HPE's 26% Pop: The AI Trade Is Spreading Past Nvidia

Hewlett Packard Enterprise (HPE) jumped roughly 26% in a single session after raising its full-year guidance — its strongest earnings surprise in years. The signal matters more than the move: the AI trade is leaking out of Nvidia (NVDA) and into the boring layers underneath it.

What Just Happened in Enterprise AI Hardware?

The picks-and-shovels layer finally showed up in the numbers. For two years, AI profits concentrated almost entirely in NVDA, whose latest quarter posted about $81.6 billion in revenue, up roughly 85% from a year earlier.

Now the spending is fanning out. HPE raised guidance and beat estimates on the back of AI-optimized server demand, and the stock responded with a ~26% single-day move.

When the most unglamorous box-maker in tech beats by its widest margin in years, the AI capital cycle has clearly moved past the chip and into the rack. That shift is what investors should be watching.

Why Does HPE's Beat Matter Beyond One Quarter?

Because it confirms a structural hand-off, not a one-off. Hyperscaler capital spending is enormous and still climbing: the five largest U.S. cloud operators are on track to deploy somewhere north of ~$650 billion on AI infrastructure in 2026, up from roughly $380 billion in 2025.

That money does not stop at the GPU. Every accelerator needs a server chassis, high-speed networking, power, and cooling around it.

So when HPE and Dell (DELL) report fatter AI order books, they are effectively reading the same demand curve as NVDA — just one layer out. DELL entered its new fiscal year with around $43 billion of AI server backlog after closing more than $64 billion in AI-optimized orders.

Who Wins as AI Spending Moves Down the Stack?

The networking and server layer wins next. Once GPU supply loosens, the bottleneck shifts to connecting tens of thousands of chips into a single fabric — and that is a networking problem.

Arista Networks (ANET) raised its 2026 AI networking revenue target to roughly $3.25 billion, while Broadcom (AVGO) supplies much of the custom silicon and switching that stitches GPU clusters together. Legacy player Cisco (CSCO) is fighting to keep its enterprise share as these fabrics get rebuilt.

Here is how the layers stack up in 2026:

The investable idea is not that these names replace Nvidia — it is that AI dollars now have to pass through all of them to reach a working data center. Each layer collects a toll.

What Should Investors Watch Next?

Watch backlog conversion, not backlog size. A ~$43 billion order book at DELL only matters if it ships at a margin that survives contact with hyperscaler bargaining power.

Three things separate signal from hype this year. First, gross margins on AI servers — thin margins mean revenue growth without much profit. Second, whether ANET and AVGO hold pricing as competition arrives. Third, whether enterprise (not just hyperscaler) demand shows up, which is where HPE and CSCO live.

If you want a framework for judging whether these beats translate into durable cash flow, our guide to fundamental analysis walks through the margin and cash-conversion checks that matter here.

What Could Break This Thesis?

The margin math could break it. Server assembly is historically a low-margin business, and bolting GPUs onto a chassis does not automatically change that.

Critics argue that companies like DELL and SMCI (SMCI) are booking enormous AI revenue at single-digit operating margins, meaning the "AI winner" headline hides a commodity economics problem. There is also concentration risk: a handful of hyperscalers fund most of this spending, so any pause in their capital plans hits the whole stack at once.

And backlog is a forward promise, not collected cash — order cancellations or delivery slips would expose how much of the optimism is already priced in. The safest read is that the AI hardware layer is real and growing, but it is a thinner-margin, more cyclical business than the chip layer that led the rally. For a broader view of how to position around themes like this, see our investment strategies primer.

How This Fits the Bigger 2026 Picture

The market is slowly repricing what "AI exposure" means. In 2024 and 2025 it meant owning the chip; in 2026 it increasingly means owning the toll roads around the chip.

That broadening is healthy for the rally because it spreads the earnings base beyond a single company. But it also raises the bar: investors now have to distinguish between high-margin enablers like AVGO and lower-margin assemblers like SMCI.

The earnings tape is doing that sorting in real time. HPE's ~26% pop was the loudest single data point, but the quieter beats across networking and custom silicon tell the same story.

Ready to analyze these stocks yourself? Search any ticker on MainRatios to see valuations from 6 legendary investors - free.