HPE and NetApp Post Record Quarters as Investors Question AI Returns

HPE and NetApp delivered record quarters on September 2, yet both stocks fell after hours despite stronger revenue, earnings, and full-year guidance. The Google News headline captured a revealing conflict. AI infrastructure demand remains intense, but investors now expect vendors to convert that demand into cash without sacrificing margins.

HPE reported record quarterly revenue of $12.2 billion, up 34% from the previous year. NetApp reported record revenue of $2.03 billion, a 30% increase. Both companies exceeded Wall Street expectations, while their management teams presented AI as a durable growth engine rather than a temporary spending surge.

The selloffs showed that beating estimates is no longer enough. Dell has established a much larger benchmark in AI systems, while cloud operators keep pushing suppliers on delivery schedules, pricing, and efficiency. For HPE and NetApp, the question has shifted from whether AI demand exists to whether each company can capture enough profitable value from it.

What the Google News Headline Left Unsaid

These were not weak reports punished by nervous traders. They were strong reports measured against expectations that had risen even faster.

HPE’s fiscal third quarter ended July 31, 2026. The company reported adjusted earnings of $1.11 per share, compared with 44 cents one year earlier. GAAP earnings reached $1.06 per share, up from 21 cents.

Revenue rose to $12.2 billion, while adjusted operating margin expanded by 770 basis points to 16.2%. A basis point equals one-hundredth of a percentage point. Operating cash flow reached $1.6 billion, and free cash flow reached $1 billion.

The company’s quarterly results described records across revenue, orders, and profitability. HPE also raised its financial outlook and said it planned to return at least 75% of fourth-quarter free cash flow to shareholders.

The Cloud and AI segment produced $9 billion in revenue, up 25% from the previous year. Server revenue accounted for $6.8 billion and grew 35%. Storage contributed another $1.3 billion, rising 10%.

HPE AI demand was even more visible in its order book. The company booked $3.1 billion in AI orders during the quarter. AI systems represented $2.4 billion, while networking products connected to AI deployments contributed about $700 million.

That left HPE with a record AI backlog of $7.6 billion. Backlog means contracted or booked demand that has not yet become recognized revenue. AI systems represented $6.8 billion of the total.

NetApp produced a similarly clear earnings beat. Its fiscal first quarter of 2027 also ended July 31. Revenue reached $2.03 billion, while adjusted earnings rose to $2.58 per share from $1.55 one year earlier.

The storage company’s earnings release reported $2.06 billion in billings, up 36%. Adjusted operating margin reached 31.9%, and GAAP net income rose 61% to $375 million.

NetApp AI storage benefited from two record product results. All-flash array revenue reached $1.3 billion, up 47%, while public cloud revenue increased 28% to $206 million. All-flash arrays store information on flash memory instead of mechanical disks, improving latency for data-intensive workloads.

NetApp also raised its fiscal 2027 outlook. It now expects annual revenue between $7.98 billion and $8.23 billion, with adjusted earnings between $9.73 and $10.03 per share.

Those results explain why the Google News summary sounded contradictory. HPE and NetApp delivered what investors usually request: growth, earnings beats, higher guidance, and expanding AI exposure. The market still found reasons to sell both.

The contrast was sharpest because the weakness came immediately after the announcements. HPE shares slipped in extended trading, while NetApp fell more than 8%, according to the initial market reaction.

That response did not erase the companies’ operating progress. It showed that investors were looking beyond the headline records toward backlog conversion, cash generation, and the quality of future growth.

AI Infrastructure Demand Is Moving Beyond Training

The quarter suggests that enterprise AI spending is broadening from model training into inference, networking, storage, and private deployments.

Training creates or updates an AI model using large collections of data. Inference occurs when that trained model processes a request and generates an answer, prediction, image, or action. It is the recurring workload that follows deployment.

HPE said increased demand for inference and agentic AI workloads helped produce record Cloud and AI results. Agentic AI refers to systems designed to plan and execute multi-step tasks with limited human intervention.

This distinction matters because training projects can arrive as concentrated clusters purchased by a relatively small number of customers. Inference demand can spread across enterprises, government programs, regional cloud providers, and edge locations.

A broader buyer base would support demand for more than graphics processors. Production deployments also need conventional servers, networking equipment, storage, security, orchestration software, and continuing support.

HPE’s earnings presentation showed how that pattern is developing. Traditional server orders increased 75% year over year, driven partly by AI-ready configurations and higher average selling prices.

