HP Enterprise’s (NYSE:HPE) stock surged by around 20% after it expanded its partnership with Nvidia (NVDA) and since it has been oscillating in the $16.5 to $18.5 range as shown in the chart below.
Now, it was already embedding artificial intelligence into its products well before the emergence of OpenAI’s ChatGPT, and its potential merger with Juniper Networks (JNPR) puts HPE in a better position to harvest opportunities in AI networking and hybrid cloud as I had elaborated in an earlier piece in January.
At that time, I was bullish, but, this thesis, which comes ahead of the company’s second-quarter 2024 (FQ2’24) financial results on June 4, and aims to provide an update on the opportunities, emphasizes caution. Thus, more visibility is needed following the revenue decline suffered during FQ1’24 since it has suffered relatively more than its industry peers.
Coming back to the price action, I explain how it is justified by identifying Generative AI-related benefits.
Opportunities in Edge-to-Cloud with Gen AI
First, as seen in the table below, the company boasts an Intelligent Edge segment where sales grew by 2.7% YoY or from $1,169 million to $1,201 million. More importantly, earnings from operations grew by 55.5%, or from $227 million to $353 million, showing this is a highly profitable segment for a company with a superior profitability grade of A.
This segment offers an array of products based on the cloud-native approach including connectivity, security functions for campus (large), branch, data center, branch office, and for work from home. Intelligent network also includes SD-WAN (software-defined wide area network) products where HPE is a leader according to Gartner’s Magic Quadrant for September 2023 alongside Fortinet (FTNT), Cisco (CSCO), and Palo Alto (PNWN) with Juniper positioned as a visionary.
Therefore, in addition to gaining market share in everything from campus, data center, enterprise routing, or wireless LAN, HPE could extend its edge-to-cloud capability. Going a step further, integrating AI into the SD-WAN architecture leads to enhanced AI networking capabilities, key to managing high-traffic and dynamic networks at the edge (outskirts).
However, this concept of AI networking has been around for years, but its evolution should get an acceleration from Gen AI, namely for inferencing purposes. Put simply, this is the actual usage of LLMs (large language models) by software developers to create AI applications like ChatGPT. This is after these models have been trained, using lots of domain data. Furthermore, contrarily to the training part which takes place in metro data centers, the inference process is expected to occur at the edge of the network, or in regional data centers and branch offices. This is where the partnership with Nvidia becomes handy, as it provides HPE with accelerator GPUs for AI servers. Noteworthily, this is in addition to networking AI opportunities.
Talking figures, the company boasted a backlog consisting of AI and HPC (high-performance computing) products and solutions valued at $3 billion in FQ4’23 as part of a $180 billion market opportunity.
Merger Opportunities Still Present but Revenues are Declining
This was the main reason for my bullish position in January, and it was based on the combination raking sales of $36.3 billion for HPE’s FY’25 which ends in October next year, which would have represented growth of 18% versus the 3.6% without the merger. The total consisted of adding HPE’s expected sales of $30.63 billion for FY’25 to Juniper’s $5.5 billion for FY-2025, given that the merger is expected to be completed by the end of this year or the beginning of 2025. However, since then, analysts have downgraded sales expectations for both companies. Hence, HPE is expected to obtain $30.45 billion for FY’25 and $5.41 billion for Juniper. This means a total of $35.86 billion, which would still represent a growth of 17.8%.
Consequently, if the merger goes through, it will still mean double-digit growth for HPE, but I believe it is important to take the downgrades seriously. For this purpose, I look at the current fiscal year, and, as pictured below its topline estimates for FQ2’24 have been reduced from $7.12 billion in March to $6.83 billion in May.
Furthermore, the revenue of $6.76 billion obtained in FQ1’24 constituted a 13.4% YoY decline and a miss of $340 million from what was expected. This was caused by weakness in the Server and Hybrid Cloud segments, which both suffered from revenue decreasing on a YoY basis, as shown in the income statement extract above.
Part of the softness was also attributed to lengthy GPU deal time or not obtaining accelerator chips fast enough to pack them in servers for shipping to customers, thereby creating revenue shortfalls. As for Juniper, its sales declined by 16.1% YoY, while it missed estimates by $80 million.
