Everything About AI
NextFin News - Nancy Tengler, CEO and chief investment officer of Laffer Tengler Investments, is telling investors to treat the latest artificial-intelligence selloff as a chance to buy rather than a signal to flee. Her message, delivered in a television interview on Sunday, Sept. 14, 2026, lands as Nasdaq-100 futures pointed more than 1.5% lower and chip stocks extended losses worldwide after three of the largest AI companies' leaders called over the weekend for a slowdown in AI development. The question is whether she is right - and whether this selloff is the latest cyclical dip in a structural boom, or the first crack in a trillion-dollar buildout that has outrun its returns.
The Selloff Has Two Triggers, and That Changes the Stakes
Markets entered the week under pressure from two directions at once. The first trigger was commentary from AI industry leaders over the weekend urging a slower pace of development. Anthropic's chief executive opened the case for a coordinated slowdown, and the chief executives of OpenAI and SpaceXAI, along with Google DeepMind's Demis Hassabis, rallied behind the call. Investors read the convergence as a warning that demand for AI services - and for the memory chips, accelerators, and data-center gear that feed them - could weaken. Nasdaq-100 futures fell more than 1.5%, and semiconductor names that had led the rally for two years were among the hardest hit.
The second pressure is macroeconomic. Brent crude climbed above $108 a barrel after Saudi Arabia shut down its East-West pipeline, a key route for bypassing the Strait of Hormuz, following multiple drone attacks. A planned meeting between Gulf states and Iran was postponed, dimming hopes for diplomatic progress. Energy inflation of that size does not stay contained: it feeds back into bond yields, and the 10-year Treasury yield was already hovering near a three-year high at roughly 4.97% on Monday. With the Federal Reserve's policy decision due Wednesday and investors expecting a rate increase, the AI trade is being squeezed between a demand scare and a financing-cost scare.
That combination matters because it is exactly the environment in which Tengler's framework has been tested before. She has spent four decades on Wall Street, and her firm's public positioning through 2026 has been consistently constructive: Laffer Tengler added positions in Palantir, Nvidia, Tesla, CrowdStrike, Microsoft, ServiceNow, and GE Vernova during a year she described as one in which technology stocks have been beaten up. Her stance is not a reflexive dip-buy. In one appearance she cautioned against chasing AI dips indiscriminately, favoring integrated leaders such as Microsoft and ServiceNow and cybersecurity names such as CrowdStrike while warning on companies that miss key pivots, citing Adobe as an example.
The tension, then, is specific: is the AI complex in the early innings of a structural shift, as Tengler argues, or is this the moment the capital-spending bill comes due?
Why the Selloff Is Happening - and Why That Is Not a Broken Thesis
The first-order explanation is simple: prices were set for uninterrupted acceleration, and the weekend commentary introduced the possibility of a pause. When a trade is crowded and valuations embed years of compounding, even a hint of slower growth can produce a disproportionate move. That is the mechanical part.
But the transmission channel runs deeper. AI equities are priced on a chain: capex spending by hyperscalers leads to revenue for chip and equipment makers, which drives earnings growth, which supports multiple expansion. A slowdown signal attacks the first link. If hyperscalers trim data-center orders, the impact propagates to Nvidia, AMD, Broadcom, Micron, ASML, and the power-and-cooling complex in sequence. The reason the move felt heavy is that the chain is long and every link is held by momentum investors with similar risk thresholds.
Yet a slowdown in the rate of acceleration is not a decline in the level of spending. Tengler's argument, repeated across her 2026 appearances, is that the AI cycle is still in its early innings and that today's leaders are fundamentally different from the dot-com cohort. The comparison is not cosmetic. During the late-1990s bubble, companies with no revenue and no path to profit commanded market value on narrative alone. Today's AI beneficiaries - the Magnificent 7 and their suppliers - generate outsized free cash flow. A portfolio-strategy report from Goldman Sachs made the point explicitly in 2026: this is not a "hope-and-hype" cycle like the dot-com era, noting that the price-to-earnings-growth ratio for technology stocks sits near 1.6x versus 3.7x at the dot-com peak.
The hook is plain: a 30% drawdown in a company earning 70% gross margins and growing revenue is a different animal from a 30% drawdown in a company burning cash on a story.
The Capex Wall: The Strongest Argument Against the Bulls
The counter-thesis deserves full weight, because it is the one that keeps portfolio managers awake. AI-related capital expenditure is projected to reach roughly $800 billion in 2026, according to analysts at Goldman Sachs and Morgan Stanley, with Morgan Stanley lifting its 2027 outlook 17% to $1.12 trillion. At those levels, the spending must eventually convert into earnings somewhere in the chain - or it becomes a write-down cycle.
There is already evidence of a return gap. A 2025 study from MIT's NANDA initiative, which examined 300 public AI initiatives, interviewed 150 executives, and surveyed 350 employees, found that 95% of organizations reported zero measurable return on an estimated $30 billion to $40 billion of generative-AI investment. If enterprise monetization lags while the infrastructure bill keeps rising, the squeeze lands first on the equipment suppliers whose orders are most discretionary. This is not a fringe concern: it is the mainstream worry that produced the weekend calls for a slower pace.
Tengler's answer to this is Jevons Paradox - the 19th-century observation that gains in efficiency lower the cost of using a resource, which in turn raises total consumption. Applied to AI: cheaper, more efficient models do not shrink the chip market; they expand the number of use cases. "More begets more," she has said, arguing that chipmakers can lose market share and still grow at a rapid pace. She has pointed to margins around 70% as evidence that the economics of the bottleneck remain intact even as competition intensifies.
