Nvidia Builds the Whole Pipeline: Substrate Partners, a $1.1 Billion Test Reservation and a 14-Day Data Center
Published on 10/21/2026 at 19:11 | Editorial boerse-global.de
Nvidia shares rose 2.0% on Thursday as the company rolled out a set of announcements that reach from the raw materials of chipmaking to the day a finished AI data center switches on. The stock also caught a lift from a broader rebound in AI-related names, with Reuters reporting that results from SpaceX and Super Micro Computer had improved risk appetite across the sector — a reminder that the day's gain cannot be pinned on any single company headline.
At the heart of the news is a supply-chain push spanning several production stages at once. Nvidia has enlisted IBIDEN, Samsung Electro-Mechanics, Unimicron, Amkor, KYEC and SPIL in an effort to expand capacity for substrates, advanced chip packaging and testing. Substrate capacity is targeted to reach 2.5 times current levels within three years.
The commitments go well beyond making more chips. Amkor aims to lift its capacity for advanced packaging by more than 300% by the end of 2028, while KYEC is to build test facilities in the United States. Nvidia is backing that build-out with a $1.1 billion advance payment to reserve capacity — a sign the company is putting money behind the reservations rather than simply announcing them through partners. These are expansion targets and planned facilities, not production capacity that is already online.
Google Cloud Sets a Date
A second strand concerns how quickly that hardware reaches customers. Nvidia and Google Cloud said the A5X instances, built on Nvidia Vera Rubin, will become generally available next month. Google Cloud plans to scale the corresponding infrastructure across multiple locations to as many as 960,000 graphics processors.
Should investors sell immediately? Or is it worth buying Nvidia?
Two things are worth keeping apart here: general availability is announced, while the 960,000-GPU figure describes a planned expansion rather than installed capacity. No revenue for Nvidia can be derived from the build-out plan as yet.
From Six Months to 14 Days
On the physical side, Nvidia used GTC Berlin to unveil DSX POD, a modular reference design for complete Vera-Rubin AI factories. Partners are to build, integrate and validate the systems before delivery. According to the company, setup at a prepared site could shrink from six months to as little as 14 days.
That caveat does the heavy lifting. The claim is a conditional performance promise tied to a location where the necessary power and services are already in place — not a blanket commitment for any data center project. The approach is about speeding deployment through prefabricated modules, not about shortening the total duration of an arbitrary build.
Nvidia at a turning point? This analysis reveals what investors need to know now.
The strands fit together: Google Cloud describes planned access to Rubin compute, the supply-chain partners are to expand the capacity that feeds it, and pre-integrated systems are meant to cut the time needed to bring it online. What remains is the implementation gap. Capacity expansion stretches over several years, and faster installation depends on prepared sites. Thursday's news offers concrete targets and the conditions for accelerated delivery — not proof that any of them have been met.
Ad
Nvidia Stock: New Analysis - 21 October
Fresh Nvidia information released. What's the impact for investors? Our latest independent report examines recent figures and market trends.