Reason for Revenue Decline by Looking Across the Industry
Looking across the industry, others seem not to have suffered to the same degree. Thus, Cisco, which also has a partnership with Nvidia, suffered from a revenue decline of 12.7% during its latest reported quarter but still managed to beat topline estimates by $70 million while Fortinet delivered both revenue growth and a topline beat (by $10 million). As for Extreme Networks (EXTR), while revenues for its latest reported quarter declined by 36.53% YoY, it still managed to beat estimates by $6.40 million.
Moreover, HPE’s estimate for FY’24 revenue growth of 0.06% (picture below) is not aligned with the 8% IT spending expected to grow in 2024. Hence, according to Gartner, data center systems and communications services should grow by 10% and 4.3% respectively.
Looking for an explanation for the weakness in networking, it could be related to the regulatory approval for the merger.
In this case, the regulatory review process has seen the U.S. Department of Justice requesting the two stakeholders for additional information. This supposes antitrust concerns and potential risks to competition within the industry. This could also imply regulators asking HPE to dispose of part of the assets for the deal to go through. Moreover, the two have overlapping products, namely in the switching portfolio.
While the problem may not be dramatic, this can still create uncertainty in the minds of customers concerning product roadmaps, or whether a product will be supported after the merger and for what period. This was the case when VMWare was acquired by Broadcom (AVGO) last year. Moreover, the semiconductor giant’s pricing (licensing) strategy could have motivated customers to switch from its ESXi product, which is primarily on-premise, to VMware Cloud. This uncertainty resulted in competitor Nutanix (NTNX) launching a promotion including free deployment and migration services for customers wishing to ditch VMware.
This idea that uncertainty around overlapping could benefit competitors including Cisco, Extreme, Fortinet, and Palo Alto is also mentioned by Dell’Oro Group, depending on whether they capitalize on the situation which is usually the case in a competitive landscape.
A Hold While Waiting For Visibility during FQ2’24 Earnings Call
Now, the longer it takes for the $14 billion merger to be completed, the higher the degree of uncertainty, and the more likely for customers to think twice before ordering from HPE or Juniper, especially for long-term projects. As a result, demand could be impacted, increasing the probability of analysts revising the topline lower. Alternatively, the company could again miss topline estimates of $6.83 billion because weakness in networking is likely to persist. Also, it could fall short of the EPS expectation of $0.39 if it has to provide discounts to achieve sales targets. This can result in volatility risks for the stock, just as on February 29 when it lost 3% after missing topline estimates.
Therefore, it is crucial to assess whether the company is continuing to lose market share in the networking business. For this matter, HPE and Juniper should account for just 9% of the enterprise network market share after the combination, compared to 43% for Cisco whose ethernet AI switches could provide around $3 billion of sales opportunities as per my earlier thesis. Therefore, HPE faces competition in super-smart networks too and, to this end, an update on the progression of the AI backlog is crucial.
Staying within the AI realm, an update is needed on the supply of GPUs to equip the HPE ProLiant Gen11, which is compatible with Nvidia’s sophisticated H100 GPU chips. This is an area to watch closely during the forthcoming earnings call as, based on-demand complementarity, the company’s storage business should also benefit. Along the same lines, another item to look for is the HPE GreenLake, or the pay-as-you-go hardware purchasing scheme for customers, whose ARR exceeded $1.4 billion in FQ1’24 as it also benefited from AI.
Looking at valuations, HPE remains underpriced based on several metrics including forward price to sales of 0.82x where it trades at a discount of 72% relative to the IT sector. However, it was trading at an even lower P/S of 0.71x in January and was priced at $17.72.
Given the revenue estimates of $35.86 billion for the merger instead of $36.3 billion, as I mentioned above, I reduced my January target of $19.8 to $19.56 ((35.86/36.3)x 19.8). Now, this represents a 6% upside from the current share price of $18.4, but I have a Hold position while waiting for further visibility. This is especially so in an IT market expected to grow faster this year than in 2023, as in such instances, laggards tend to be punished hard by the market.
Finally, an update on the M&A is required, especially in case of any blocking points, but combining Juniper’s portfolio will boost HPE’s edge-to-cloud computing strategy as it provides an optimized way for software developers located on the outskirts to access centralized Gen AI servers.
Read the full article here