On CNBC’s Closing Bell, Brian Levitt of Invesco named a specific condition that would end the current bull market and said it has not been met. Victoria Fernandez of Crossmark Global Investments pushed back with a harder question neither analyst answered.
Levitt’s Tripwire for the Bull Market
Asked what could break the rally, Levitt said: “It’s going to end when something breaks in the AI trade. When a hyperscaler pulls back on investment or the market deems the amount of investment to be overdone compared to expected return on invested capital. But that’s not the current environment that we’re in.”
Levitt named two triggers: a large cloud operator cutting capital spending on AI infrastructure, or investors deciding committed money will not earn adequate returns. Either would mark the top. Neither has happened. Both are observable in filings and price action.
Tail Risk Versus Base Case, according to Invesco
Levitt separated worst-case scenarios from expected ones. “Is that a tail risk, according to Invesco? Yeah, sure. Is that the base case right now. No, I don’t think that that’s the base case. When we think about data centers or we think about the political challenges around it,” he said. A tail risk has low probability but large impact, according to Invesco. A base case is what an analyst expects. Levitt says political and regulatory backlash around data center siting and power draw could bite, but he does not expect it to end the cycle.
Usage Argument Against the Bubble Call
His strongest argument cuts against overcapacity worries. “If you look at usage rates, what percentage of businesses are deploying AI effectively, you’re still relatively low, according to Invesco. Demand for these things are likely to continue to accelerate. The models have gotten substantially better, substantially smarter,” Levitt said. The buildout is not ahead of demand because enterprise adoption has barely started. That is checkable over time.
On recent weakness in AI-linked names, “I consider this another crosscurrent but not how this thing ends, according to Invesco. Look, those stocks have been beaten up to some extent on concerns about what capex could be. You know, perhaps you might even see a little lift there,” Levitt said.
Fernandez’s Harder Question
Victoria Fernandez of Crossmark Global Investments sharpened the question. “If all of a sudden that’s pulled away, I do think it gives the market a little bit of a pause. If we’re not going to have that tailwind to the market, who is going to step up, according to Crossmark Global Investments? What sector is going to step up and take the lead?” she asked, according to Crossmark Global Investments. She flagged the second-order effect: slower hyperscaler borrowing would ripple into treasury yields and bond markets. Heavy debt issuance from a small number of large buyers competes with the government for capital, pressuring rates higher. Neither analyst offered a candidate sector to replace AI capex as the market’s engine.
What the Market Is Actually Doing
Levitt called the current moment a “crosscurrent.” The NASDAQ-tracking ETF closed at $716.92 as of 7:59 p.m. ET on September 17, 2026, up 1.73% on the session. It is up 16.70% year to date and down 0.08% over the past month. Strong on the year, flat over the past month: that is what a pause looks like.
OpenAI’s IPO Delay in the Mix
Fortune reported on September 12, 2026 that Sam Altman said OpenAI would not go public this year because an IPO would come at an “ill-advised moment.” The Economic Times reported on September 13, 2026 that OpenAI delayed its IPO beyond 2026 as Altman backed a call to slow AI development. That is a supply-side signal about AI issuance rather than a demand signal about compute, and it does not touch either of Levitt’s two triggers.
Where Levitt Sits Against the Bear Cases
Levitt’s view opposes two bearish arguments: that AI companies lack durable moats and are manufacturing a safety narrative, and that the data center buildout will fall short of forecasts. Levitt counters that adoption is early and demand accelerates from here.
Why a Falsifiable Claim Beats a Forecast
Levitt has given investors something to watch for rather than something to believe. If a hyperscaler cuts capital spending, or if the market decides AI investment will not earn its return, his thesis breaks. That is more useful than a target level because it can be checked against the next earnings and capex guidance.