By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panel is divided on HPE's recent performance, with some seeing genuine AI demand and others suspecting a backlog-fueled surge. The EPS beat may not be sustainable, and margin compression is a key risk.
Risk: Margin compression due to component inflation or backlog conversion at lower ASPs
Opportunity: Durable, multi-year demand for on-premises AI inferencing with stable ASPs
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Hewlett Packard Enterprise (HPE) was poised to open more than 23% higher on Tuesday after the server and networking products maker reported record second quarter earnings, fueled by a boom in enterprise investments in AI infrastructure.
The company also raised its full-year outlook while accelerating its long-term financial goals by two years, forecasting that demand for its servers will remain strong well into 2027.
“Traditional server orders increased triple digits as customers continue to modernize their compute infrastructure and invest in AI inferencing,” CEO Antonio Neri said during the company’s earnings call on Monday afternoon.
“Orders more than doubled, significantly outpacing revenue, resulting in a record company backlog,” he added.
Unlike COVID, when customers were double booking, Neri said, “We don't see that at all. We have no cancellations.”
While training artificial intelligence models requires expensive graphics processing units (GPUs), AI inference, which performs tasks, can be done on standard central processing unit (CPU) servers.
Management explained that corporate customers prefer CPU servers because they can run AI locally and securely on-premises. The trend is boosting HPE sales and driving higher profit margins.
Revenue in the quarter reached $10.7 billion, up 40%. Adjusted earnings per share of $0.79 increased 108%, blowing past the company’s outlook.
HPE shares jumped nearly 25% over the past two sessions to hit all-time highs, driven by peer Dell’s blowout results last week,
“Hewlett Packard (HPE) — a stock that has never featured on our retail leaderboard — finished as the second most-bought stock today,” a Vanda Research note stated on Monday after the bell.
“In fact, retail investors have bought as much HPE over the last 2 sessions as they did in the prior 11 months combined,” the note read.
HPE has surged by more than 90% year to date.
Ines Ferre is a senior business reporter for Yahoo Finance. Follow her on X at @ines_ferre.
Click here for the latest technology news that will impact the stock market
Read the latest financial and business news from Yahoo Finance
Four leading AI models discuss this article
"HPE's record backlog is likely a demand-pull phenomenon that will compress margins and growth rates as it converts to revenue, and retail FOMO into all-time highs is a classic exhaustion signal."
HPE's 40% revenue growth and 108% EPS beat are genuine, but the article conflates two separate tailwinds: AI infrastructure spending (real, cyclical) and a backlog surge that may reflect pull-forward demand rather than sustainable growth. The 'no cancellations' comment is reassuring but also suspicious—it suggests customers locked in orders at peak pricing before potential margin compression. Retail FOMO (buying as much HPE in 2 days as 11 months prior) is a classic late-cycle signal. The real risk: if CPU-based inference commoditizes faster than expected, or if customers realize they overbought, backlog converts to revenue at lower margins in 2026–27.
If HPE's backlog truly reflects multi-year demand visibility through 2027 at current margins, and if on-premises AI inference becomes the dominant architecture (vs. cloud GPU training), then the stock's 90% YTD gain and forward guidance acceleration are justified—not a bubble.
"Retail-driven momentum after a 90% YTD run leaves HPE exposed to any AI capex pause or competitive share loss."
HPE's 40% revenue growth and record backlog reflect real enterprise demand for CPU-based AI inference on-premises, with orders more than doubling and no reported cancellations. This supports the accelerated 2027 targets and explains the EPS surge to $0.79. Yet the 90% YTD advance, 25% two-day pop, and retail investors buying more in two days than the prior 11 months combined signal froth. Much of the AI server upside may already be priced in, leaving little margin if Dell or hyperscalers capture share or if inference spend normalizes post-2025.
Sustained on-prem inference demand could prove larger and stickier than expected, allowing HPE to compound at elevated margins and justify further multiple expansion even from current levels.
"HPE is successfully pivoting its legacy server business into a high-margin AI inferencing play, justifying a permanent expansion of its forward P/E multiple."
