“I characterize this as worse than the pandemic because in the pandemic it was all about getting the right products and services to support mission critical customers like hospitals that we’re serving society at large,” said Neri in an interview with CRN.
HPE CEO Antonio Neri told CRN that the current memory supply crisis, which has resulted in sharp price increases in the cost of memory, is a bigger problem than the supply constraints the industry faced during the global COVID-19 pandemic.
“I characterize this as worse than the pandemic because in the pandemic it was all about getting the right products and services to support mission critical customers like hospitals that were serving society at large,” said Neri in an interview with CRN. “This is about meeting the customer demand through a massive inflection point called AI, and this hypercycle in AI is driving a dislocation in supply-demand mismatching and a violent increase in cost.”
The COVID-19 pandemic — which began in 2020 and ended in 2023 — resulted in a dramatic increase in IT hardware prices and supply constraints with long lead times for products as companies scrambled to provide work at home products for customers.
The current memory supply crisis has caused HPE and other vendors to raise server and storage product prices and implement strict new measures including the ability to reprice server orders up until day of shipment.
The current memory crisis is complicated by the fact that there is also a “technology transition” to HBM (High Bandwidth Memory) version 4 from version 3, said Neri. “That is why this is a multifaceted, complex situation,” he said.
The current crisis has its roots in the impact that the COVID-19 global pandemic had on component and memory makers, said Neri.
“You have to go back all the way to 2022 and 2023 to understand how we got in this situation,” he said. “Many of the (memory) suppliers after they made a big COVID investment (to provide additional supply) to support the human disaster and actually lost a lot of money. They could not recoup the capex (capital expenditure) investment because demand declined. Now they need to make another huge investment to meet demand and obviously they are looking at 2028, 2029 and 2030 as the years to get the payoff from that investment. That is why this is a complex situation.”
Neri’s comments came after HPE reported better-than-expected non‑GAAP earnings for its first fiscal quarter ended Jan. 31 of 65 cents per share. That was six cents above the Zacks consensus analyst estimate of 59 cents per share. HPE reported sales of $9.3 billion for the quarter compared with the Zacks consensus estimate of $9.32 billion.
Neri told analysts that rising memory prices — which have resulted in tough new measures including the ability for HPE to reprice server and GreenLake orders up until shipment —will persist well into 2027.
“We need that ability to cancel an order to the customer who placed the order,” he said. “We work with the partner and the customer to figure out how to best fulfill that demand, understanding there is a cost increase. The question is how can we capture that demand in a very cost effective way if we have the supply?”
Neri said the key for HPE is to provide “as much as possible line of sight” to channel partners in the context of what supply is available and what configurations HPE can offer to help remedy the situation.
Here is an edited excerpt from the interview with Neri.
What is your view on the first fiscal quarter results?
We had a strong quarter where we grew revenues 18 percent obviously adding Juniper (which HPE acquired last July for $13.4 billion). We have strong momentum across both segments of the market (Networking and Cloud & AI).
We delivered a record profitable quarter which resulted in (non‑GAAP earnings of) 65 cents (per share), well above our outlook range, and we translated that into strong free cash flow of $708 million. In general, Q1 tends to be a consumer of free cash flow, or cash versus this time it is a positive — a generation of free cash flow.
I’m very pleased with the performance in both segments. But definitely networking stood out because we outperformed our own expectations. So we grew revenues on a normalized basis seven percent but the orders momentum is very, very strong.
We grew orders on a normalized basis by high single-digits in campus and branch, and we grew data center switching mid 40 percent year over year, routing mid 20 percent year over year. And then when you think about networks for AI, which is a combination of switching and routing, particularly in the context of AI and AI buildouts, we are raising our networks for AI cumulative orders intake to $1.7 billion to $1.9 billion up from $1.5 billion that we communicated at SAM (Securities Analyst Meeting last October).
So when you put it all together we are raising our network revenue outlook to the high single-digits for the year. And now networking represents more than 30 percent of our company revenues and more than half of the profit of the total company.
So networking clearly is paying off for us. The Juniper acquisition is paying off, and we just completed phase one of the integration, which at the core was onboarding employees, working the strategy, executing a product roadmap, and then completing the integration of the sales force. So now we have one integrated networking sales force. That gives us the confidence to now scale revenue synergies as we go forward. So that’s on the networking side. Very positive. Super excited. The team is doing an awesome job, and we are on track to deliver the Juniper synergy targets that we committed for (Fiscal Years) 2026, 2027 and 2028.
