Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leonec ullamcorper mattis, pulvinar dapibus leo.
AI bubble bursting fears are growing, but Andreessen Horowitz’s September report shows tech earnings up about 56% against stock gains above 20%. The Business Perspective explains both sides of the AI bubble bursting debate, including why Treasury Secretary Bessent dismissed the concern on October 3.
⚡ Key Takeaways
AI bubble bursting talk has a new problem: the numbers. Per Andreessen Horowitz’s September report, tech earnings per share are up 55.7% this year while tech stocks have gained 22.1%. Treasury Secretary Scott Bessent, meanwhile, dismissed bubble fears on October 3. The Business Perspective lays out both sides, because the answer hinges on what happens to AI spending next.
Here’s the thing: a bubble is a story about prices outrunning profits, and right now profits are the ones sprinting. That doesn’t make the doubters wrong. It changes what they’re really betting on, and that’s where this guide starts.
📋 In this article
A bubble isn’t just a big rally. It’s a rally that earnings can’t justify, and that gap is what both camps are measuring. Same market, two very different readings of AI bubble bursting risk.
| Question | Bull reading | Bear reading |
|---|---|---|
| Prices vs profits | Tech EPS up 55.7% vs a 22.1% price gain (a16z) | Earnings forecasts can change fast if AI spending slows (Benzinga) |
| Valuations | Tech P/E multiple down 21.6% (a16z) | OpenAI and Anthropic took 43% of global AI venture capital in H1 2026 (PitchBook) |
| Spending | Hyperscaler spending pushed chip earnings to unprecedented levels (a16z) | A capex slowdown would hit Nvidia and suppliers first (WisdomTree’s Sam Rines) |
| Financing | Big players are investing and growing revenue (Bessent, per Bloomberg) | Money circulating between the same firms may inflate values (critics) |
Bessent made the case as a macro optimist. He said ripple effects from the Iran war are masking underlying US strength, pointing to strong consumer spending and wage growth in line with inflation, per Bloomberg.
He was just as dismissive of the doom camp. Bloomberg separately reported that he called alarmism without solutions from parts of the AI community “not leadership” and urged the industry to police itself.
That said, the same Bloomberg report lists surging AI spending among the forces pushing borrowing costs higher. So even the official who sees no bubble treats the AI boom as a market mover.
| Metric (tech sector, year to date) | Reading |
|---|---|
| Price gain | +22.1% |
| Earnings-per-share growth | +55.7% |
| P/E multiple change | −21.6% |
| Multiple vs five-year average | Roughly 20% below |
Source: a16z State of Markets report (Sept 2026), as covered by Benzinga on October 1.
Read that table slowly. Prices rose less than half as fast as earnings, so investors pay less for each dollar of tech profit than they used to. Benzinga adds that the report sees the rally broadening beyond the Magnificent Seven.
One caveat matters. Benzinga notes the data is a September snapshot, and earnings expectations can change quickly if AI spending slows. An AI bubble bursting test is only as good as next quarter’s earnings.
The money is already lopsided. Per PitchBook’s Q2 2026 AI report, AI startups raised $407 billion in the first half of 2026, more than all of 2025, with OpenAI and Anthropic absorbing 43% of it. We tracked the biggest rounds in our list of AI startups that raised $100M+ in 2026.
That concentration cuts both ways. Strong companies get funded at scale, but the risk of an AI bubble bursting is highest where capital is most concentrated, as our venture capital market share analysis shows.
Valuations running ahead of revenue are the clearest stress points. Cognition raised $2 billion at a $48 billion valuation with no publicly disclosed revenue, which we covered in Cognition’s $48B valuation trap. OpenAI, meanwhile, loses $1.60 for every $1 it earns, according to our OpenAI revenue analysis.
For a founder-side view of how VCs think about raising, Rise of Startups has a candid read: 10 Secrets VCs Won’t Tell You About Raising Funding. And for a primer on where AI shows up in real products, see IoT Insights Hub’s guide to AI + IoT technologies.