The debate, stripped of rhetoric, is a race between two curves: the capex curve rising toward $1 trillion and the monetization curve catching up. The bull case requires the second to follow the first with only a manageable lag. The bear case requires the lag to stretch long enough to force a capex cut.
Cyclical or Structural: The Call That Decides the Trade
This is where the investment decision actually lives. If the AI buildout is a cyclical boom, the current selloff is the start of a mean-reversion that will not fully retrace - and buying the dip is catching a falling knife. If it is structural, the selloff is a discount on a multi-year trend.
Three pieces of evidence point to structural, with a cyclical overlay. First, the demand driver is a change in the cost of intelligence, not a one-off product cycle. Semiconductors have always been cyclical - inventory builds, order pauses, correction. But the underlying demand for compute has shifted regimes the way demand for electricity did a century ago. That is why Tengler frames the sector as early innings rather than late cycle.
Second, the balance sheets on the buying side can sustain the spend longer than in prior bubbles. The hyperscalers funding the buildout are among the most cash-generative companies in history. A dot-com company raised equity to buy servers it could not monetize; today's buyers are funding data centers from operating cash flow and are simultaneously seeing AI features pull through into existing subscription revenue.
Third, the selloffs so far have behaved like cyclical interruptions within a structural uptrend. After the January 2025 DeepSeek-driven rout - when Nvidia's shares fell 17% in a single session, erasing a record $589 billion in market value, the largest one-day loss for any company in U.S. stock-market history - the group recovered as the narrative shifted back to demand. Broadcom fell more than 17% and Micron dropped almost 12% that day, and both retraced much of the loss as the panic faded. The June 2026 tech slide, which pushed Nvidia briefly behind Apple in market value, also proved temporary. Each episode was triggered by a narrative shock, not by an earnings collapse.
The cyclical overlay is real, however. Semiconductor demand will oscillate with inventory and order timing, and the group can give back 20% to 30% in a risk-off month without the structural thesis being damaged. Tengler's own behavior reflects this distinction: she buys on weakness but is selective about which names, favoring those with durable pricing power and recurring revenue over those dependent on a single product pivot.
The Second-Order Trade Most Investors Are Missing
The first-order reaction to a slowdown call is to sell chips. The second-order implication is that the beneficiaries rotate within the AI complex rather than exiting it.
If development slows at the frontier but deployment accelerates across enterprises, the relative winners shift from pure-play accelerator makers toward the companies that integrate AI into workflows and charge for it - software and cybersecurity names with pricing power. That is consistent with Tengler's preference for Microsoft, ServiceNow, and CrowdStrike, and with her firm's additions to Palantir and GE Vernova, the latter a bet on the power infrastructure that every data center requires regardless of which chip wins.
There is also a cross-asset implication. If AI capex moderates but does not collapse, the biggest casualty is not equities - it is the bond market's inflation premium. A slower buildout reduces the demand shock to electricity, copper, and construction, which in turn reduces the term premium that has pushed the 10-year yield toward 5%. The equity selloff and the bond selloff are being driven by the same story; a resolution in one helps the other.
And the third-order gap: the market has priced AI as a binary - either the boom continues uninterrupted or the bubble bursts. The most likely outcome is neither. Spending growth decelerates, monetization improves gradually, and the winners are picked by margins and cash conversion rather than by who grew fastest in the prior quarter. Investors positioned for the binary will be whipsawed in both directions.
What to Watch: A Falsifiable Forward Look
The practical reading of Tengler's position is not "buy everything that fell." It is that a structural trend is being repriced by a cyclical scare, and that selectivity matters more than timing. The beneficiaries in a slower-growth AI environment are the companies with pricing power, recurring revenue, and exposure to the infrastructure that deployment requires - software, cybersecurity, power, and the chip designers with the widest moats. The most exposed are the marginal suppliers whose orders depend on uninterrupted capex acceleration, and the companies betting on a single pivot without a proven revenue path.
The forward look splits by horizon. In the short term, sentiment and positioning dominate; a crowded trade can overshoot to the downside regardless of fundamentals. Over the medium term, the Federal Reserve's Wednesday decision and the incoming inflation data will set the discount rate for the whole complex. Over the long term, the question is whether AI monetization closes the gap with capex - and the evidence so far says it is behind, not that it is impossible.
Scenarios:
- Base case: capex growth decelerates but stays elevated; AI equities chop sideways to higher as earnings catch up. Trigger: hyperscaler guidance that trims growth but reaffirms absolute spending.
- Upside case: enterprise AI revenue accelerates faster than expected, validating the Jevons Paradox trade. Trigger: software and cybersecurity earnings showing AI-driven price and volume gains.
- Downside case: the monetization gap forces a capex cut in 2027. Trigger: two consecutive quarters of declining data-center capital expenditure from the largest hyperscalers.
The falsifying signal for the bull thesis is concrete: if the largest hyperscalers report two straight quarters of lower data-center capital expenditure while AI revenue growth stalls below 20% annually, the "early innings" call is wrong and the selloff is the start of a cyclical down-leg, not a buying opportunity.
You want to take advantage of these periods of confusion or melancholy around the stock's future performance and add to your holdings if you own it and initiate if you don't.
Tengler has long urged investors to use periods of market melancholy as entry points, a stance she has applied to Nvidia and other AI leaders through multiple selloffs. Her closing argument, distilled from four decades of navigating technology cycles, is that the market pays a premium for conviction precisely when the narrative turns dark. The AI trade is not dying; it is growing up. And growth that is maturing is priced differently from growth that is stalling - the difference is who can tell them apart before the earnings prove it.
Explore more exclusive insights at nextfin.ai.
