Nvidia picked Schneider Electric to build 14-day AI data centers as Barclays warns peak pricing may end

Nvidia named Schneider a factory builder of its 14-day DSX POD data centers on October 21, the same day Schneider launched its RapidScale design for Vera Rubin, while Barclays warned the day before that peak pricing for power-gear makers is ending. Volume and price now point opposite ways on the same company.

Richard TangRichard Tangfour-minute read
Share
Nvidia chief executive Jensen Huang receives the National Medal of Technology and Innovation from President Trump in Washington on October 8, 2026
1 / 6Slide 1 of 6
Nvidia chief executive Jensen Huang receives the National Medal of Technology and Innovation, Washington, October 8, 2026. On October 21 his company named Schneider Electric one of six builders of its 14-day DSX POD data centers.

Nvidia−4.19% — Nvidia, down 4.19 percent today named Schneider Electric−0.78% — Schneider Electric, down 0.78 percent today on October 21 as one of six companies, alongside Dell Technologies−6.94% — Dell Technologies, down 6.94 percent today, HPE−4.02% — HPE, down 4.02 percent today, Lenovo, Bull and Vertiv−6.78% — Vertiv, down 6.78 percent today, that factory-build its new DSX POD, a complete AI data center assembled off-site and installed in as little as 14 days at a prepared site against about six months today, with first deliveries early 2027 35. The same day, Schneider launched RapidScale, its own prefabricated reference design built for Nvidia's Vera Rubin platform 6.

Three days earlier it had printed a record quarter: €11.5 billion of revenue, up 17% organically, data-center demand up triple-digit, and a second upgrade to its 2026 target in the year 1. That formalizes the volume story. The money argument ran the other way a day before the keynote.

Barclays says the peak price is ending

Barclays−2.06% — Barclays, down 2.06 percent today, led by analyst George Featherstone, warned on October 20 that "peak cycle economics could soon be normalized" for power-gear makers: AI buyers are pressing on price, and new capacity is arriving in exactly the two categories where lead times had spiked, power and medium-voltage equipment 2. Reported talks between Nscale, a compute-rental firm known as a neocloud, and Anthropic$350B — Anthropic, private, latest valuation $350B price capacity below CoreWeave's−6.61% — CoreWeave, down 6.61 percent today September benchmarks per gigawatt-year 2.

Paris had already voted. On October 20 Schneider fell 3.5%, the steepest drop in the CAC 40, as credit fear around neoclouds swept the equipment makers; CoreWeave, the largest of them, has seen its bond yields near 15% 7. The stock sits 22% below its August peak, Siemens Energy−1.75% — Siemens Energy, down 1.75 percent today is 31% off its April high, and ABB has given back about the same since June, figures Barclays itself reports 2.

The AI-gear drawdowns are deep: 22% to 31% from recent peaks

-0%-10%-20%-30%-40%Siemens Energy-31%Schneider Electric-22%ABB-22%
Data
Value
Siemens Energy-31%
Schneider Electric-22%
ABB-22%
Share-price declines from recent peaks as reported by Barclays: Siemens Energy since its April high, Schneider Electric since its August peak, ABB roughly the same since June; all as of October 20, 2026.2

Volume is not the open question

The demand side keeps confirming itself. Schneider's Systems business, the line prefabricated data centers are sold through, grew 22% organically in the quarter, faster than any other part of the company, and the group ended with a record backlog and what management calls increasing visibility into 2027 and 2028 1.

Nvidia sees the same build-out from above: its supply and capacity commitments, agreements to buy parts and book factory space, hit $279 billion at the end of July, $160 billion of it added in a single quarter, per its own 10-Q 3. "Demand is extraordinary. Capacity is the constraint," said Jensen Huang 3. When the buyer at the top of the chain commits like that, the volume link is locked.

Price and watts per token are the open links

Price is where the evidence is thinner and the risk sits. Nothing in this week's releases discloses what a DSX POD sells for or what Schneider's economics on the Nvidia arrangement are, so the question is whether factory standardization protects margin or turns power gear into a commodity box. The test arrives with POD pricing and the early 2027 deliveries 35.

Nvidia's own software adds a third force, and it pushes against volume of gear rather than with it. DSX trials squeezed 24% to 49% more token throughput from a fixed power budget, and in a grid-event trial a cluster cut its power draw by up to 40% while critical work kept running 34. More compute per megawatt means less power equipment per unit of AI, which is why efficiency, not only buyer pressure, can normalize what Schneider charges.

The stock is refusing to pick a side

So the volume story and the price story point opposite ways on the same company, and the 22% slide since August is the market declining to pick a side 2. The chain runs from standardized factory-built data centers to more AI installations to more Schneider Systems revenue; the price hop in between is unproven, and it is the one the next two quarters will settle.

Standardized volume growing faster than price falls makes the drawdown an entry; buyers paying commodity rates for factory-built boxes makes Barclays right that normalization has started.

Note: no disclosed figures exist for Schneider's POD economics or the terms of the Nvidia arrangement; the verdict above rests on reported volume, drawdowns and DSX efficiency results, not undisclosed pricing.

Get Data Center by email, free

Who is building AI's data centers, and who is paying for them. In our daily email, every weekday.

Loading BCS

Reader comments

0 comments

    More in Data Center

    All in this topic →

    Data Center

    Trane Technologies' $12.5 billion backlog waits on suppliers; CEO sees margin relief in six to nine months

    At the Baird conference on November 12, Dave Regnery put the constraint on Trane's $12.5 billion backlog in its suppliers' plants rather than its own, with relief expected in six to nine months. With detailed 2027 guidance deferred to fourth-quarter results, that timeline is the only forward margin guide investors hold.

    Data Center

    Legrand puts data centers at 32% of revenue just as Thailand asks 49 data-center builds to pause

    The French group's Southeast Asia chief put numbers on its AI exposure: data centers are about a third of revenue and close to €3 billion a year, just as Thailand and Singapore start rationing the power and water new builds need.

    Data Center

    MARA sold 13,597 bitcoin for $1.1 billion, cut debt 37%; two AI leases by Dec. 31 decide the stock

    Traders blamed the miner for bitcoin's crash to $62,006, but its $1.1 billion coin sale is a balance-sheet story: lenders repaid in full, a bitcoin treasury converting into AI data centers, and a December 31 lease deadline that decides the stock.

    Data Center

    Rackspace signed a $380 million Together AI deal to buy Nvidia GB300 racks; no lender has committed

    The definitive agreement looks like a win for Rackspace, but the company said the GB300 hardware will be financed separately and no lender commitment exists yet. Until that loan closes, the $610 million AI book and the megawatt target behind it are conditions to be met, not numbers to book.

    Data Center

    Everpure CEO and co-founder trusts sold $17 million eight days before Wednesday's earnings report

    Forms filed November 12 show $17.1 million of insider sales on November 10, at $98.88 to $104.75, about a third below the October peak and eight days before the November 18 report. The report has to reconcile a 50 percent growth call from Guggenheim with a company that burned $136 million of operating cash last quarter.

    Data Center

    Gartner sees $4 trillion of AI spending in 2027, but a positive AI return now costs 18 times more

    Gartner's own release credits part of the server bill to memory-price inflation, and its analysts told chief information officers this week that positive returns now cost 18 times more to reach than two years ago. Anyone marking AI vendor revenue against the first number has to price the second.

    Subscribe to TheInference for free