
Last 4 earnings calls · through Q3 FY2026
The bull case is that AI data center build-outs and enterprise inferencing are lifting demand for servers, storage and networking at the same time.
HPE argues that owning Juniper's routing and switching lets it sell a full stack into its existing server accounts and win larger deals than either company could alone.
Management has moved its long-term profit and cash targets forward, which bulls read as proof the integration and cost programs are ahead of plan rather than a one-off.
Rising cash generation and the sale of the China stake are paying down acquisition debt, which bulls expect to leave room for steady returns to shareholders.
The bear case is that HPE cannot ship what customers order because memory and wafer capacity are short, so growth depends on allocations it does not control.
Bears argue that much of the recent growth comes from raising prices rather than selling more units, and that this fades once memory costs and supply settle.
As bigger AI systems deals ship, the revenue mix shifts toward lower-margin hardware, which bears say will pull gross and operating margins back toward historical levels.
Bears note that HPE still competes with larger network vendors and that its data center switching trajectory rests on a small number of very large customers.
In short: Orders outgrow revenue in networking and servers, but supply limits how much of that demand HPE can ship.
Management first targeted $1.5 billion of cumulative AI networking orders by fiscal 2026, then lifted the goal to at least $2 billion and later to $2.5 billion to $3 billion as AI data center demand grew.
HPE's largest product line returned to strong growth as customers bought traditional servers to run agentic AI and inferencing, which also pulls through storage, software and services attachments.
Management said networking orders grew about 3.5 times faster than revenue, so the binding limit on growth is component supply rather than customer demand, and conversion timing sits largely outside HPE's control.
Management said unit volumes were constrained and much of the increase came through average selling prices, which could unwind once memory costs and component availability normalize.
In short: Profit and cash flow are rising fast, though a larger AI systems mix is expected to pull gross margin down.
HPE had committed to at least $3 of earnings per share in fiscal 2028; management now expects $3.4 at the midpoint in fiscal 2026 and has raised the free cash flow goal twice since.
Cash generation of that scale, against $225 million in the prior twelve months, funds the dividend, buybacks and the debt taken on for Juniper without new borrowing.
Gross margin reached 40.1% last quarter on favorable deal mix and pricing; management now expects a larger share of AI systems revenue to pull the rate back down toward its historical range.
The Cloud & AI segment that houses AI systems operates at a much lower margin rate than networking, so a richer AI mix pulls the company's overall margin rate down each quarter it ships more heavily.
In short: Juniper adds routing, switching and security, while memory makers set HPE's server costs.
HPE merged the Juniper and Aruba sales teams into one organization and can now sell campus, data center switching, routing and security together, which is what the largest customer deals require.
The win puts HPE's QFX switches and PTX routers into very large AI cloud sites, and its own routing silicon is a differentiator that only a couple of competitors can match at that scale.
Orders in that product line run well ahead of shipments, so reported revenue depends on component availability; the shortfall shows how easily HPE's networking growth can slip into a later quarter.
DRAM and NAND make up more than half of a traditional server's material cost, so memory makers' capacity decisions rather than HPE's own pricing determine server profitability.