HPE is undergoing a valuation re-rating, moving from a legacy hardware laggard to an AI infrastructure play. The 108% EPS growth and record backlog suggest the market has structurally underestimated the demand for on-premises AI inferencing. While the 90% YTD rally is steep, the shift toward higher-margin, specialized server configurations provides a fundamental tailwind. However, the retail-driven buying frenzy mentioned by Vanda Research is a classic contrarian signal of short-term exhaustion. I am bullish on the infrastructure trend, but cautious on the entry point given the parabolic move. Investors should watch for margin compression if supply chain costs for high-end components spike in Q3.
The massive backlog may be a symptom of supply chain bottlenecks rather than purely organic demand, creating a 'phantom' growth narrative that will evaporate once lead times normalize.
"HPE has a real AI-infrastructure tailwind, but sustained upside hinges on durable enterprise capex and the profitability of CPU-based AI inference."
HPE's Q2 beat and record backlog signal a genuine AI-infrastructure cycle with on-prem CPU-based inference appealing to enterprises' security and control needs. Revenue rose 40% to $10.7B and adjusted EPS $0.79, up 108%, justifying a higher multiple as management accelerates long-term goals by two years. The thesis rests on AI demand broadening beyond GPUs and translating into durable server demand and margin uplift from standard CPU nodes. Yet risks exist: the surge could be backlog-fueled or cyclical, 2027 visibility is optimistic, and competition or a pivot to services could compress margins. A durable upside requires sustained capex beyond a few quarters.
The rally could prove temporary if backlog is pulled forward demand or if AI spend slows; CPU-inference on-premises may prove less scalable/cost-effective than anticipated, pushing enterprises toward GPUs or hybrid models.
"The backlog's margin profile—not just its size—determines whether HPE's valuation re-rating sticks or reverses in 2026."
Nobody's addressed the margin math directly. If HPE's backlog is real multi-year demand at *current* pricing, gross margins should expand—yet the article doesn't specify. Gemini flags supply-cost risk, but the inverse matters more: if HPE locked in component costs before inflation, they're actually margin-protected. That's bullish. But if backlog pricing already assumes margin compression (customers negotiating volume discounts), the 108% EPS beat is partly accounting fiction, not earnings power.
"Locked costs likely fail to cover the full backlog horizon, risking ASP compression on 2026-27 conversions."
Claude assumes locked component costs will protect margins through the backlog, yet this ignores how HPE's AI server configs rely on volatile high-end parts whose pricing HPE likely cannot fully hedge beyond 2025. If inference demand normalizes faster than the 2027 targets imply, those locked orders convert at compressed ASPs, turning the EPS beat into a one-time pull-forward rather than structural expansion.
"HPE's proprietary power and cooling technology creates a margin buffer that mitigates the risk of volatile component pricing."
Grok, your focus on volatile component costs misses the moat: HPE’s proprietary cooling and power management systems. These aren't just commodity servers; they are specialized infrastructure. If HPE is selling 'AI-ready' stacks, they aren't just selling hardware—they're selling energy efficiency, which is the primary bottleneck for on-prem data centers. Margin compression is less likely if the value-add is power density, not just raw compute. The real risk is not component inflation, but customer churn to hyperscalers.
"HPE's energy/density moat and AI-ready stacks can sustain margins even if backlog contains pull-forward demand."
Grok’s emphasis on backlog as a potential pull-forward risk is reasonable, but it misses a real margin moat: HPE’s energy/density advantage and integrated AI-ready stacks. If customers prize power efficiency and space savings as much as compute, pricing power could hold even as demand normalizes, limiting margin compression. Backlog quality matters—if these are durable, multi-year commitments with stable ASPs, the EPS beat isn’t just a one-off pull-forward but could be sustainable.
The panel is divided on HPE's recent performance, with some seeing genuine AI demand and others suspecting a backlog-fueled surge. The EPS beat may not be sustainable, and margin compression is a key risk.
Durable, multi-year demand for on-premises AI inferencing with stable ASPs
Margin compression due to component inflation or backlog conversion at lower ASPs