Nvidia Builds the Grid, the Software Moat and the Quantum Bridge
Published on 09/20/2026 at 12:41 | Editorial boerse-global.de
Nvidia spent the past week doing what it does best: planting flags in territory it does not yet own. On Thursday the chipmaker joined forces with Google and Emerald AI to launch the AI Energy Management Alliance, a vehicle for turning power-hungry AI data centers into flexible loads that can be folded into existing electricity grids rather than treated as a headache for utilities.
The move addresses a bottleneck that CEO Jensen Huang had already flagged publicly. Speaking at a Goldman Sachs conference on September 10, he described the global build-out of AI infrastructure as still being in its early innings, while naming available land, supply-chain capacity and — above all — electricity constraints as the chief obstacles to bringing new facilities online.
A benchmark debut and a quantum orchestration layer
Hardware and software advances landed alongside the energy initiative. The Vera Rubin NVL72 platform made its debut in the MLPerf Inference v6.1 industry benchmark, notching top scores in performance tests for AI inference workloads. On the software side, Nvidia widened its quantum computing toolkit: the open-source CUDA-Q platform gained CUDA-Q Logical, an orchestration layer that lets developers design and validate fault-tolerant applications for future quantum systems. The addition arrived on September 14, following a demonstration with Quantum Machines days earlier in which programs ran across classical processors and qubits at microsecond speeds.
That quantum push is not an isolated experiment. Tools such as the QUOPS benchmarking suite from Sandia National Laboratories are anchoring Nvidia deep inside scientific infrastructure, and the company's acquisition of direct access to the global developer community points the same way. What CUDA became for machine learning, CUDA-Q is meant to become for quantum informatics — and once a research team builds its workflow on that foundation, switching to a rival stack becomes a costly proposition. The combination of a software marketplace, compute infrastructure and a future quantum link forms an ecosystem that is hard for competitors to crack.
Should investors sell immediately? Or is it worth buying Nvidia?
Demand signals from the top
Underneath the technical announcements sits demand that shows no sign of cooling. Huang said Thursday that Nvidia expects to sell twice as many chips next year as it does in the current one, citing continued government investment in national AI programs and steady capacity expansion at large cloud data centers. That figure is a demand assessment, not a formal quarterly forecast.
The company also extended its media footprint, unveiling accelerated software development kits and dedicated microservices for media and entertainment workflows under the Nvidia AI for Media banner in Amsterdam on September 9.
Insiders trim, fundamentals hold
Financial strength gives management room to fund all of this at once. Revenue climbed to USD 96.2 billion in the second quarter of fiscal 2027, powered by a data center business that contributed USD 89 billion on its own. A 75 percent gross margin generates the cash flows needed to finance next-decade markets in parallel with today's core business.
News that CEO Jensen Huang and CFO Colette Kress sold shares to cover tax obligations drew brief attention, but such transactions under pre-arranged trading plans are routine and leave the fundamental picture untouched. The market, for its part, appears increasingly willing to credit management's projected 70 percent growth rate for fiscal 2028.
A stock near its high, and a regulatory backdrop
Nvidia shares closed Friday at EUR 193.10, leaving the stock 4.6 percent below its 52-week high of EUR 202.50 — a peak set in mid-May. The year-to-date gain stands at 20 percent, and some investors may be tempted to lock in profits. The broader case, though, rests on more than the current chip generation: Nvidia is writing the standards for the next decade of data processing, from AI training today to quantum computing tomorrow.
Not everything is smooth. Reuters reported Tuesday that Huang attended a state banquet hosted by US President Donald Trump for Chinese President Xi Jinping, and the company also faces antitrust scrutiny — the US Department of Justice is reportedly examining a licensing arrangement between Nvidia and AI chip startup Groq. None of that has deterred the company from pressing ahead with its plan to embed both hardware components and software control systems firmly in data centers worldwide.
Ad
Nvidia Stock: New Analysis - 20 September
Fresh Nvidia information released. What's the impact for investors? Our latest independent report examines recent figures and market trends.