Private Cloud AI orders grew at a triple-digit rate. HPE describes Private Cloud AI as an integrated hardware and software system for operating AI workloads within a customer-controlled environment.

That approach appeals to organizations that cannot send every document, transaction, or customer record into a public AI service. Banks, healthcare providers, industrial companies, and government agencies often need tighter control over sensitive information.

The company also reported that GreenLake customer count grew 18%. GreenLake is HPE’s consumption-based platform for managing infrastructure across private data centers, edge environments, and public clouds.

Networking formed another part of the growth story. HPE reported $2.9 billion in networking revenue, up 75% as reported. However, much of that increase reflected the Juniper Networks acquisition. On a normalized basis, networking revenue grew 10%.

That distinction prevents the acquisition from being mistaken for entirely organic expansion. Still, networking orders grew 36% on a normalized basis, while HPE booked another $700 million in products designed for AI networks.

The company also disclosed a post-quarter inference agreement with an unnamed hyperscale customer valued at $3.5 billion. A hyperscaler operates computing infrastructure at enormous scale and typically serves cloud, internet, or AI customers.

HPE separately announced a gigawatt-scale networking agreement with Oracle for a large AI cloud buildout. A gigawatt-scale facility requires electrical capacity comparable to a major power station, illustrating how infrastructure requirements now extend beyond computing chips.

NetApp occupies another layer of this stack. Models cannot produce useful business results unless organizations can locate, govern, move, and retrieve the data feeding those models.

NetApp AI storage revenue does not appear as one isolated reporting segment. Instead, the opportunity runs through all-flash systems, public cloud services, data management software, and partnerships with infrastructure providers.

The company released StorageGRID 12.1 to support distributed AI data pipelines through a federated global namespace. That term describes a unified method for locating information stored across different sites or systems.

NetApp also acquired DataPelago, a company developing data infrastructure software for AI workloads. The acquisition price was not disclosed, so its financial contribution remains unclear.

The operating pattern nevertheless supports a broader conclusion. The AI infrastructure cycle is expanding into the systems surrounding accelerators. This expansion gives established enterprise vendors more ways to participate, but it also creates more execution points where projects can stall.

Record Demand Creates a Harder Conversion Test

HPE and NetApp now face pressure from the gap between booked demand and the cash that reaches shareholders.

For HPE, the central test is backlog conversion. Its $7.6 billion AI backlog signals strong customer commitments, but supply limitations continue to restrict shipment timing.

Components needed for large AI systems do not arrive independently. Accelerators, memory, networking equipment, cooling systems, racks, and power infrastructure must be available within the same deployment window.

A shortage in one category can delay the entire system. Customer acceptance can create another delay because revenue recognition often depends on delivery, installation, testing, or contractual milestones.

This helps explain why investors treated the backlog as both an asset and a risk. It creates future revenue visibility, but it also raises expectations for production schedules and working-capital management.

HPE’s reported gross margin reached 40.1%, while adjusted gross margin reached 40.4%. Those gains suggest the company managed its overall product mix well during the quarter.

However, AI systems can carry lower margins than software, services, or specialized networking products. Large customers also have substantial purchasing leverage, especially when a contract involves billions in infrastructure.

HPE needs to preserve profitability while competing for those deals. It must also avoid allowing inventory, supplier commitments, and customer financing arrangements to consume the cash produced by growth.

The company’s post-quarter hyperscaler agreement illustrates that balance. A $3.5 billion contract is significant, but its value depends on delivery timing, component availability, payment terms, and the margin retained after hardware costs.

The Oracle arrangement introduces another uncertainty. HPE issued warrants allowing Oracle to buy HPE shares under undisclosed conditions. Warrants can strengthen a commercial partnership, but they can also dilute existing shareholders if exercised.

The companies did not disclose the number of shares, exercise price, or performance conditions in the initial announcement. Investors therefore lacked enough information to measure the potential dilution.

NetApp faced a different conversion problem. Its revenue, billings, operating margin, and adjusted earnings all exceeded expectations. Free cash flow still fell 35% to $401 million from $620 million one year earlier.

Free cash flow represents cash generated after capital spending. It matters because accounting earnings can rise while changes in receivables, inventory, taxes, or other working-capital items reduce available cash.

NetApp’s GAAP operating cash flow fell 25% to $503 million. The company did not present the decline as a failure of product demand, but investors focused on the mismatch between accelerating revenue and weaker cash generation.

That is the heart of the reversal. HPE AI demand created a record backlog that still requires execution. NetApp AI storage drove record revenue, but the resulting quarter produced less cash than the prior-year period.

