HPE CEO Antonio Neri said HPE GreenLake is an “important component” of HPE’s AI strategy because it provides customers “confidence and control of their data which is the fundamental value when it comes to AI.”

Hewlett Packard Enterprise CEO Antonio Neri said HPE’s GreenLake pay-per-use on premise cloud service is helping fuel AI server sales growth.

Neri said HPE GreenLake is an “important component” of HPE’s AI strategy because it provides customers “confidence and control of their data which is the fundamental value when it comes to AI.”

HPE converted approximately $1.3 billion of orders into AI systems revenue in the third fiscal quarter ended July 31, up approximately 40 percent increase from the last quarter.

AI system orders, overall, grew $1.6 billion sequentially in the quarter to a cumulative $6.2 billion since the first quarter of Fiscal Year 2023.

As for HPE GreenLake sales momentum, Neri told CRN that the hybrid cloud powerhouse added 3,000 new HPE GreenLake customers in the quarter with an overall increase of 39 percent in the all important high-margin HPE annualized revenue run rate.

“On hybrid cloud, we now have 37,000 customers on the (HPE GreenLake) platform, which is 3,000 more customers than we had in Q2 and 10,000 more customers than we had a year ago,” said Neri. “That’s driving AI systems growth of 39 percent (to $1.3 billion) and the mix of ARR (Annualized Revenue Run Rate) improved again to 71 percent software and services.”

That 71 percent mix of software and services that comes with HPE GreenLake and the ARR financial model is driving higher margins for HPE and its partners. In fact, HPE’s ARR mix was up 300 basis points year over year due to the stronger sales of AI services tied to AI server sales and the Aruba Central platform.

Overall, HPE’s ARR in the third fiscal quarter grew 39 percent to $1.7 billion.

HPE is seeing what HPE Senior Vice President and Chief Financial Officer Marie Myers is calling a “strong appetite” for enterprise AI customers to leverage HPE GreenLake to “alleviate investment pressures as well as to retain flexibility to grow” workloads.

With AI server sales momentum, HPE server sales in the quarter were up 35 percent from the year ago period to $4.3 billion with a 10.8 percent operating margin compared with 10.1 percent in the year ago quarter. That operating margin is up 70 basis points year over year but down 20 points sequentially.

HPE server sales were up 27 percent through partners,said Neri. He said the server segment overall out performed expectations “thanks to an acceleration in converting AI system orders to revenue.”

AI system revenues now make up 30 percent share of total system revenues, up from just 10 percent in the year ago quarter.

Neri said the strong server sales growth shows that there is no “cannabilization” of traditional servers by the increase in AI server sales. That’s because, Neri said, “it does not make a lot of sense to put a traditional workload that’s running fairly efficiently on a traditional server into a GPU. Most of the GPU is consumed to do all this stuff with AI. So why would you do that?”

Here is an excerpt from the discussion with Neri.

What were the highlights of the quarter?

First of all obviously the headline is that we delivered a very strong order performance. We saw impressive revenue growth driven by number one, the AI systems conversion to revenue. Strong demand continuing in AI, but we saw demand continuing across all businesses.

We saw sequential order growth across all businesses, and we did all of that while improving profitability on a quarter over quarter basis. And that’s why, on the revenue side, we delivered 10 percent year over year (revenue growth), which was on the high end of our guidance. And on the EPS side, we delivered 50 cents, which is two pennies above the high end of the guidance. So based on this performance, we are raising again, GAAP and Non-GAAP EPS, for the full year 2024. That’s the financial view.

The only add to that is that we closed the first tranche of the H3C sale. Remember we have been talking about this for some time. That first tranche has closed and we have collected $2.1 billion already. And remember this was a complicated process because we are a partner (selling 30 percent of ishares in H3C to Chinese IT provider Unisplendour) in China, selling to a Chinese partner and had a lot of regulatory approvals to go through.

We are on track to close the ($14 billion acquisiton of) Juniper (Networks) by the end of calendar 2024 or the beginning of calendar year 2025. In fact, we have already got approval from the European Union, UK, India, Japan and many others.

Last but not least, we are not thinking just AI or networking. We are also thinking about hybrid cloud, because they’re all synergistic. We announced and closed the Morpheus transaction at the end of August.

What do you see with regard to AI performance in the quarter?

