Sign in

    Hewlett Packard Enterprise Co (HPE)

    Q1 2024 Summary

    Published Jan 10, 2025, 5:10 PM UTC
    Initial Price$15.29October 31, 2023
    Final Price$15.29January 31, 2024
    Price Change$0.00
    % Change0.00%

    What went well

    • Strong demand for AI systems is driving significant growth opportunities: HPE's cumulative orders for AI systems reached $4 billion, with a backlog of $3 billion, nearly tripling year-over-year. This robust demand is expected to drive a very strong second half, predominantly driven by AI systems revenue. , ,
    • Expansion of Annual Recurring Revenue (ARR) through HPE GreenLake platform: A significant portion of AI system orders is going through HPE GreenLake, contributing to ARR growth. HPE GreenLake now serves over 31,000 customers, up approximately 8% quarter-over-quarter (about 3,000 new customers), enhancing margin expansion over time.
    • Stabilization and sequential improvement expected in servers and storage segments: HPE sees signs of stabilization with sequential order improvement in traditional servers and storage, shifting to higher-value Gen11 servers, which have higher pricing and configurations. This shift is expected to drive revenue growth and improve operational services attachment rates.

    What went wrong

    • Declining demand in the Intelligent Edge segment and softness in the networking market: HPE experienced a significant revenue shortfall in the Intelligent Edge business due to weakened customer demand, particularly in Europe and Asia. This softness in the networking market is expected to persist throughout the year, impacting revenue growth.
    • Delayed revenue recognition from AI server orders due to GPU supply constraints and customer readiness: The company is facing delays in converting AI server orders into revenue because of tight GPU availability and customers taking longer to prepare data center space, power, and cooling requirements. This affects HPE's ability to capitalize on strong AI demand in a timely manner.
    • Elevated GPU lead times impacting order fulfillment and revenue conversion: Despite some improvement, GPU lead times remain elevated at over 20 weeks, limiting HPE's capacity to fulfill orders and convert its $3 billion AI backlog into revenue. This extended lead time poses a risk to revenue growth in the highly competitive AI server market.

    Q&A Summary

    1. AI Systems Revenue and Backlog
      Q: How is AI revenue impacting your ARR growth?
      A: A significant portion of our $4 billion in cumulative AI orders is going through HPE GreenLake, contributing to ARR growth. We shipped around $400 million in AI revenue and ended the quarter with a $3 billion AI backlog, nearly tripling year-on-year.

    2. GPU Supply and Lead Times
      Q: Have GPU lead times improved, and how does that affect backlog conversion?
      A: GPU lead times have come down but are still elevated at 20-plus weeks. We continue to need more supply against our $3 billion backlog. However, we're confident in converting GPU orders into revenue as we see improvements in supply and customer acceptances.

    3. Intelligent Edge and Networking Outlook
      Q: What's changed in Intelligent Edge, and is the Juniper deal affecting demand?
      A: We saw an acceleration of demand softness in January, but we do not have a channel inventory problem. Customers are taking longer to install shipped products. We haven't lost any deals due to the Juniper acquisition announcement, and we expect a slight improvement in the back half with Q2 being the trough.

    4. Server and Storage Growth Expectations
      Q: Why do you expect sequential growth in servers and storage?
      A: We see signs of stabilization with sequential order improvement. There's a mix shift to Gen11 servers, which are higher-priced, and we'll pass on cost inflation. Units are stable or improving, driving Operational Services attach rates. In storage, AI is a pull-through demand, and Alletra is growing, contributing to ARR through disaggregated software subscriptions.

    5. Profitability of AI Servers
      Q: How does accelerated compute affect server profitability?
      A: Our Server segment delivered strong performance within the 11% to 13% target range. Selling an XD system can be 20× the value of a traditional server, and an EX system up to 35×. This allows us to optimize margins through configurations and service attachments, driving better shareholder outcomes.

    6. AI Inferencing Demand
      Q: How do enterprise AI inferencing deployments differ from training, and when will they impact revenue?
      A: Enterprises focus on fine-tuning models with their data in secure environments, leading to inferencing at the edge. Use cases include retail and banking. Growth in this area will materialize in the second half and into fiscal '25, given lead times. The Juniper acquisition enhances our ability to deliver these solutions.

    7. Shift to GPU Accelerated Computing
      Q: Will x86 applications shift to GPUs as Moore's Law ends?
      A: We believe traditional CPU-based architectures and GPU-accelerated AI applications will coexist. Not all applications will move to GPUs; some inferencing solutions will continue to run efficiently on CPUs. The transition will take time, and the mix depends on how AI applications are constructed.

    Related Publications