Capital · United States

HPE’s Rahim: Pace of change in the data center is ‘extraordinary’

Credited to Network World · networkworld.com · 6d ago

Photograph · Network World

Following its record-setting third quarter, HPE said it expects cumulative AI network orders for fiscal 2026 to top $3 billion, exceeding its previous forecasts. The company also raised its fiscal 2027 networking revenue target, projecting growth percentages in the high teens to low 20s. Specific to data center networking, HPE is estimating revenue to grow at roughly 50% CAGR through fiscal 2029.

“The pace of change in the data center is, quite frankly, extraordinary. AI has fundamentally reshaped how infrastructure is designed and deployed,” Rami Rahim, executive vice president, president and general manager of HPE Networking, told analysts and others at the annual HPE Networking Investor Day. “And the reason is simple: You can invest billions of dollars into GPUs, but if the network can’t keep up with the traffic and the performance demands, those resources are not fully utilized. So, as a result, networking has become a much more strategic part of the AI infrastructure stack.”

“We are taking standards-based Ethernet to a new layer of the data center stack for the first time, so it’s unlocking a massive TAM [total addressable market]. And if you think about the capacity requirements within the data center, scale-out is orders of magnitude greater than scale-across, and scale-up is orders of magnitude greater than scale-out,” said Rahim (pictured above at a 2025 event). “The capacity requirements are stunning in this area, so it unlocks a significant new TAM for us.”

Rahim pointed to HPE’s integrated AMD Helios package as an example of what an integrated Juniper/HPE will deliver. In December, HPE said it will support AMD’s Helios AI rack scale architecture with integrated scale-up Ethernet networking. The system basically will be a scale-up turnkey Ethernet package that will feature a purpose-built Juniper Networks switch based on Broadcom’s 102.4 Tbps. Tomahawk 6 network silicon and employ the Ultra Accelerator Link over Ethernet (UALoE) specification.

“We are bringing together Juniper’s networking technology with HPE’s compute, liquid cooling, rack scale engineering to deliver an integrated Ethernet-based scale-up solution. More importantly, it’s opening an entirely new addressable market in networking, either by selling our AMD Helios networking trades as part of an HPE integrated rack solution or independently to other solution providers. We believe Helios represents more than $1 billion networking opportunity over the next two years with networking trade orders having already exceeded $200 million,” Rahim said.

In addition, HPE announced this week that it won a $1.2 billion AMD Helios order from cloud infrastructure provider Vultr. “It’s HPE’s first order for the new integrated AMD Helios system, which features our new purpose-built HPE networking scale-up switch and software,” Rahim added.

Another more traditional networking area is also taking on increased importance in the AI networking arena, and that is routing, Rahim said: “As AI clusters grow beyond individual data centers, customers need to connect infrastructure across buildings, campuses, and geographies with enormous bandwidth, with low latency, and with high power efficiency. That’s where our PTX portfolio and our express silicon truly excel. The important point here is that this is not a new market we are trying to enter. We are applying decades of routing innovation, custom silicon expertise, and deep incumbency in some of the world’s largest cloud and service provider networks to a new wave of AI-driven demand. As a result, we expect our routing revenue to grow at low to high 20% CAGR from fiscal 2026 through 2029.”

Industry experts were generally enthusiastic about HPE’s strategy and presentation at the event.

“HPE’s market strategy spans a broad array of critical networking categories, including Data Center Networking, Routing, Campus & Branch, and Security. Addressing a diverse client base that ranges from traditional enterprise environments to hyperscalers and neocloud providers, the company projects significant market share gains across these key segments,” wrote Ron Westfall, vice president and practice leader for infrastructure and networking for HyperFRAME Research, in a blog post about the event. “Over the next three years, HPE expects to outpace overall industry growth rates in Data Center Networking, Routing, and Campus & Branch, positioning the unit as a high-margin growth engine for the broader enterprise.”

“We believe that HPE’s new long-term networking strategy is uniquely positioned to succeed because it pairs $800 million in expected post-Juniper cost synergies with a complete, full-stack AI architecture that tightly integrates scale-up and scale-out switching directly into high-density compute infrastructure,” Westfall wrote.

“As such, organizations must prioritize evaluating this proposition because single-vendor disaggregation eliminates the performance bottlenecks, latency risks, and integration complexity inherent in attempting to orchestrate disparate AI hardware, routing, and campus networks. Through unifying AI-native routing, Wi-Fi 7 campus modernization, and embedded zero-trust security under one operational umbrella, HPE provides enterprises and cloud providers with a future-proof foundation optimized for the demanding economics and scale of next-generation AI workloads,” Westfall wrote. Some other observations from HPE Networking Investors Day:

Beyond the enterprise: The campus and branch market is entering a widespread, multi-year modernization cycle tied to a Wi‑Fi 7 refresh, Rahim said. He expects to see a combined Aruba and Juniper Mist portfolio, larger installed base, and global go-to-market model. Campus and branch revenue is expected to grow at a high-single-digit percent CAGR through fiscal 2029.

