Nasdaq In Freefall: Is the AI Trade Ending?

After the NASDAQ fell about 5% and the S&P 500 dropped 4%–5% on Thursday, Travis asks whether the AI trade is ending. Lou says AI is not dead, but the “land grab” period may be over: for two years, underinvesting was the main risk, while now investors are demanding evidence of return on investment. Emily attributes the volatility to two forces moving against each other: political pressure over data centers and the continued growth implied by hyperscalers’ capital spending. She expects no sustained AI-spending downturn until hyperscalers actually cut CapEx, something the next earnings cycle may clarify.

Lou notes that current AI suppliers, including NVIDIA, are recipients of spending that has already occurred, so their reported growth can continue even as investors sell stocks based on lower future expectations. Travis compares the situation with the telecom build-out during the dot-com era: CapEx continued through the December 2000 quarter, but the market had peaked in March 2000, and the NASDAQ 100 was down roughly 50% a year later. Emily agrees that large software companies can quickly turn spending on or off, eventually affecting hardware companies down the chain.

Lou says being on an airplane during Thursday’s selloff—“a small metal tube” with no practical access to trading—was nearly ideal because it prevented a reaction to the moment. He is nervous about the next six months but still sees durable AI applications, whether AI becomes an “imaginary friend on your shoulder” or a productivity tool. He says a bubble can contain lasting winners, citing 1999 as an example, and argues investors should stop assuming every AI-related company will work out. Emily sees both risk and opportunity, but says mispricing requires accepting intense volatility.

Travis points to Duolingo’s roughly 35% post-earnings jump as evidence that some supposedly doomed non-AI stocks may be recovering. The company’s daily active users grew strongly despite predictions that people would simply build language apps with ChatGPT. Lou cautions that Duolingo remains down substantially for the year and that its success still involves AI disruption: management prioritized user growth even at the expense of profits, which declined. He credits the company for acting before disruption arrived, while warning that surviving AI does not automatically make a company a winning investment.

Uber reported about 500 AI-agent food orders but only about four AI-agent rides; one user described the ride experience as poor and said they would use the app directly next time. Lou says that single quarter does not prove Uber is safe, because both Uber and Duolingo remain down for the year. Emily rejects using the small numbers as proof that agents will not disrupt Uber. She notes that AI assistants are already asking to integrate with Uber, which could become valuable partnership distribution even as it changes how customers access ride-hailing.

Emily says broad market narratives—such as “software is dead”—ignore differences between companies. She views Datadog as an example of a business that may benefit from AI: rather than replacing its function with an internal tool, companies may need Datadog to manage valuable data and control agents’ access to it. She contrasts that with collaboration software, which she believes could face greater pricing pressure from AI. Travis and Lou caution that investors should distinguish a company that survives the next quarter from one that can remain strong for five or ten years.

Tesla plans to end Cybertruck production and shift manufacturing capacity to its CyberCab, following the Model S and Model X out of the company’s central lineup. Lou says abandoning a product that did not sell well is sensible, but questions whether Tesla still wants to be primarily an auto company amid possible expansion into energy, AI, and self-driving. Emily notes the shift from earlier expectations of 250,000 Cybertrucks and says the Cybertruck never fit Tesla’s stated goal of producing low-cost, mass-market EVs. They also question why Tesla has not delivered a major product refresh.

Emily says Google Search is in serious trouble because more searches end with AI summaries, making them harder to monetize than the old model in which users clicked sponsored or organic links and Google was paid. She worries smaller advertisers will be priced out if AI answers contain fewer, more expensive ads, sending them toward Meta or YouTube. Lou says “search” is evolving and Alphabet may eventually monetize a different form of it, though he concedes that the old ten-blue-links model is threatened. He believes Alphabet can probably adapt; Emily says adaptation will not replicate the past twenty years’ economics.

Lou says the S&P 500 has never been truly diversified because it is market-cap weighted: as companies such as Meta and NVIDIA become larger, they occupy a larger share of the index. He sees this concentration as potentially useful for investors who want broad exposure to AI without selecting individual winners, but acknowledges that the largest companies are currently concentrated in the same AI contest. Emily, who is 32, says she has never owned an S&P 500 index fund and prefers total-market funds with small- and mid-cap exposure. Travis explains the difference between market-cap, price, and equal weighting.

Emily says Apple does not need to lead in AI because it controls the consumer portal that large-language-model providers want to reach. Her main concern is not AI leadership but losing consumers, and she sees no evidence of that yet. Lou suggests Apple could remain model-agnostic by acquiring or using services such as Perplexity rather than committing to one model. They joke about Apple’s poorly received flip-phone launch and note that Apple has previously arrived late to categories while still catching up. Travis admits that, after buying a new iPhone and Mac, he remains deeply tied to Apple’s ecosystem.

Lou says the Cybertruck’s failure should not be confused with a failure of EVs: electric powertrains are better in principle, though batteries and charging infrastructure still need to improve. Emily agrees but says EV adoption has been slower and steadier than expected, with few suitable three-row vehicles in the United States. Lou mentions the Rivian R1S and Volkswagen ID Buzz as disappointing options, while Emily notes that cheaper, accessible Chinese EVs are already being sold internationally. Lou estimates that EVs reaching more than 50% adoption is still decades away.

Emily is unsure whether OpenAI will win or lose. ChatGPT gives it strong retail recognition, but individual users are expensive in compute and relatively difficult to monetize; she says OpenAI needs more enterprise customers, where Anthropic has shown stronger demand. Lou thinks OpenAI’s technology and people give it a place in a potentially commoditized market, but he is not sure that makes it an attractive investment. They discuss reports that an Anthropic valuation expectation had fallen from $1.5 trillion toward $1 trillion and suggest tighter funding could force AI companies to improve efficiency rather than simply build more data centers.

Emily rejects the idea that Big Tech automatically wins because it has customers, cash-rich legacy businesses, and the ability to fund companies such as SpaceX and Anthropic. The dot-com era showed that companies could invest in the internet and still be disrupted, so hyperscalers must keep adapting even if an AI bubble bursts. Lou agrees that the largest companies may remain viable but warns that their future businesses may not resemble the past decade’s unusually strong growth and cash generation. Higher rates and more companies competing for capital could affect them across the economy.

Emily chooses Axon Enterprise after the stock sold off on reduced guidance and a large investment in a third-party drone maker. She says the guidance change mainly reflects federal procurement contracts delayed beyond the current and next quarters, while software growth is accelerating and management remains focused on 30%-plus long-term growth. Lou selects Zebra Technologies, which makes barcode inventory systems and warehouse-automation technology. He says the quarter was weak partly because memory is scarce and more expensive, but demand looks solid into 2027 and beyond; at roughly 15 times expected earnings, he sees a buying opportunity. Dan chooses Zebra for the show’s watchlist.