This is the link most bubble talk skips. The Fed’s Philip Jefferson said the economy is being buffeted by a cascade of shocks, including rising energy prices and the surging AI buildout, per CNBC. Bloomberg adds that the 10-year Treasury yield briefly hit its highest level since 2002 this week.
The Business Perspective covered the corporate side in our guide to why corporate bond yields crossed 6%. Cheap money fueled the AI buildout. Pricier money tests it, and projects financed with debt feel the pressure first.
Bessent’s counter is that the energy shock will fade and that the rise in yields reflects global trends, not a US-specific problem.
| Signal | Why it matters | Source |
|---|---|---|
| Hyperscaler capex guidance | It feeds the chip earnings behind the 56% | a16z via Benzinga |
| 10-year Treasury yield | Sets the cost of money for AI projects | Bloomberg |
| Corporate credit risk premiums | Show investor appetite for risk, at a six-month high | Bloomberg |
| Share of venture dollars going to AI | Measures how concentrated the bet is | Dealroom via TBP |
This looks more like a selective repricing risk than a market-wide AI bubble bursting, but we hold that view loosely. Earnings growth of 56% is a real cushion. It just isn’t a guarantee, because a cushion built on AI spending is only as stable as the spending.
Nobody can say for certain. Andreessen Horowitz’s September report shows tech earnings per share up 55.7% against a 22.1% price gain, which argues against a classic bubble. Skeptics counter that those earnings depend on heavy AI spending that could slow, so an AI bubble bursting looks more like a repricing than a collapse.
In an Axios interview published October 3, Treasury Secretary Scott Bessent dismissed concerns about an AI-driven bubble. Per Bloomberg, he argued that Microsoft, Google and Meta are investing heavily and generating significant revenue growth for Anthropic and OpenAI, and he said the energy shock pushing up yields will fade.
a16z’s September 2026 State of Markets report, as covered by Benzinga, shows tech stocks up 22.1% this year while earnings per share grew 55.7%. The sector’s P/E multiple fell 21.6%. a16z concludes that profits, not prices, are doing the work behind the rally.
Skeptics point to heavy AI spending, funding concentrated in a few companies and circular investment flows that may inflate valuations. Per PitchBook data in The Business Perspective’s coverage, OpenAI and Anthropic absorbed 43% of global AI venture capital in the first half of 2026. OpenAI also loses $1.60 for every $1 it earns, per The Business Perspective’s analysis.
It could shrink funding rounds, lower valuations and delay IPOs, because AI took 86% of US venture dollars, per Dealroom data tracked by The Business Perspective. Startups with real revenue and low burn would likely cope better than companies valued mainly on narrative. That is analysis, not a forecast.
Higher rates raise the cost of the money that funds AI spending and reduce what investors pay for future profits. Bloomberg reports surging AI spending is helping push borrowing costs higher, and the 10-year Treasury yield briefly hit its highest level since 2002 this week. Bessent says the rise matches global trends.
Here’s where the evidence leaves us. Earnings are growing about 2.5 times as fast as tech stock prices, and multiples sit below their five-year average. That’s not what a classic bubble looks like.
But the cushion rests on spending, and spending is exactly what skeptics doubt. If hyperscalers pull back, the chip earnings behind the 56% would feel it first, and AI-heavy venture funding next.
At The Business Perspective, we’ll keep tracking the AI bubble bursting question as capex guidance, yields and funding data come in. For now, treat the bull and bear cases as two readings of the same numbers, and watch the next earnings season before you pick one.
Source note: data comes from Bloomberg (Oct 3, 2026), Benzinga’s coverage of a16z’s State of Markets report (Oct 1), CNBC (Sept 30), SmartAsset, TradingKey, PitchBook and Dealroom data as reported by The Business Perspective. This article is analysis, not financial advice.