If you look at AI, we booked $1.6 billion in new orders in the quarter. That brings the total cumulative orders to $6.2 billion of which $3.5 billion has been booked in the last four quarters.

So if you take the last four quarters, $3.5 billion and $1.6 billion in just Q3, you can see there has been an acceleration (in AI sales).

Of the current backlog, which we exited the quarter with $3.4 billion in backlog in AI, new orders obviously contributed to that. But as the dollars got bigger, the percentage from enterprise is still 15 percent. So while the mix has not shifted, the dollars have grown correspondingly (in enterprise) as we went up with $1.6 billion in AI (orders).

What did you see in the networking business in the quarter?

In networking we see a market recovery along with our peers. We saw orders growth sequentially, and it was driven across all geographies and across wireless LAN, data center switching and SASE. Campus switching is taking a little longer to recover.

What are you seeing in the hybrid cloud segment?

On hybrid cloud, we now have 37,000 customers on the (HPE GreenLake) platform, which is 3,000 more customers than we had in Q2 and 10,000 more customers than we had a year ago. That’s driving AI systems growth of 39 percent and the mix of ARR (Annualized Recuring Revenue) improved again to 71 percent software and services.

On the storage side (of the business), we saw double digit year over year growth with HPE Alletra in orders. But obviously on the revenue side that lags because a significant portion of our revenue gets deferred because of our subscription model to the platform.

How did the channel perform for HPE in the quarter?

First of all of the $7.7 billion we reported in sales $4.1 billion came through the channel, which is 57 percent of our sales. That is five percent growth year over year.

The cloud business, which is the hybrid cloud segment, (through the channel) had 14 percent year over year growth and 5 percent quarter over quarter. HPE Alletra is basically 6 percent of the total indirect (sales) in Q3, but Alletra went from 6 percent to 19 percent. So you can see the acceleration.

In the server market, we grew servers through the channel 27 percent year over year.

In the networking business, we grew 7 percent quarter over quarter through the channel.

What kind of tailwind for AI sales at this point in time particularly with HPE Private Cloud AI?

The momentum for service providers and the large language (GenAI) model builders continues to be very strong. And you see that in the order bookings. Enterprise is growing and you can see that because, as I said, 15 percent of our total (AI) orders are coming from enterprise on a larger number. And that is not even with HPE Private Cloud for AI on the market. That just became available yesterday for ordering.

We announced it at (HPE) Discover and it took us six or seven weeks to make it available which is pretty fast by the way considering how much work it takes.

That gives me tremendous hope and excitement about the growth we see (in AI) because customers are now maturing their business models, and they are becoming more confident in the ability to use this technology. And then with the sovereign clouds (with AI) there is huge interest, but it takes time, because the design and procurement process are elongated with the government (approval) and other things that are involved in it along the way. Sometimes venture capital has to be aligned to the opportunity and the like.

I would say one segment is very strong. The other ones are getting stronger and stronger. So as we go into the future, you’re going to see growth related to the other segments that keeps raising the tides in this space.

I’m also very pleased that the traditional server business has been very strong. We grew orders and revenue double digits year over year, which basically demonstrates that there is no cannibalization (with the AI server business). Why is that? It’s simple. It’s because it does not make a lot of sense to put a traditional workload that’s running fairly efficiently on a traditional server into a GPU. Most of the GPU is consumed to do all this stuff with AI. So why would you do that?

So do you think we are seeing a renaissance in server sales going forward?

I don’t know if it’s a renaissance. I would say that it is the typical cycles where people need to modernize (their servers). The market goes up and down with these cycles. At the same time, remember that we are also driving intentionally a transition to Gen11 (servers) and more than 60 percent of our servers in the traditional space are now on Gen11. So we have been doing that over the last several quarters.

And then a year from now, whatever it is, I don’t know the exact time frame, but depending on the silicon availability, we’re going to be on Gen12.

So this is the nature of the business, but what I see is structural pricing, because of the density of the servers, gets bigger and bigger. That drives AUPs (Average Unit Prices) higher. Then obviously you have to deal with the dynamics of the commodity (components) market, which in the DRAM space is slightly up. And the NAND (Flash Memory) space is up more than DRAM. That also drives AUPs.

But listen we saw (server) units growth on a year over year basis and I am happy with that. But in the end it is all about what I call the synergistic (effect) between networking, the cloud and AI, all delivered through (HPE) Greenlake.