AI capex boom widens profit margin gap to 100 points

The AI capex boom has split profit margins by 100 points across the value chain, with chipmakers earning 41 percent operating margins while model developers lose 59 percent, Apollo chief economist Torsten Sløk said.

"AI boom's profits are currently being funded by investors rather than earned from customers," Sløk said in a blog post published Friday. "The upstream margins are real, but they are paid for out of capital raised by the layer losing money, not out of cash generated by end demand."

Sløk categorized AI companies into four segments — models and applications, cloud and compute, energy and grid, and silicon and equipment — using data from PitchBook and Bloomberg for companies including OpenAI, Anthropic, Microsoft, Amazon, Constellation Energy, Nvidia, AMD, and Micron. The silicon and equipment category posted the highest operating margin at 41 percent, while models and applications recorded a negative 59 percent operating margin. Cloud and compute providers, which include Microsoft's Azure and Amazon Web Services, sit in between, while energy and grid operators such as Constellation Energy benefit from surging electricity demand from data centers.

Goldman Sachs projects AI investments to exceed $1 trillion in 2026, yet the technology has shown little measurable impact on economy-wide productivity or profit margin growth outside the Magnificent Seven. Should AI financing slow, the lopsided margin structure threatens to destabilize the entire industry, Sløk warned.

The divergence extends beyond the AI value chain itself. The AI capex boom is lifting profit margins for the Magnificent Seven — Microsoft, Amazon, Alphabet, Nvidia, Meta, Apple, and Tesla — while the broader S&P 500 has yet to see comparable margin expansion. This concentration of profitability in a handful of mega-cap technology names has reinforced concerns about index-level risk, with the S&P 500's earnings growth increasingly dependent on a narrow set of companies. The top seven stocks now account for a disproportionate share of index earnings, leaving the broader market vulnerable to any slowdown in AI-related spending.

The pattern echoes earlier technology investment cycles. During the late-1990s dot-com boom, infrastructure providers such as Cisco and Lucent captured outsized margins while application-layer companies struggled to monetize. The difference today is the scale: Goldman Sachs projects AI-related capital spending to exceed $1 trillion in 2026, a figure that dwarfs prior technology investment waves. The dot-com cycle ultimately ended when downstream monetization failed to materialize, and Sløk's analysis suggests a similar dynamic could unfold if AI applications do not generate sufficient revenue.

Can AI's End Customers Deliver Returns Fast Enough?

Sløk's central concern is whether the capital deployed upstream will eventually translate into downstream revenue. The models and applications layer — the segment closest to end customers — is currently the least profitable, with a negative 59 percent operating margin. For the value chain to remain sustainable, this segment must either grow revenue or continue raising capital.

"The bottom line is that the most profitable part of the AI value chain depends on the least profitable part continuing to grow revenue or raise capital," Sløk said. "Capital can bridge the gap for a while, but not indefinitely. And therein lies the risk: Will the ROI show up for AI's end customers fast enough to sustain the spending that is generating those upstream margins?"

For investors, the implications are twofold. First, the Magnificent Seven's margin outperformance may not be sustainable if downstream monetization fails to materialize. Second, the broader market's lack of margin expansion suggests the AI boom has yet to translate into economy-wide productivity gains, which could limit the durability of the current earnings cycle. If AI spending slows, the 41 percent operating margins at the silicon and equipment layer could compress rapidly, dragging down the very companies that have driven much of the S&P 500's gains over the past two years.

The stakes extend beyond equity markets. If AI capital spending fails to generate returns for end customers, the resulting slowdown could ripple through credit markets, where technology companies have issued significant debt to fund data center construction and chip purchases. A pullback in the projected $1 trillion of AI capex would also hit energy and grid operators that have invested heavily to meet data center electricity demand, creating a second-order impact on utilities and power infrastructure.

This article is for informational purposes only and does not constitute investment advice.