Neither signal invalidates the growth story. Both signals raise the standard for the next report.

Investors are effectively asking HPE to ship its backlog without surrendering margin. They are asking NetApp to turn higher billings and earnings into stronger cash conversion.

These expectations are demanding because AI infrastructure projects can be large and irregular. A small number of shipment delays, customer acceptances, or payment schedules can alter quarterly cash results.

The market’s response therefore reflects timing and valuation, not just operating direction. HPE shares had more than doubled during 2026 before the report, according to the initial coverage. A strong quarter had already been anticipated.

When expectations rise that far, an earnings beat can still disappoint. Investors begin comparing the result with an optimistic internal benchmark rather than the published analyst consensus.

Dell Raises the Bar for HPE and NetApp

The main competitive pressure comes from infrastructure vendors already converting AI demand at greater scale.

Dell provides the clearest comparison for HPE. Its server business targets many of the same hyperscale, sovereign, enterprise, and cloud infrastructure customers.

During Dell’s fiscal second quarter of 2027, the company reported $60.9 billion in AI orders and $16.4 billion in AI server revenue. It ended the quarter with a $95 billion AI backlog.

Those figures came from Dell’s earnings call, which also said the company’s AI customer count exceeded 6,500. Dell presented its engineering, global supply chain, financing, and deployment capacity as competitive advantages.

The comparison requires care. Companies can define AI orders, systems, revenue, and backlog differently. A direct ratio may therefore create false precision.

Even with that limitation, Dell’s disclosed scale establishes a demanding market reference. HPE must show that its smaller backlog can produce attractive economics or offer a differentiated mix of networking, private cloud, and enterprise services.

HPE’s Juniper acquisition is central to that argument. The combined portfolio now extends from servers and storage into campus networks, routing, security, and data center switching.

Management expects $600 million in annualized cost synergies from the Juniper integration by the end of fiscal 2028. HPE also said net leverage fell to 1.8 times adjusted earnings before interest, taxes, depreciation, and amortization.

That result placed leverage below its 2.0-times target more than one year earlier than planned. It reduces one concern surrounding a large acquisition, although integration work remains unfinished.

HPE can compete by selling more of the complete deployment. An enterprise may prefer one supplier for servers, networking, private cloud management, and ongoing support instead of coordinating several vendors.

That strategy also creates execution risk. Combining product portfolios, sales teams, channel programs, and support operations takes time. Cost reductions can damage growth if they remove expertise customers need during complex deployments.

NetApp faces a more fragmented opponent set. Dell, Pure Storage, IBM, cloud platforms, and specialized data infrastructure companies all compete for parts of the enterprise storage budget.

Its advantage rests on data management across hybrid environments. Customers may keep sensitive information on-premises while running selected AI jobs in public clouds. They need consistent controls and efficient movement between those locations.

NetApp’s public cloud growth suggests that hybrid demand remains healthy. Its all-flash growth also shows that customers continue modernizing local storage for performance-sensitive workloads.

Yet AI buyers increasingly evaluate storage as part of an entire data pipeline. They consider data preparation, vector retrieval, governance, model access, observability, and lifecycle management alongside raw capacity.

This forces NetApp to prove that its platform remains strategically important as AI applications move closer to cloud databases and specialized data services. The DataPelago acquisition appears designed to strengthen that case.

The competition is not simply HPE against Dell or NetApp against another storage vendor. The larger contest is over which supplier captures the profit surrounding expensive AI accelerators.

Chipmakers capture value through scarce computing components. Cloud providers capture value through consumption. Infrastructure vendors must show that integration, deployment, networking, and data management deserve durable margins of their own.

That is why Google News readers should not interpret the selloff as evidence that AI infrastructure demand has collapsed. The numbers show the opposite. The market is questioning how that demand will be divided.

What the Record Numbers Do Not Prove

One strong quarter cannot establish that every backlog dollar will ship on time or that every AI investment will produce lasting returns.

HPE says AI has become a multiyear growth driver. Its order growth, backlog, and hyperscaler agreement support that position, but they do not independently verify future conversion rates.

The company still depends on suppliers for processors, memory, networking components, and other equipment. Multi-year supply agreements can improve access, yet they also create commitments based on forecasts that might change.

Large AI deployments consume extraordinary amounts of power. They require suitable land, cooling, grid connections, and construction capacity. Hardware can be available while a customer’s data center remains unprepared to accept it.