Network protection: As networking and security converge, Rahim said HPE continues to embed security throughout the network by combining SASE, network access control, firewalls, identity-based policy enforcement, and AI-powered operations. HPE’s security revenue is expected to grow at a high-single-digit percent CAGR through fiscal 2029.

Partners: On Nov. 1, HPE will converge the Juniper and Aruba programs under one unified partner program, HPE Partner Ready Vantage. “Partners drive most of our networking business, and this milestone will enable all 60,000 HPE partners to sell the full networking portfolio through one unified partner program. Before the acquisition, only around 10% of HPE Aruba Networking and Juniper partners overlapped, and so this creates a significant cross-selling opportunity as we bring the combined HPE Networking portfolio to a much broader set of sellers and customers.”

Service growth: “Today, services represent about one-third of our HPE Networking business. Services deepen customer relationships, add higher margin recurring revenue, and improve the quality of our products driving durable earnings,” Rahim said. HPE sees an opportunity to bring that business model and the culture to Aruba across a much larger installed base. “So, by fiscal 2028, we expect to increase services attach rates by five percentage points, supporting both recurring revenue growth and margin expansion,” Rahim said.

Power & SitesUnited States

$165bn Project Jupiter: What Oracle’s force majeure notice means for AI data centre risk

Credited to Capacity Media · capacityglobal.com5d ago

Useful

Oracle has issued a force majeure notice over Project Jupiter, its flagship Stargate data centre campus in southern New Mexico, preserving the right to defer rent if the site misses its 2028 start because its power supply is not ready.

The notice, first reported by Bloomberg, went to a unit of Blue Owl Capital, which owns developer STACK Infrastructure. Oracle is not seeking to exit as anchor tenant. According to Bloomberg, if both sides accept that a force majeure event tied to power has occurred, Oracle could secure a three-year delay on rent once payments begin. It would still owe other costs in the interim, and the full lease term once rent starts.

Oracle maintains the project is on track. “Project Jupiter remains on our planned schedule,” the company said. Oracle spokesman Michael Egbert described such notices as “commonplace in developments of this scale”.

Flooding then reached the site on 30 September, although Oracle and STACK said in a joint statement that no critical infrastructure was damaged and the schedule is unchanged.

Why did Oracle issue a force majeure notice?

The data centre buildings are under construction. The power to run them is not yet approved.

Project Jupiter is designed to run on a private microgrid of Bloom Energy solid-oxide fuel cells supplying up to 2.45GW. That supply chain has three regulatory links, and none is complete.

The air permit. The New Mexico Environment Department has until 23 November to decide on the microgrid’s air quality permit, according to Source New Mexico. A state Supreme Court stay paused the proceeding in August and was lifted on 17 September. The original hearing officer recused himself in August, and no replacement had been named by late September. At a July scheduling meeting, a lawyer for microgrid developer Yucca Growth Infrastructure said the company stood to lose $325m a month from November without a decision.

The gas supply. Energy Transfer’s Green Chile pipeline, which would fuel the microgrid, still needs Federal Energy Regulatory Commission authorisation. Its start date has slipped by roughly six months to February 2027.

The land. The New Mexico State Land Office cancelled rights-of-way for parts of the pipeline route in March and rejected a request to reconsider in July, saying they were not in the state land trust’s best interests.

Who carries the risk?

In a 14 September statement, Oracle stressed that the campus and the microgrid are separate facilities on adjacent sites, run by separate companies. That structure, increasingly common as hyperscalers turn to on-site generation, spreads delivery risk across contracts that do not automatically move together.

The economics sit in the lease. According to a person familiar with the deal, quoted by Reuters, securing power is Oracle’s responsibility under the contract, Oracle covers the project’s debt costs and it cannot terminate the lease. Blue Owl, with about $3bn of equity in the project, earns a 9% development-stage yield that rises to around 11% once the campus is complete.

Invoking force majeure extends the period of lower development-stage rent. Blue Owl still collects full rent for the planned term, only later. The notice therefore delays the landlord’s return rather than shifting debt onto lenders, and it tests how far force majeure can stretch when the tenant itself holds the power obligation. Markets reacted anyway: Oracle shares fell more than 3% on the day, according to CNBC, and the Financial Times has reported the roughly $18bn construction loan trading at stressed levels.

Bloom Energy chief executive KR Sridhar told Bloomberg TV this week that its units are “fungible” and can be redirected if a project slips. A developer holding a single campus has no such option.

What does this mean for the data centre market?

Force majeure is now a negotiated clause, not boilerplate. Tenants on AI campuses will seek explicit protection against permitting delays on power and fuel. Developers will push for tighter definitions and shorter relief periods.

Behind-the-meter power moves risk rather than removing it. Bypassing the utility avoids interconnection queues, but adds air permits, pipeline approvals and emissions scrutiny, each with its own opponents. It also concentrates exposure to gas supply and prices, a risk Capacity examined after the Middle East supply shock.

Local opposition is now a balance-sheet issue. Permit challenges, land office decisions and state politics now feed directly into rent schedules and the cost of capital. Lenders have already started pricing community opposition into financing terms, and nearly $200bn of US data centre projects are stalled. New Mexico fits a wider US pattern that looks very different in Europe and Asia.

