyone who watched the dot-com era unfold, but the dollar amounts this time around are staggering. Microsoft, Google, Amazon, and Meta collectively committed over $200 billion in AI-related capital expenditures in 2024 and early 2025, according to their own earnings reports. And the smaller companies chasing them? They’re stretching even harder.

The core problem isn’t AI itself. The technology works. Large language models, computer vision systems, and predictive analytics tools are producing genuine value in specific applications. But there’s a widening gap between what AI can actually deliver today and what the market is pricing in. Franklin pointed to this disconnect directly — companies are buying AI tools and services not because they’ve identified a clear ROI, but because they’re terrified of being left behind.

Fear as a purchasing driver. That’s the bubble.

The spending spree has a shelf life, and the reckoning is already starting

Goldman Sachs published research in mid-2024 questioning whether the massive AI infrastructure buildout would generate sufficient returns, with head of global equity research Jim Covello arguing that AI technology is “enormously expensive” and that the costs may not justify the benefits for most use cases. That skepticism has only grown. Sequoia Capital’s David Cahn estimated in a widely circulated analysis that the AI industry would need to generate $600 billion in annual revenue just to cover infrastructure costs — a number that dwarfs current AI revenue figures by an almost absurd margin.

So where does that leave the average enterprise CEO? Stuck. Franklin described a dynamic where leaders feel compelled to announce AI strategies, hire AI teams, and acquire AI vendors regardless of whether those investments connect to measurable business outcomes. The pressure comes from investors, from boards, from competitors issuing press releases about their own AI transformations. It’s reflexive at this point. Almost automatic.

And the vendors know it. AI software companies are selling hard into this anxiety, often packaging existing capabilities under an AI label to command premium pricing. The Wall Street Journal has reported on “AI washing” — the practice of companies overstating their AI capabilities to attract investment. The SEC has even taken enforcement actions against firms for misleading claims about their use of artificial intelligence.

The talent market reflects the same distortion. AI engineer salaries have skyrocketed, with some senior roles commanding $800,000 or more annually at major tech firms, according to data from Levels.fyi. Companies are paying these premiums not always because they have defined work for these engineers but because not having an AI team feels existentially dangerous.

Franklin isn’t alone in her assessment, though she’s unusually blunt about it. Salesforce CEO Marc Benioff has made comments suggesting that many AI deployments are failing to meet expectations. Even Sam Altman, whose company OpenAI sits at the center of the current AI boom, has acknowledged that most AI startups will fail — a statement that reads differently when you consider how much venture capital has flooded into the space. CB Insights data shows AI startup funding hit record levels in 2024, with billions flowing into companies that have minimal revenue and unclear paths to profitability.

Here’s what makes this moment different from previous tech bubbles: the underlying technology genuinely is powerful. The internet was transformative even after the dot-com crash wiped out trillions in market value. AI will likely follow a similar trajectory — real long-term value, but a painful correction between the hype cycle and sustainable adoption. The companies that survive will be the ones spending on AI with discipline rather than panic.

That distinction matters. Discipline versus panic. Franklin is essentially warning her peers to know the difference.

The pressure won’t ease anytime soon. Earnings calls will keep featuring AI buzzwords. Boards will keep asking about AI strategy. But the smartest operators — the ones paying attention to what Franklin is saying — will start demanding proof before writing checks. Not pitch decks. Not demos. Proof. Measurable, repeatable business value tied to specific AI implementations.

The bubble doesn’t have to pop catastrophically. It can deflate. But only if enough leaders are honest about what they’re seeing — and what they’re not.