NVIDIA's $2 Billion Bet on Nebius Signals a New Phase in the AI Infrastructure Arms Race

On March 11, 2026, NVIDIA announced a $2 billion strategic investment in Nebius Group, an Amsterdam-headquartered AI cloud company that rose from the ashes of Russia's Yandex empire. The deal—NVIDIA's third $2 billion investment in just ten days—is more than a financial transaction. It is a clear signal that the chipmaker is moving aggressively to lock in the next generation of AI infrastructure partners as the race to build gigawatt-scale "AI factories" accelerates globally.

The Deal: What NVIDIA Is Buying

According to an SEC filing, NVIDIA agreed to purchase shares representing approximately 8.3% of Nebius at $94.94 per share [1]. The investment comes with far more than capital: Nebius will receive early access to NVIDIA's next-generation Rubin GPU platform, Vera CPUs, and BlueField storage systems [2]. The two companies will collaborate on AI infrastructure deployment, fleet management, inference optimization, and what NVIDIA calls "AI factory" design and support [3].

The partnership's ambition is staggering in scale. Together, NVIDIA and Nebius plan to deploy more than 5 gigawatts of NVIDIA systems by the end of 2030 [2]—enough power to run a small country. For context, a single gigawatt can power roughly 750,000 homes.

Nebius shares surged 16% on the announcement, pushing the company's market capitalization to roughly $28 billion [4].

From Yandex to Nebius: The Phoenix Story

Nebius Group's origin story reads like a geopolitical thriller. The company's predecessor, Yandex N.V., was the Dutch parent entity of Russia's dominant search engine and technology conglomerate, co-founded by Arkady Volozh in the late 1990s [5]. At its peak, Yandex was valued at $30 billion and was often called "the Google of Russia."

When Russia invaded Ukraine in February 2022, Yandex N.V.'s NASDAQ-listed securities were suspended from trading amid international sanctions [5]. The company spent the next two years unwinding its Russian operations, ultimately selling all Russian assets to a consortium of domestic investors in July 2024. What remained was a collection of international businesses—AI infrastructure, autonomous vehicle technology, cloud computing—staffed by over 1,000 former Yandex employees [5][6].

Yandex N.V. formally renamed itself Nebius Group N.V. and resumed NASDAQ trading in October 2024 under the ticker NBIS [5]. Volozh, a Kazakh-born mathematician who had run Yandex for 25 years, returned as CEO with a new mission: building the infrastructure layer for the global AI revolution.

"Nebius has been built for AI since day one—not adapted from a general-purpose cloud," Volozh said in a recent interview [7]. The distinction matters. While traditional hyperscalers like Amazon Web Services and Microsoft Azure offer AI capabilities atop their general-purpose platforms, Nebius and its neocloud peers are architecting their entire stack around GPU-intensive workloads from the ground up.

The Neocloud Phenomenon

The NVIDIA-Nebius deal is the latest manifestation of a structural shift in cloud computing. A new class of providers—dubbed "neoclouds"—has emerged to challenge the hyperscaler giants by specializing exclusively in AI and high-performance computing workloads [8].

CoreWeave, the sector's dominant player, has eclipsed $5 billion in annual revenue faster than any other cloud platform in history, began trading on NASDAQ in late March 2025, and currently trades around $138-$139 per share [8]. Lambda Labs, another key competitor, offers both cloud GPUs and on-premises GPU clusters. Other players include Crusoe Cloud, Nscale, Vultr, and Fluidstack [8].

The neoclouds compete on a simple value proposition: specialization and price. According to industry analysis, an hour of DGX H100 compute costs approximately $98 from hyperscaler platforms, while neoclouds deliver the same resource at roughly $34 per hour—a 66% price reduction [8]. This pricing advantage, combined with purpose-built infrastructure optimized for AI training and inference, has attracted a flood of customers and capital.

