Hewlett Packard Enterprise has signed a $1.2 billion agreement with cloud infrastructure provider Vultr to deploy AMD's new Helios rack system across AI data centers. The deal marks HPE's first commercial rollout of the Helios platform and lands alongside a broader upward revision of the company's long-term revenue forecasts, a signal that demand for AI infrastructure continues to outrun expectations even among hardware vendors accustomed to rapid growth.
What the Vultr Agreement Signals
HPE CFO Marie Myers framed the Vultr contract as more than a single transaction. Speaking on Yahoo Finance's Market Domination, she described the unveiling of HPE's accompanying network switch - built around roughly 1,700 copper cables - as a highlight of the announcement, calling it emblematic of the engineering density now required to support modern AI workloads. The switch and the Helios rack system are designed to work together, reflecting how data center hardware providers are increasingly selling integrated stacks rather than standalone components.
For Vultr, a cloud infrastructure company competing against far larger hyperscalers, access to AMD's latest GPU architecture through an established enterprise partner like HPE offers a way to scale AI compute capacity without building silicon relationships from scratch. For HPE, the deal extends a pattern: last quarter the company disclosed a separate data center agreement with Oracle, suggesting a deliberate strategy of diversifying its AI infrastructure customer base beyond any single hyperscale buyer.
Why AMD, and Why Not Exclusively
Myers was explicit that HPE's embrace of AMD's Helios architecture does not represent a shift away from other chip suppliers. "We've always used all different types of silicon from different vendors," she said, positioning the AMD partnership as one piece of a multi-vendor strategy rather than a bet on a single supplier. That approach matters in a market where GPU supply has been constrained and where enterprise customers are wary of depending on one chipmaker for mission-critical AI infrastructure. Keeping multiple silicon partners gives HPE negotiating leverage and gives customers choice, both of which have become competitive differentiators as AI hardware costs climb.
Networking as the Growth Engine
Alongside the Vultr news, HPE raised its networking segment outlook, now projecting data center networking revenue growth in the low-to-high 50s percentage range through 2029. Myers pointed to two converging trends driving that forecast: continued expansion in AI data center buildout, and rising enterprise interest in self-operating, autonomous networks. She cited ServiceNow's stated goal of reaching full network autonomy by 2028 as an example of the ambition now shaping networking demand - a goal Myers suggested only a small number of infrastructure providers are currently equipped to support.
The framing she used - "AI for networks, networks for AI" - captures a dynamic reshaping enterprise technology budgets. Networking equipment is no longer a secondary consideration behind GPU procurement; it has become a bottleneck and a differentiator in its own right, since AI training and inference workloads depend heavily on how efficiently data moves between processors.
What It Means for the Broader Market
Deals of this size underline how AI infrastructure spending has shifted from experimental budgets to long-term capital commitments. Enterprises and cloud providers are locking in multi-year hardware arrangements, which gives suppliers like HPE and AMD more predictable revenue but also raises the stakes if AI demand growth slows or if customers face their own cost pressures. Investors evaluating these announcements should weigh them against the broader pattern of concentrated AI capital expenditure across a relatively small number of large contracts, rather than treating any single deal as proof of durable, broad-based demand.