Goldman Flags "Flight to Quality" in AI: DigitalOcean Emerges as Profit-Linked Alpha Play

The market has long priced in a massive, sustained wave of AI capital expenditure. The scale is staggering: analysts project a $539 billion capex wave this year, set to grow to $629 billion in 2027. This spending has been a primary engine for tech stocks, with the sheer size of the investment creating a powerful narrative of growth. Yet, the setup is shifting. The growth rate of this spending is expected to slow down, creating a new divergence between the initial hype and the emerging financial reality.

For years, consensus estimates have consistently underestimated actual AI spending. The pattern is clear: analyst estimates have consistently underestimated capex spending related to AI. This persistent gap meant the market was often surprised by the strength of the investment cycle, leading to upward revisions in forecasts. The payoff has been tangible for the broader economy. Technology companies' massive spending on artificial intelligence accounted for half or more of U.S. growth last year, effectively propping up an otherwise anemic economy. In other words, the AI capex wave was not just a corporate story-it was a material driver of national economic expansion.

This creates the core expectation gap. The market has priced in a multi-year boom in spending, but the slowdown in the growth rate of that spending now introduces a new question: is the economic payoff keeping pace? The recent divergence in stock performance among AI leaders shows investors are already being selective. They are rotating away from infrastructure companies where capex is debt-funded and operating earnings growth is under pressure, while rewarding those with a clearer link between spending and revenue. The initial wave of optimism is being replaced by a more granular assessment of which companies can actually convert this massive investment into profits.

The Flight to Quality: Separating Winners from the Noise

The market's initial euphoria around AI is giving way to a sharper, more selective game. Investors are moving from a broad "AI will change everything" narrative to demanding visible earnings quality. This is a classic "flight to quality," where capital rotates away from speculative infrastructure and toward companies with a clearer path to profits. The shift is already visible in the data: since June, the average stock price correlation across the large public AI hyperscalers has declined from 80% to just 20%. In other words, the herd mentality has broken down, and investors are now picking winners based on fundamentals, not just hype.

This rotation is a direct response to the financial reality of the capex cycle. The market is rotating away from AI infrastructure companies where operating earnings growth is under pressure and spending is debt-funded. These are the companies that have led the charge in building data centers, but their operating margins are getting squeezed. The expectation gap here is clear: the market had priced in growth from massive spending, but now it's questioning whether that spending is translating into healthy profits. The focus is shifting to the next beneficiaries-the platform and productivity companies that can more directly monetize the AI wave.

DigitalOcean exemplifies this new quality play. It is not a household-name AI giant, but it offers reliable access to AI-capable infrastructure. The company's model provides a tangible, scalable service that customers pay for, creating a more predictable revenue stream. This is the kind of "quality" the market now seeks: a clear link between spending and revenue. DigitalOcean's own performance shows accelerating growth, fitting the profile of a beneficiary that can ride the AI wave without the same level of financial strain as pure infrastructure builders. In this new phase, the market's patience is reserved for those who can demonstrate they are not just spending, but profiting.

The Earnings Payoff: When Will It Materialize?

The market is waiting for the payoff from all that AI spending, but the timeline is lagging behind the hype. Goldman Sachs strategist Ben Snider projects that AI-related productivity gains will contribute about 0.4% to S&P 500 earnings growth next year, rising to 1.5% in 2027. That's a meaningful but still modest boost to the broader index's expected earnings growth. The key risk is a guidance reset if this payoff materializes slower than priced in. The market has already priced in a powerful earnings acceleration, so any stumble in the AI contribution could force downward revisions.

Progress is also uneven, potentially concentrating earnings momentum at the top of the index. Snider notes that larger companies are reporting more visible progress than smaller firms. This dynamic could keep the bulk of profit growth in the hands of the largest S&P 500 constituents, which are already set to account for around 46% of the index's profit growth in 2026. For investors, this means the earnings story is becoming less about broad-based AI adoption and more about the performance of a select few giants. The expectation gap here is between the market's broad optimism and the reality of a concentrated, selective payoff.

The most critical risk, however, is a "diminishing probability" that all current market leaders generate enough long-term profits to justify today's valuations. Goldman's own warning highlights this tension: the magnitudes of current spending and market caps alongside increasing competition suggest a diminishing probability that all of today's market leaders generate enough long-term profits to sufficiently reward today's investors. This is the core expectation arbitrage. The market has priced in a near-perfect outcome where every AI giant succeeds. The reality of scaling costs, competitive pressures, and the lag to earnings means that probability is falling. The upcoming earnings season will be a key test of whether the payoff is closing the gap-or widening it.

Catalysts and Risks: The Path to a "Capital Discipline" Phase

The maturing AI supercycle now hinges on a few near-term signals that will confirm or challenge the thesis of a slowdown. The first and most immediate test is the upcoming earnings season. Investors need to see if the guidance from giants like Microsoft, Meta, and Alphabet aligns with the expected deceleration in capex growth. The market has priced in a multi-year boom, but if these leaders signal they are pulling back from the projected $539 billion capex wave this year, it would validate the "priced in" narrative and force a reset. Conversely, if guidance stays aggressive, it could prolong the infrastructure spending phase and delay the shift to profitability.

The primary physical risk to this entire cycle is a gigawatt ceiling on power supply. As data centers multiply, the availability of reliable, affordable electricity is becoming a hard constraint. This is a classic capital allocation problem: companies will be forced to prioritize spending on the highest-return AI activities, likely accelerating the shift away from pure infrastructure build-out. This power bottleneck acts as a natural brake, ensuring that not all planned spending can be executed, which could help temper the financial strain on debt-funded projects.

The key catalyst for a true shift in capital allocation will be the visible monetization of AI beyond the data center. The market is waiting for evidence that spending is moving from building capacity to generating measurable productivity gains and platform revenue. This is where the "flight to quality" becomes a tangible investment theme. The recent divergence in stock performance, where correlations among AI hyperscalers have declined from 80% to just 20%, shows investors are already rotating toward companies with a clearer link between spending and revenue. The path to a "capital discipline" phase is paved by this selective rotation, where capital flows to the platform and productivity beneficiaries that can demonstrate a return on the massive investments already made.