Synergy Research Group projects that neocloud revenues, currently growing at over 200% per year, will reach $180 billion by 2030 [8].

Source: Futurum Group / Company Earnings Reports

Data as of Mar 12, 2026CSV

NVIDIA's Investment Blitz

The Nebius investment does not exist in isolation. In a breathtaking week-and-a-half of capital deployment, NVIDIA invested $2 billion each in photonics companies Lumentum and Coherent on March 2—bringing its combined strategic investments in that period to $6 billion [9]. The Lumentum and Coherent deals targeted a different but related bottleneck: the advanced optical interconnect technology required to link GPUs in next-generation data centers at scale [9].

NVIDIA's broader investment portfolio tells an even larger story. By the end of 2025, the chipmaker's public equity portfolio exceeded $13 billion [10]. Key positions include a multi-billion dollar stake in CoreWeave, and the company has announced a staggering commitment to invest up to $100 billion in OpenAI as each gigawatt of deployment is activated—with the first gigawatt expected on NVIDIA's Vera Rubin platform in the second half of 2026 [10].

NVIDIA has also forged a multiyear, multigenerational strategic partnership with Meta, spanning on-premises, cloud, and AI infrastructure, enabling deployment of millions of Blackwell and Rubin GPUs [10]. Europe is also in focus: NVIDIA has partnered with multiple European governments and firms to build sovereign AI infrastructure across the continent [10].

The pattern is unmistakable. NVIDIA is no longer merely selling chips—it is strategically investing across the entire AI value chain, from silicon and optics to cloud infrastructure and application layers, creating a web of dependencies that reinforces demand for its hardware.

Nebius's Expansion Plans

The NVIDIA investment arrives at a pivotal moment for Nebius. In the days leading up to the announcement, the Independence, Missouri city council voted on March 3, 2026 to approve a Chapter 100 industrial development incentive plan for Nebius's planned AI factory campus [11]. The project is massive: a 1.2-gigawatt facility on approximately 400 acres, representing a $6.6 billion total investment [11].

The Independence campus would be Nebius's largest facility in the United States, expected to create approximately 1,200 construction jobs and 130 permanent positions. Under the approved agreement, Nebius will make Payments in Lieu of Taxes (PILOT) projected at over $650 million to the city, local school districts, and other jurisdictions over 20 years [11]. Power delivery is slated for the second half of 2026, connecting to Independence Power & Light, the city's municipal utility [11].

This is not Nebius's only expansion. The company is building a 300-megawatt cloud campus in New Jersey—currently its largest active project—and announced plans in February 2026 to build an AI factory in Birmingham, Alabama [5]. Its data center network also spans the U.K., France, Finland, and Iceland [12].

In February 2026, Nebius also acquired Tavily, an agentic search company, for approximately $400 million, signaling ambitions that extend beyond raw infrastructure into the software and applications layer [5].

Financial Trajectory: Explosive Growth, Deep Spending

Nebius's Q4 2025 financial results, reported on February 12, 2026, reveal a company growing at breakneck speed while burning through capital at an equally aggressive pace [13].

Group revenue reached $228 million in Q4, representing 547% year-over-year growth. The core AI cloud business was even more explosive, with revenue surging 830% year-over-year and 63% sequentially to $214.2 million—roughly 94% of total revenue [13]. Annualized run-rate revenue reached $1.25 billion by year-end 2025, exceeding internal guidance [13].

The company posted positive adjusted EBITDA of $15 million at the group level, with the AI cloud segment delivering $51.8 million in adjusted EBITDA at a 24% margin [13]. However, the net loss was $249.6 million, reflecting massive infrastructure buildout costs. Capital expenditures surged to $2.06 billion in Q4 alone—up from $417.5 million a year earlier—as Nebius acquired AI processors and expanded data centers [13].

The company held $3.7 billion in cash and equivalents at year-end, bolstered by $834 million in operating cash flow during the quarter [13]. The NVIDIA investment adds another $2 billion to Nebius's war chest, providing substantial runway for its gigawatt-scale ambitions.

