The Bubble Isn’t-And Is

May 5, 2026

Artificial intelligence firms accounted for 61% of global venture capital (VC) investment in 2025, or $258.7B out of $427.1B, more than doubling AI’s share since 2022 (30%), according to OECD analysis. These numbers have reignited a familiar debate: is AI in a bubble?

In August 2025, Sam Altman said, “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes. He compared the moment to the dot-com bubble: “When bubbles happen, smart people get overexcited about a kernel of truth.”

MIT Sloan’s Thomas Davenport & Randy Bean don’t mince words in their 2026 AI predictions, comparing the present situation with the dot-com bubble, they say, “It’s hard not to see the similarities to today’s situation, including the sky- high valuations of startups, the emphasis on user growth over profits, the media hype, the expensive infrastructure buildout.”

The question isn’t whether AI will transform industries. It will. The question is whether today’s capital allocation reflects that transformation—or distorts it.

The case for a Bubble

A Boston Consulting study in late 2024 of 1,000 companies adopting AI found that only 4% achieved significant value creation, while another 22% advanced beyond the proof‑of‑concept stage to generate some value.4 Another report by the MIT Project NANDA claims that despite billions of dollars in enterprise investments in Gen AI, 95% of organizations have no measurable P&L impact. These numbers reveal a potential disconnect between capital deployment and value creation.

Circular financing compounds the concern. Circular financing in AI infrastructure deals occurs when major suppliers like NVIDIA and Microsoft invest in startups and data-centre operators, who in turn commit to long-term purchases of GPUs, servers, and cloud capacity. Supplier then recognize revenue from these contracts, boosting their valuations, while startups gain credibility and guaranteed access to infrastructure. Analysts estimate that AI suppliers, cloud platforms, and developers have committed nearly US$1T in interconnected deals. A prime example of such an interconnected deal is the one between NVIDIA and OpenAI. In September 2025, NVIDIA pledged up to $100B to help finance OpenAI’s massive data‑centre expansion. In exchange, OpenAI agreed to purchase and deploy millions of NVIDIA GPUs in those facilities.

Such deals are being flagged as circular arrangements as they can inflate demand signals, distort reported revenues and valuations, and hide real weaknesses.

The case against a bubble

The counter-thesis merits equal scrutiny.

“AI itself is not a bubble. That’s a crazy concept,” said Mary Callahan Erdoes, CEO of JPMorgan Asset & Wealth Management, in November 2025. “We are on the precipice of a major, major revolution in the way that companies operate.

AI companies are generating substantial revenue. OpenAI’s growth trajectory is also closely aligned with compute growth. OpenAI’s annual recurring revenue rose from $2B in 2023 to $6B in 2024, and reached $20B+ in 2025 — a curve that mirrors its compute scaling from 0.2 GW in 2023 to 0.6 GW in 2024 and approximately 1.9 GW in 2025.9 Anthropic grew its run-rate revenue from $100M+ in Jan 2024 to $14B in Feb 202610. NVIDIA recorded a revenue of $215.9B11 in FY 2026 (up 65% YoY) and a net income of $120B (up 65% YoY)

Tech Sector Quality Gap: 2000 vs 2026

Compare the scenario today with the dot-com era. Cisco traded at 200 times earnings in 2000 on speculations13, while NVIDIA trades at a 41.33 P/E ratio (TTM). The earnings quality gap is also noteworthy: in 2000, 74% of internet companies had negative cash flows, tech companies accounted for 33% of the S&P 500 market cap, but only 15% of its earnings. By contrast, a March 2026 analysis shows that the tech sector makes up over 43% of the S&P market cap and 35.9% of its earnings, a much tighter ratio. Despite high capital expenditures, Microsoft, Google, Amazon, and Meta collectively generated $209B in free cash flow in Q3 2025 (TTM).

JPMorgan’s 2026 outlook stated it plainly: “Despite AI bubble and valuation concerns, we see current elevated multiples correctly anticipating above-trend earnings growth, an AI capex boom, rising shareholder payouts, and easier fiscal policy.”

Conclusion

The evidence suggests caution, not panic. Private markets show signs of froth. “Median pre-money valuations for AI companies are reaching premiums of 39.3% above non-AI companies at Series B and 56.4% at Series C.”. But leading private companies like OpenAI and Anthropic are posting real revenue growth. Public markets, meanwhile, seem to be earnings justified. The “Magnificent 7” companies have reported higher (year-over-year) earnings growth than the other 493 companies in the S&P 500 over the past several quarters.

AI Valuation Premium Over Non-AI Companies

History offers guidance. When the dot-com bubble burst, Amazon’s stock collapsed. Jeff Bezos made “discipline, efficiency, and eliminating” his watchwords. In Q4 2001, Amazon posted its first profit.

The correction didn’t kill the internet. It killed the undisciplined. AI’s reckoning, if it comes, will follow the same pattern.

FinStrat Insights