One AI Buildout, Three Laggards: UBS's Palantir, AT&T, and Spotify Picks Are Three Different Bets

UBS rolled out a list of its highest-conviction tech, media, and telecom calls in mid-September, and three of the most striking names shared a label that usually sends a stock to the bargain bin: "laggard." Palantir, AT&T, and Spotify have all trailed the market in 2026, yet the bank rates each a buy, with Palantir's target lifted to $250 and Spotify's held at $690. On the surface that reads as contrarian cheerleading for beaten-down names. Read the reasoning and it is actually one macro call wearing three different outfits — and knowing which outfit you're buying is the whole game.

The shared thesis is simple to state: the artificial-intelligence spending cycle is early, so the money is flowing now and the paydays come later. The numbers behind it are enormous. The five largest U.S. cloud and AI providers have collectively committed $660 billion to $690 billion of capital spending this year, roughly double 2025, and Goldman Sachs has global AI investment crossing $1 trillion in 2026. Yet on the demand side, UBS found only 8% of surveyed companies have deployed agentic AI at scale. That gap — infrastructure cost being paid today, enterprise revenue arriving tomorrow — is the clock these three stocks are all running against. It is the classic capital-cycle lead: the buildout is discounted first, the monetization has not shown up in results yet.

What makes the three interesting is that they are not three versions of the same bet. They sit at three different points on that single clock, and each one has something different that has to go right.

Palantir: paying for the monetization, priced like it already happened

Palantir is the purest "who gets paid when AI actually runs a business" trade. UBS calls it the best AI enabler in the market and argues it faces little competition in turning frontier models into useful enterprise software — marketing, supply chains, security, operational decisions. After attending the company's customer conference, the bank said demand is outstripping capacity and repeatedly flagged that adoption is still early. Its projected revenue growth is the eye-popper: UBS models 88% this year and the company describes its own growth as exceeding 90%, with a 63% three-year compound rate forecast.

But that strength is already in the price, and this is where the "laggard" label gets dangerous. Palantir trades well above the broad software market, at roughly 51 times projected 2027 free cash flow and a forward price-to-earnings multiple of about 75 to 80 times. UBS argues that premium is justified against faster-growing peers like Snowflake and CrowdStrike — a "discount" only relative to the most expensive stocks in the software sector. What dragged the shares down in 2026 is exactly the question UBS had to answer: investors worried that large language models would render data-software companies like Palantir unnecessary. The entire bull case rests on the opposite being true — that LLMs are too imprecise to replace what Palantir does — and on a torrid growth rate continuing long enough to make a 50-plus-times-cash-flow multiple look reasonable. The most upside, and the most that has to go right.

Spotify: the consumer side of the same curve

Spotify is a different animal, and it is the tell for which "AI cycle" UBS actually means. This is not an infrastructure-spend story at all. It is a subscription business that crossed 300 million premium users and posted a record 33% gross margin, with revenue growth in the mid-teens. The AI angle here is operational — better personalization and cheaper content economics pushing margins higher — rather than building data centers that show up on someone else's income statement. The stock has gone essentially nowhere in 2026, closing in the mid-$500s, comfortably below its high. For a compounder whose user base and margin are still climbing, the lag is a timing problem, not a broken thesis, and UBS's $690 target implies meaningful room back to that high.

The catch on Spotify is that the market already believes it. The shares came off a massive run before stalling, so the re-rating from "money-losing streamer" to "cash-generating subscription company" has largely happened. The bet now is that pricing power and margin expansion keep compounding for several more years at a pace the current multiple already anticipates. It is a steadier, lower-octane version of the Palantir trade — a growth story the market partly trusts rather than one it still doubts.

AT&T: the income floor the cycle hides under

AT&T is the odd one out, and it's worth taking seriously precisely because it does not fit the AI narrative at all. It is a slow-growth, dividend-paying telecom trading near the bottom of its 52-week range at about $25, with a roughly 4.3% yield, around 11 times earnings, and a UBS target of $31 set in late 2025. On the surface it is a defensive income holding that happens to be in UBS's tech basket. But look at why it lagged and it's the mirror image of the other two. Free cash flow declined meaningfully earlier this year and leverage is rising as the company pours money into a fiber buildout. The dividend is only as safe as the cash that funds it, and for now that cash is under pressure.

AT&T is the part of the call where the three forces that drive every long-term market — demographics, debt, and technology — collide most visibly. It is old-economy infrastructure financing a technology upgrade on borrowed money, promising the payoff later. In a world where a $1 trillion AI buildout is gorging on capital, a 4% yield funded by a thinning free-cash-flow stream is the trade that looks safest and may carry the most operational risk if the debt picture keeps deteriorating.

Why the shared label matters less than the location

Notice what UBS did not do: it did not tell you these three have identical risk. It told you they all sit on the same upward curve and got left behind as the market crowded into the obvious winners — the chip makers and cloud giants who already ran for the door when the spending first went parabolic. That is the real investment lesson, and it maps onto the current mood cleanly. Risk appetite right now reads greedy — the crypto fear-and-greed gauge sits near 71, deep in greed territory, and total crypto market cap is near $2.8 trillion. When everyone is this excited about one theme, the crowded trade is the front of the curve, and the laggards are where the disagreement lives. That is exactly the sort of setup where a macro investor checks the lead indicators instead of leaning on sentiment.

But greed is also a warning, not a confirmation. The honest framing of UBS's call is that you are choosing where along one capital cycle to get paid, and each name demands a different thing from the future: Palantir needs a doubling-down of enterprise AI adoption to justify a 50-times-cash-flow multiple; Spotify needs a proven growth story to keep compounding; AT&T needs fiber to convert debt into cash that keeps a dividend alive. They only look like the same trade because they share one banner. The reason to care about any of them is the same as the reason to care about the whole market right now — the money to build AI has been spent, and the question of who finally gets to collect on it is very much open.