HPE Surges Nearly 20%: The Hidden AI Infrastructure Story the Market Finally Priced In

The AI trade of the past two years has always come down to the same move: repricing the picks-and-shovels names. From Nvidia to Broadcom to SMCI, every few months a new name gets dragged into the spotlight. This time it's HPE's turn — surprising in some ways, inevitable in others.

Surprising, because for a long time HPE was filed away under "legacy IT" — stuck at single-to-low-double-digit P/E multiples, treated as a shadow play on Dell. Inevitable, because as long as GPUs stay constrained, whoever is racking them up and selling them to enterprises is going to capture real value. On June 2, the market put that thesis on the table with a single 20% candle.

What Drove the Beat

HPE reported fiscal Q2 2026 results (note: HPE's fiscal year runs from November, so this doesn't align with the calendar quarter). The numbers landed hard — revenue of $10.68B against a $9.79B consensus, up 40% YoY; non-GAAP EPS of $0.79 versus a $0.53 estimate. A clean, broad beat.

The segment breakdown is where it gets interesting. The Cloud & AI Systems segment posted revenue of $7.71B, up 22.9% YoY, with server revenue of $5.5B, up 32.7% — direct evidence that AI servers are shipping in real volume. Networking came in at $2.69B, up 148% YoY — a figure that looks shocking until you add the crucial context: the $14B Juniper Networks acquisition, which closed March 5, began consolidating this quarter, and the networking segment's explosion is almost entirely structural rather than organic. The data center networking sub-segment alone was up 233% YoY.

What really forced sell-side models to restart was the guidance. Full-year revenue growth was raised from 17–22% to 29–33% (Juniper consolidation accounts for a substantial portion). Non-GAAP EPS guidance was lifted from $2.30–$2.50 to $3.35–$3.45, and full-year free cash flow guidance came in at no less than $3.5B. On the AI side, management was explicit on the call: Q2 AI systems orders of $2.1B, AI Systems backlog hitting a record $5.9B. It was the first time the market heard HPE deliver something close to Nvidia's "demand is insatiable" narrative.

The market reaction was swift and violent. HPE opened high and kept climbing, closing up 19.47% — roughly $47 to $56 — its largest single-day gain since going public. Intraday it briefly touched +25%; after-hours buying pushed it toward +30% at one point. On a day when the S&P 500 itself was hitting new highs, a 19% single-name jump is rare enough — doubly so for a legacy server company with a $75–80B market cap.

Why the Market Is Re-Rating HPE

Understanding this move requires understanding HPE's prior valuation trap. For years the market wrote HPE off as a legacy IT vendor — servers, networking, storage, plus the Cray HPC (high-performance computing) business acquired earlier. Sprawling, slow-growing, thin-margined. Like Dell, it got priced as a "cycle + depreciation + slow dividend" story, multiple perpetually compressed.

This quarter changed the valuation anchor, on three distinct pieces of logic.

First, HPE is a critical channel for Nvidia GPUs reaching enterprise customers. Many enterprises can't get direct H100 or B200 allocation and have to source through system integrators. HPE slots GPUs into its own reference architecture rack solutions and sells the full package — essentially capturing the overflow demand from Nvidia's supply-constrained ecosystem. The $2.1B in Q2 AI orders and $5.9B backlog are the numerical proof of that thesis.

Second, GreenLake adds a private-cloud premium for the AI era. The biggest friction in enterprise AI deployment isn't compute — it's putting models, data, and compliance controls into a manageable environment. Public cloud raises trust and data sovereignty concerns; pure on-prem is too heavy a lift for most. GreenLake's "hardware in your data center, pay per use" model lands squarely in that middle ground. With GreenLake ARR growing at a high-double-digit YoY clip and enterprises signing longer-term contracts, this is the part of the business that deserves multiple expansion — and that the market has consistently underpriced.

Third, Juniper plugged the networking gap. The $14B deal closed March 5, bringing Juniper's data center networking assets, Mist AI (an AI-driven network operations platform), and SD-WAN into the HPE fold. That completes the server + networking + private cloud trifecta. Layer in the legacy HPC business and the AI server ramp, and HPE can now tell a credible "legacy giant transitioning to full-stack AI infrastructure" story — the kind of story that commands a very different multiple than "commodity server vendor."

Dell, SMCI, Cisco — Will They Get Re-Rated Too?

The most direct chain reaction from HPE's surge is a recalibration of peer valuations.

Dell (DELL) is the closest comp — PowerEdge AI server orders are building similarly, but Dell's most recent outlook read as conservative, and the market has stayed on the fence over AI server margin sustainability. Now that HPE has made the case that "AI servers are a real, accelerating business," Dell's next earnings will go under the microscope. The same sell-side models are now under pressure to revise up.

Super Micro (SMCI) is a more complicated picture. SMCI was once the purest AI server play in the market, but after peaking in 2024 (pre-split price around $1,200; a 10-for-1 split was completed October 1, 2024), it's seen a steep drawdown compounded by audit disputes and a credibility overhang that hasn't fully cleared. HPE's quarter effectively serves as an industry-wide endorsement of "AI server demand is still accelerating" — if SMCI can deliver a credible forward guide in its next report, the valuation recovery runway is substantial.

Cisco (CSCO) plays a different angle. Its UCS server line has long been overshadowed by HPE and Dell, but Cisco holds the Splunk data assets and a complete networking stack — its AI-era pitch leans more toward full-stack integration. The problem is that HPE, via Juniper, has now assembled a remarkably similar stack. Whatever "full-stack" premium Cisco commanded is starting to get diluted. Whether the market re-rates Cisco as the next AI infrastructure play hinges on what AI order data looks like in its next earnings disclosure.

Worth flagging the macro backdrop: on June 2, the S&P 500 closed at 7,609.78, a new high. Marvell jumped 32% post-earnings. Broadcom was up roughly 3–5% ahead of its own report. AI compute chain sentiment had just flipped from two weeks of hesitation back into full-on euphoria — HPE's surge is an extension of that broader move, not an isolated event.

Signals to Watch

Near-term, four data points will determine whether this re-rating has real legs or fades as a one-quarter pop.

First, Dell's next earnings: AI server revenue and backlog. The peer comparison will either validate HPE's narrative or deflate it. Second, whether HPE can quantify Juniper synergies next quarter — cost saves, cross-sell order flow — because the market needs to separate how much of that +148% networking growth is acquisition-driven versus organic. Third, GreenLake ARR growth: if it holds in the high-double-digit range, the multiple holds; if it decelerates, the re-rating thesis gets wobbly. Fourth, Nvidia's next earnings call: how often — and how specifically — Jensen Huang calls out HPE as a channel partner. A mention from Nvidia is another layer of validation the market will latch onto.

Longer term, the real question is whether HPE can carve out a position in AI inference. Training workloads have largely been locked up by hyperscalers. Inference — running models at scale inside enterprise private clouds — is precisely where GreenLake should have a structural edge. The pace at which that demand materializes will determine whether HPE is a one-cycle re-rating story or a multi-year secular rerate.

The near-20% jump has pulled attention back to a name the market had written off. The question now is whether that thesis can sustain a run — or whether June 2 turns out to be a one-time valuation correction.

This content is for informational purposes only and does not constitute investment advice, trading advice, or any guarantee of returns.