Hewlett Packard Enterprise Co.’s artificial intelligence servers sold like crazy during its fiscal second quarter, powering the company to an impressive earnings and revenue beat, and the company’s stock soared in extended trading.

For the quarter, HPE reported earnings before certain costs such as stock compensation of 79 cents per share, blowing away the Street’s consensus estimate of just 53 cents per share. Revenue for the period came to $10.68 billion, up 40% compared to the same period one year ago and comfortably ahead of the Street’s $9.79 billion target. The quarter was the company’s biggest earnings-per-share beat since February 2018.

AI is, of course, the main reason for HPE’s barnstorming results. The company said that its overall Cloud & AI revenue came to $7.71 billion, surpassing the Street’s target of $6.87 billion.

However, it was the server business that really set the cat among the pigeons. Revenue from that sub-division of the Cloud & AI segment topped $5.45 billion, well ahead of the $4.66 billion analyst forecast. Moreover, although the profit margins on AI server sales are notoriously slim, HPE was able to bump up its overall net earnings to $624 million, from a net loss of $1.05 billion one year ago.

The company’s networking business was also a strong performer, with revenue there rising 148%, to $2.7 billion in the quarter.

President and Chief Executive Antonio Neri (pictured) hailed the company’s “exceptional quarter” and its much higher-than-anticipated profitability, saying it reflects both the company’s strong execution and the “healthy demand” it’s seeing across its business. “Customers continue to invest in modernizing their infrastructure and scaling AI, and our performance shows the strength of our combined networking portfolio and the value we are delivering to shareholders,” he added.

Adding to the company’s impetus, officials said they’re raising its full-year earnings guidance by a full dollar per share. That means it now projects earnings of between $3.35 to $3.45 per share, up from its earlier estimate of $2.30 to $2.50 per share. In contrast, Wall Street analysts are looking for full-year earnings of just $2.43 per share.

On a conference call with analysts, Neri said bookings for traditional servers are up by triple digits from where they were a year ago, amounting to the biggest backlog in the company’s history. He explained that customers in security-focused industries have significantly increased their demands for on-premises AI servers and infrastructure, as opposed to cloud-based AI resources. That suits HPE just fine, as it primarily caters to enterprises and government agencies, rather than cloud hyperscalers, which prefer to buy up unbranded servers and equipment in bulk.

Analyst Patrick Moorhead told CNBC that HPE is focused on the higher-margin opportunities that these customers provide, as opposed to its main rival Dell Technologies Inc., which targets neoclouds such as CoreWeave Inc. “Their bump-up has everything to do with increasing the profitability of AI, and the fact that they’re hitting their targets a year and a half in advance,” he said.

HPE still faces challenges in negotiating the global memory crunch, Neri conceded. He said he anticipates that costs will remain “elevated” until 2027 at least.

It’s hard to pick many faults with HPE’s latest quarter, said Holger Mueller of Constellation Research. He said the company’s performance was undoubtedly stellar in every way, helping it to move past the $10 billion in quarterly revenue milestone for the first time in its history. “It was driven by the AI boom of course, and related to that, its networking division is doing very well too, albeit it received a big bump from the acquisition and amalgamation of Juniper into that segment,” Mueller said. “Most impressive was the translation on the bottom line, with HPE executing a $1.6 billion swing to profit, having lost money four quarters ago. That’s a remarkable feat by Antonio Neri and his team.”

According to Neri, the company is expecting big things from its next-generation AI hardware, including the new ProLiant edge-computing servers it announced in April and the latest ProLiant server rack based on Nvidia Corp.’s Vera central processing units, which was unveiled early Monday at the Computex conference in Taiwan. It’s designed for cost-efficient AI inference workloads.

“This is going to be a major growth driver,” said Nvidia CEO Jensen Huang during a keynote. He added that millions of the Vera CPUs are currently in production ahead of a rollout later this year.

Neri told analysts that the new server, which will launch globally in the fall, has been optimized especially to run “agentic” AI workloads, where autonomous bots perform tasks on behalf of humans. “These workloads require exceptional CPU performance to enable real-time reasoning across agentic AI and financial services applications,” Neri said. “We are delivering a new class of infrastructure to help customers accelerate insights and operate with confidence in the most demanding environments.”

For the current quarter, HPE said it’s looking for earnings of between 84 and 89 cents per share on sales of between $11.5 billion and $12.1 billion. In contrast, Wall Street analysts are modeling earnings of just 66 cents per share on revenue of $10.9 billion. Investors lapped up the company’s results, and its stock jumped more than 25% after-hours, adding to some impressive momentum that has seen it rise more than 95% in the year to date.

Photo: SiliconANGLE