For cloud and AI we did what we said we would do. We grew orders in traditional server thanks to our channel partners, obviously, low double-digits.
We grew our Alletra MP orders by high double-digits. In fact, we grew orders 42 percent year over year. Our Private Cloud AI or the AI factory for enterprise grew for the fourth consecutive quarter.
Our software, which obviously is the Hybrid Cloud Ops suite with a focus in particular on virtualization with VM Essentials grew for the third consecutive quarter on a sequential basis.
Our GreenLake platform now has almost 50,000 customers on the platform. The number of devices connected to the platform is growing very rapidly.
The Juniper Mist and Aruba Central group was up 28 percent year over year. Our Wi-Fi 7 access points grew 10x year over year.
For the contribution of the channel in the quarter, we saw the indirect business, the channel business, growing 48 percent year over year. That now accounts for 71 percent of the total HPE business compared to 65 percent in Q4.
The results were super strong. That is why we are raising our networking revenue to the high single-digits. We are raising our earnings per share by five cents at the mid-point and we are raising our free cash flow to at least $2 billion for fiscal year 2026, and are paying down the debt in alignment with our commitments as part of the Juniper deal. So we exit Q1 with a net debt ratio of 2.6 times.
What has been the situation with rising commodity component prices and availability with rising memory prices?
On the commodity side, obviously, it is a challenging situation. So there we focus on three key things. Number one is trying to secure as much supply through capacity allocation with our partners. So we are extending our multi-year agreements where doable and possible. Number two is obviously we have to provide as much as possible line of sight to our channel partners in the context of what supplies are available and what configurations we can offer. Because there is more supply in one area than others. So shaping demand is going to be important. And then number three is the fact that we will have to raise prices together with our channel partners.
What is the customer response to the rising memory prices and how does this compare to the global pandemic?
Last week I was in Europe for Mobile World Congress and then I went to London where I had more than 20 meetings with customers.
Customers understand this. There is a realization this situation will last well into 2027. So what they want is transparency. What they want is direct communication working with our partners to be able to shape that demand. But I have to tell you, there was not one customer that told me: ‘Sorry Antonio I don’t want the product anymore.’ Actually it was the opposite. They wanted to know how they can get the product and how we can work these pricing issues as fairly and constructively as possible.
So with our partners we have to do the best job to preserve profitability and serve our customers the best we can. But this situation will last at least well into 2027 and we just need to deal with it the way we normally do, which is through the partnership, transparency and the ability to shape demand.
How is demand being impacted by this situation and how does it compare to the COVID-19 global pandemic in 2020?
Demand is very strong. I think it will continue to be strong even with the cost and challenges we see in the market. Everybody needs technology.
I characterize this as worse than the pandemic because in the pandemic it was all about getting the right products and services to support mission critical customers like hospitals that were serving society at large.
This is about meeting the customer demand through a massive inflection point called AI, and this hypercycle in AI is driving a dislocation in supply-demand mismatching and a violent increase in cost. There is also a technology transition happening underneath not just from DRAM to HBM (High Bandwidth Memory), but even with HBM there is a transition from version 3 to version 4 as we increase the number of cores in the GPUs and also in the CPUs. So that is why this is a multifaceted, complex situation.
You have to go back all the way to 2022 and 2023 to understand how we got in this situation. Many of the (memory) suppliers after they made a big COVID investment (to provide additional supply) to support the human disaster and actually lost a lot of money. They could not recoup the capex (capital expenditure) investment because demand declined. Now they need to make another huge investment to meet demand and obviously they are looking at 2028, 2029 and 2030 as the years to get the payoff from that investment. That is why this is a complex situation. We as a company will do our best working with our partners.
Any comment on how HPE is handling the memory crisis versus Cisco, which canceled compute promotions and incentives including deal registration?
I cannot comment on what Cisco is doing or not doing. We need that ability to cancel an order to the customer who placed the order. We work with the partner and the customer to figure out how to best fulfill that demand, understanding there is a cost increase. The question is how can we capture that demand in a very cost-effective way if we have the supply. I always say a quick no is better than a long yes.
Is the networking success playing out faster than you thought given how quickly you are integrating Juniper Networks?
Well, the team is doing a fantastic job. Look we did extensive planning. I’m not surprised about the progress on that integration because obviously I was part of that planning process.
I’m really pleased with the momentum we have in the market and the amazing innovation we were able to bring to the market during that integration process.
Look at all the innovation we announced at (HPE) Discover (Barcelona), NRF (National Retail Federation) and Mobile World Congress. Look the story (of our AI innovation and the impact in the market) will be told years out (from now),
