HPE's $7.6B AI Server Backlog Signals Supply Crunch Ahead

Record AI demand collides with component shortages

Hewlett Packard Enterprise is sitting on a $7.6 billion backlog of AI server orders—a figure that highlights both the explosive demand for on-premises AI infrastructure and the supply chain bottlenecks constraining the industry. The company reported $3.1 billion in new AI orders during its fiscal third quarter, but actual shipments lagged significantly behind bookings as memory and storage components remain scarce.

HPE's third-quarter revenue climbed 34% year over year to $12.2 billion, while adjusted earnings surged 152% to $1.11 per share. The results, first reported by AI Watch, prompted management to raise both full-year and fiscal 2027 guidance, with the company now projecting at least $5 billion in free cash flow for 2027.

Why it matters

The widening gap between orders and shipments at a major infrastructure provider reveals a critical constraint in the AI buildout: enterprises want to deploy AI systems in their own data centers, but hardware makers can't secure enough components to meet demand. This supply-demand imbalance is driving prices higher and could reshape competitive dynamics as customers prioritize vendors who can actually deliver hardware.

Supply constraints drive pricing power

CEO Antonio Neri acknowledged that tight supply has persisted throughout the year and warned it will "last for a longer period of time, which obviously will have consequences on cost and pricing." The networking segment illustrates the mismatch most clearly—orders grew 36% year over year while normalized revenue increased just 10%.

Server revenue grew 35% in the quarter, but that growth came primarily from higher average selling prices rather than increased unit volumes. Management expects some improvement in shipment capacity during the fourth quarter, though the supply situation remains constrained.

Gross margin reached a record 40.4% in the quarter, which the CFO described as "a confluence of everything coming together at once." However, management expects margins to decline as AI systems—which carry lower margins than traditional servers—account for a larger share of the product mix.

Networking business provides margin cushion

HPE's networking division, now generating roughly 30% of total operating profit, should help offset some margin pressure from AI systems. Networking products carry higher margins than the company average, with management projecting mid-to-high 20% margins for fiscal 2027 compared to approximately 13% for the server segment.

Traditional server bookings also showed strength, rising 75% as customers purchased AI-enabled hardware. The combination of AI-specific orders and AI-capable traditional servers suggests enterprises are preparing their infrastructure for a range of workloads rather than betting exclusively on specialized AI systems.

These details were first reported by AI Watch based on HPE's third-quarter earnings announcement.