The concentration of demand also deserves attention. Hyperscalers and large AI infrastructure operators place enormous orders, but those buyers can renegotiate designs or delay deployments when technology changes.

Accelerator generations now advance quickly. A customer may postpone acceptance if a newer architecture offers better performance or lower operating costs. Vendors then face inventory risk or pressure to redesign systems.

HPE’s backlog therefore represents visibility, not guaranteed timing. Investors will need evidence that orders become revenue without creating excessive inventory or receivables.

Margins present another uncertainty. HPE reported excellent companywide profitability, but future quarters may include a different mix of large systems, networking equipment, software, and services.

A higher share of competitive hyperscale contracts can lift revenue while limiting gross profit. Conversely, networking and support services can improve the economics surrounding those systems.

NetApp’s central risk is cash conversion. Its first-quarter free cash flow decline does not establish a structural problem, because quarterly working-capital movements can reverse.

Still, the decline matters when revenue and earnings are accelerating. Investors will expect management to explain whether receivables, tax timing, purchasing decisions, acquisition costs, or other items caused the gap.

NetApp’s all-flash growth also needs context. Some customers are expanding capacity for AI, but others may be refreshing conventional storage or responding to changing component conditions.

The company describes its platform as infrastructure for AI and hybrid multicloud projects. The reported segment data does not isolate how much revenue came directly from production AI workloads.

That limitation does not weaken the reported revenue. It limits how confidently outsiders can attribute every record to artificial intelligence.

The competitive comparison contains similar uncertainty. Dell’s much larger AI backlog appears formidable, but backlog definitions and customer mixes differ. Scale can also increase exposure to low-margin contracts and supply commitments.

Investors are evaluating all three companies against a market that rewards credible AI exposure but punishes weak evidence of economic capture. Revenue growth opens the conversation. Cash flow, margin, and repeatability settle it.

This explains the seemingly harsh response to two record quarters. The market did not reject the demand evidence. It refused to treat demand as a complete investment thesis.

Three Signals Will Decide Whether the Selloff Was Right

The next one to three months should reveal whether investors identified genuine execution problems or simply demanded perfection from strong results.

The first signal is HPE’s AI backlog conversion. Investors should compare future shipments with the $7.6 billion backlog reported at the end of July.

A rising backlog paired with accelerating revenue would strengthen the argument that HPE is expanding capacity while preserving future visibility. A rising backlog with flat shipments would suggest that supply or deployment constraints are becoming more serious.

The mix will matter as much as the total. HPE needs AI systems, networking, private cloud, and services to reinforce one another. Revenue dominated by lower-margin equipment would weaken the economic case even if reported sales remained high.

The second signal is NetApp’s free cash flow. Its $401 million result must be evaluated against the company’s next operating cash flow, receivables, and working-capital disclosures.

A rebound would support the view that the first-quarter decline reflected timing. Continued weakness alongside record revenue would raise harder questions about collections, spending requirements, or the quality of reported growth.

Investors should also watch whether all-flash and public cloud growth remain balanced. NetApp needs performance-oriented systems and cloud data services to contribute together if its hybrid platform strategy is working.

The third signal is the competitive response. Dell’s $95 billion AI backlog gives it scale, but HPE can still differentiate through Juniper networking, private cloud systems, and enterprise relationships.

Watch for new hyperscaler awards, sovereign infrastructure projects, and production inference deployments. Those contracts will show whether HPE’s announced deals represent a repeatable sales model.

For NetApp, the relevant evidence will come from customer adoption of AI-oriented data services. Product announcements matter less than recurring cloud consumption, flash demand, and identifiable production deployments.

The market reaction also deserves a second look after regular trading absorbs the reports. Extended-hours prices can move sharply on thin liquidity and incomplete interpretation.

Still, the initial judgment carried a clear message. Investors want infrastructure companies to demonstrate that AI orders become profitable revenue and that accounting earnings become cash.

That is the useful conclusion behind the Google News headline. HPE and NetApp did not report an AI demand problem. They reported enough demand to make execution, margins, and cash conversion the new sources of doubt.

Enterprise buyers should follow the same signals. A large backlog can affect delivery schedules, configuration choices, support capacity, and negotiating leverage. Cash pressure can influence how aggressively a supplier invests in products and partnerships.

The next earnings reports will test two distinct promises. HPE must turn scarce components and large orders into timely, profitable systems. NetApp must turn storage growth and billings into stronger cash generation.

If both companies deliver, the September selloff will look like a reaction to elevated expectations. If conversion weakens, investors will have identified the limits of record demand before the headline numbers showed them.