Source: Nebius Group Q4 2025 Earnings Report

Data as of Feb 12, 2026CSV

The $690 Billion Infrastructure Sprint

Nebius and NVIDIA's partnership unfolds against the backdrop of an unprecedented global investment surge in AI infrastructure. The five largest U.S. cloud and AI infrastructure providers—Microsoft, Alphabet, Amazon, Meta, and Oracle—have collectively committed to spending between $660 billion and $690 billion on capital expenditure in 2026, nearly doubling 2025 levels [14]. Amazon alone has projected $200 billion in 2026 spending, up from $131 billion in 2025, while Google has estimated between $175 billion and $185 billion [14].

Gartner forecasts worldwide AI spending will total $2.52 trillion in 2026, a 44% increase year-over-year, with AI infrastructure adding $401 billion in spending as technology providers build out AI foundations [15]. AI-optimized Infrastructure-as-a-Service spending alone is projected to reach $37.5 billion in 2026, roughly doubling from $18.3 billion in 2025 [16].

These figures dwarf anything seen in previous technology cycles. The AI infrastructure buildout has become the single largest capital deployment program in the history of the technology industry.

Strategic Implications and Risks

The NVIDIA-Nebius deal raises several important strategic questions.

Concentration risk. NVIDIA is constructing an investment portfolio that creates deep financial ties with its own customers and partners. If the AI infrastructure boom slows or demand disappoints, NVIDIA faces exposure not only through reduced chip sales but through equity losses in companies like Nebius, CoreWeave, and potentially OpenAI. NVIDIA's stock, trading at approximately $186 with a market cap near $4.5 trillion [17], already prices in sustained AI hypergrowth.

Geopolitical dimensions. Nebius's Yandex origins remain a subject of scrutiny. While the company has fully divested its Russian assets, its founder Arkady Volozh was briefly subject to EU sanctions (later removed), and some investors may remain cautious about the company's heritage [5][6]. The fact that NVIDIA—an American company subject to U.S. export controls on AI chips—is investing billions in a company founded by a Russian-born entrepreneur speaks to NVIDIA's confidence in Nebius's clean break from its past.

Valuation concerns. Some analysts have questioned whether Nebius's $28 billion valuation is justified for a company that reported only $228 million in quarterly revenue and a $250 million net loss [18]. The bearish case rests on the assumption that neocloud margins will compress as hyperscalers improve their own GPU offerings, and that Nebius's rapid growth may not be sustainable as competition intensifies.

Energy and environmental footprint. The aggregate power requirements are staggering. Nebius alone plans to deploy 5 gigawatts of NVIDIA systems by 2030 [2]. Across the industry, hundreds of gigawatts of new power capacity will be needed—raising serious questions about grid capacity, renewable energy sourcing, and the carbon footprint of the AI revolution. Nebius has touted its Independence campus's closed-loop cooling system and commitment to minimizing water consumption [11], but the sheer scale of energy demand remains a societal concern.

What Comes Next

NVIDIA's $2 billion investment in Nebius is a bet that the AI infrastructure market is still in its early innings. With hyperscaler capex approaching $700 billion, neocloud revenues growing at 200% annually, and demand for specialized AI compute showing no signs of abating, the strategic logic is clear.

For Nebius, the deal provides not just capital but the imprimatur of the most important company in the AI ecosystem—and guaranteed access to NVIDIA's most advanced chips before competitors. For NVIDIA, it secures another major customer-partner committed to deploying its technology at gigawatt scale.

The broader question is whether this investment frenzy represents a durable structural shift or an unsustainable bubble. History offers cautionary parallels—the fiber-optic buildout of the late 1990s created enormous overcapacity before demand eventually caught up. But the executives writing the checks insist this time is different. Whether they're right will determine the fortunes of companies, communities, and an entire industry for years to come.