There is a fallback. Oracle executive vice president Mahesh Thiagarajan told Doña Ana County commissioners in late July that if the permit is unavailable, the project would turn to the grid, and Senator Martin Heinrich’s proposed GRID Savings Act would make hyperscale data centres fund their own grid upgrades. Either route adds time.

Dates to watch: the air permit decision by 23 November 2026, FERC authorisation of the Green Chile pipeline, and pipeline commissioning targeted for February 2027.

What is Project Jupiter? A $165bn AI data centre campus near Santa Teresa in Doña Ana County, New Mexico, developed by STACK Infrastructure and BorderPlex Digital Assets with Oracle as tenant. It is one of the flagship Stargate sites built for OpenAI.

What does Oracle’s force majeure notice do? It preserves Oracle’s right to pay lower development-stage rent for up to three years longer if power delays stop the campus opening in 2028. It does not end Oracle’s tenancy.

Is Project Jupiter delayed? Oracle says no. The 2028 target stands, but the microgrid air permit and the gas pipeline it depends on are both still awaiting approval.

Who is affected if Project Jupiter is delayed? Mainly Blue Owl Capital, owner of developer STACK Infrastructure, whose step-up to full rent would be postponed. According to a Reuters source, Oracle covers the project’s debt costs and cannot terminate the lease.

Metro Connect USA 2027

Metro Connect USA is the largest executive-level digital infrastructure event in the U.S. The only one of its kind, this 25-year-strong gathering is where decision makers come together to make deals happen.

Original · Capacity Media

Front Method

Capital · United States

Amazon’s $1B plan to combat data center backlash draws more backlash

Credited to Ars Technica · arstechnica.com · 5d ago

Photograph · Ars Technica

Amazon praised for ending NDAs but slammed for downplaying data center pollution.

On Friday, Amazon committed to donating more than $1 billion over the next five years to communities neighboring data centers.

In a press release, Amazon Web Services’ CEO Matt Garman said that communities would decide how to prioritize funds across “education, job training, energy affordability, water, and energy preservation, and local priorities.”

Laying out other community commitments, Amazon also promised to ensure data centers wouldn’t increase power bills or wipe out local water supplies. Regarding the latter, Amazon promised to be “water positive” by 2030, while claiming that 75 percent of projects have already met this goal.

Rolling out these commitments, Amazon seemed bent on convincing Americans that the swelling data center backlash is based on lies. Echoing Donald Trump’s claims that foreign disinformation campaigns are fueling widespread protests, Garman urged that calls for data moratoriums must end or else American frontier AI might fall behind China’s.

According to Garman, US rivals are trying to “trick us into slowing down,” and Amazon’s commitments are meant to reassure Americans that everything’s okay and that we should instead plow ahead, as Trump endlessly advocates.

However, advocates at Stand.Earth—a global nonprofit that’s been fighting for environmental protections in communities hosting data centers—told Ars that Amazon is the one being “misleading” by spreading “corporate propaganda.” Seemingly in crisis mode as billions of dollars in data center projects have been blocked nationwide, Amazon’s promised donations represent “a flailing attempt at damage control for the harm its data center build-out has already caused,” Stand.Earth said.

Most problematic, Stand.Earth suggested, is Amazon’s plan to back a power plant that experts expect will become the largest source of US climate pollution. Glaringly, Amazon’s data center commitments don’t mention the power plant and scarcely address pollution, apart from promising to exclusively rely on back-up generators that use the lowest possible emissions.

“For all the claims Amazon makes about investing in carbon-free energy, it conveniently leaves out that it is constructing the single-largest source of climate emissions in the United States in Pecos, Texas,” Stand.Earth said. “Communities will face 33 million tons of harmful air pollution every year—with the power plant using the same amount of energy needed to run more than 1.9 million homes.”

Additionally, Stand.Earth called out Amazon for claiming that most data center generators sparking pollution fears in communities typically sit “idle.” Although Amazon emphasized that most generators only run for “roughly 10 hours per year,” Stand.Earth cited cases where “Amazon data centers ran on diesel back-up power for days at a time.”

Some praise: Amazon stops using NDAs

At least one part of Amazon’s deal is promising, Stand.Earth suggested. The company ending its practice of using nondisclosure agreements when building new data centers will likely help make projects more transparent.

However, “many of the key concerns communities have are not addressed,” Stand.Earth found. For example, Amazon’s promises to replace all water used for data centers seemingly won’t address “the massive water footprint left by energy infrastructure powering its data centers” or “the potential for forever-chemical contamination from nascent cooling technologies,” Stand.Earth told Ars.

The total amount Amazon is investing also doesn’t seem like much compared to what the company has spent. In the press release, Garman said that Amazon contributed the same amount, more than $1 billion, to data center communities over the past three years.

“Compared to the $220 billion the company is investing in data centers in 2026 alone, Amazon’s commitment of $1 billion over five years for community support is a drop in the bucket,” Stand.Earth said. “It is difficult to read this statement as anything other than a plea from Amazon to continue exploiting communities at minimal cost.”