Nebius Launches $3.75 Billion Debt Offering Following Meta Partnership

In the span of seven days, Nebius Group has gone from a company most retail investors had never heard of to one raising capital at a pace that rivals the hyperscalers it serves. On March 17, the Amsterdam-based AI infrastructure firm announced a $3.75 billion convertible notes offering — its second multi-billion-dollar debt raise in six months — just one day after disclosing a $27 billion deal with Meta Platforms and less than a week after Nvidia committed $2 billion in equity [1][2][3].

The offering consists of two tranches: $2 billion in convertible senior notes due 2031 and $1.75 billion due 2033, with an option for initial purchasers to buy an additional $562.5 million [4]. Proceeds will fund data center expansion and the purchase of next-generation GPU hardware. For a company that reported just $228 million in quarterly revenue three months ago, the scale of capital being deployed is staggering — and telling about the current state of the AI arms race.

The Week That Changed Nebius

The cascade of announcements began on March 11, when Nvidia revealed a $2 billion strategic investment in Nebius at $94.94 per share, acquiring roughly 8.3% of the company [3]. The deal was more than financial: it formalized a partnership in which the two companies will collaborate on AI infrastructure deployment, fleet management, and the design of next-generation AI factories. Nebius committed to deploying more than 5 gigawatts of Nvidia systems by the end of 2030.

Five days later, Meta disclosed a five-year agreement worth up to $27 billion, making it one of the largest AI infrastructure contracts ever signed [2]. The deal includes $12 billion of dedicated capacity across multiple locations — based on one of the first large-scale deployments of Nvidia's Vera Rubin platform, with delivery starting in early 2027 — plus up to $15 billion in additional available compute capacity. It was a ninefold expansion of an initial $3 billion agreement signed in November 2025, bringing Meta's total contracted spend with Nebius to $30 billion [5].

Nebius shares surged 14% on the Meta news, pushing the stock above $130, up roughly 550% from the $20 range where it traded when it relisted on Nasdaq in October 2024 [6].

From Russia's Google to Europe's AI Backbone

Nebius's origin story is one of the most unusual in tech. The company is the successor to Yandex N.V., the Dutch holding company of Russia's dominant search engine and technology platform, founded in 1997 by Arkady Volozh and Ilya Segalovich [7]. Yandex went public on Nasdaq in 2011, raising $1.3 billion.

But Russia's invasion of Ukraine in February 2022 upended everything. Nasdaq suspended trading in Yandex securities, and the company found itself caught between Western sanctions and Russian government pressure. In July 2024, Yandex N.V. sold all of its Russian assets to a consortium of Russian investors. What remained was an Amsterdam-based shell with roughly 1,300 engineers, a portfolio of AI intellectual property, and several international subsidiaries [7].

Rebranded as Nebius Group, the company resumed Nasdaq trading in October 2024 and pivoted entirely toward becoming a Western-focused AI infrastructure provider. Volozh, a Kazakhstan-born mathematician who had built Yandex into a $30 billion company, now had to rebuild from near-zero revenue with a fundamentally different business model.

The Neocloud Model

Nebius operates as what the industry calls a "neocloud" — a cloud provider built exclusively for AI workloads rather than general-purpose computing. Unlike Amazon Web Services or Microsoft Azure, which offer everything from web hosting to database management, neoclouds focus on the GPU-dense, power-hungry infrastructure required for training and running large AI models [8].

The model has proven extraordinarily attractive to hyperscalers who cannot build data centers fast enough to meet their own AI ambitions. Even companies like Microsoft and Meta, which collectively plan to spend hundreds of billions on AI infrastructure in 2026 alone, need external partners to bridge the gap between demand and their own construction timelines [9].

Nebius's competitive advantage lies in its engineering heritage. The team that built Yandex's internal AI infrastructure — training models at scale years before the current generative AI boom — now applies that expertise to building AI-optimized data centers for the world's largest tech companies.

Source: FRED / U.S. Treasury

Data as of Mar 13, 2026CSV

A Balance Sheet Built on Convertible Debt

The March 17 offering is Nebius's second major capital raise in six months. In September 2025, the company raised $3.75 billion through a combination of $2.75 billion in convertible notes and roughly $1 billion in new shares at $92.50 apiece [10]. That earlier offering was split into two series: $1.375 billion of 1.00% notes due 2030 and $1.375 billion of 2.75% notes due 2032, with conversion prices at a 50% premium to the offering price [10].

The new March 2026 offering follows the same convertible playbook but at even larger scale. When combined with the September raise, Nebius will have issued more than $6.5 billion in convertible debt within half a year — a volume that would be remarkable for any company, let alone one with an annualized revenue run rate of roughly $1.2 billion as of year-end 2025 [11].

The strategy carries clear logic: convertible notes offer lower coupon rates than traditional debt, and if the stock continues to rise, the notes convert into equity rather than requiring cash repayment. But it also means significant dilution for existing shareholders if the conversion prices are reached, and substantial debt servicing obligations if they are not.

Nebius closed 2025 with $3.7 billion in cash on its balance sheet [11]. With the new raise, it will have the firepower to fund the massive capital expenditures required to fulfill its Meta, Microsoft, and Nvidia commitments.

The Contract Backlog

The sheer scale of Nebius's contracted revenue is what distinguishes it from a typical high-growth startup burning through cash. Consider the roster:

• Microsoft: Up to $19.4 billion through 2031 for GPU infrastructure at Nebius's Vineland, New Jersey data center, supporting Azure demand [12]
• Meta: Up to $30 billion over five years ($3 billion initial deal plus $27 billion expansion), with deliveries starting early 2027 [2][5]
• Nvidia: $2 billion equity investment plus a strategic partnership spanning infrastructure deployment and AI factory design [3]

Management has reaffirmed a 2026 annualized run rate target of $7 billion to $9 billion, with more than 50% of that capacity already under long-term contracts [8]. If achieved, it would represent a roughly 7x increase from the $1.2 billion ARR reported at year-end 2025.

The Neocloud Arms Race

Nebius is not operating in a vacuum. The neocloud sector has become one of the hottest segments in technology, with revenues passing $5 billion in Q2 2025, growing at 205% year-over-year, and forecast to reach $180 billion by 2030 [13].

CoreWeave, the sector's first mover and largest player with a market capitalization near $71 billion, has built a $55.6 billion backlog and reported nearly $1 billion in quarterly revenue [13]. Lambda Labs, backed by a $1.5 billion funding round, has its own multibillion-dollar deal with Microsoft [13].

But Nebius trades at a meaningful discount to CoreWeave: roughly 3.5x its projected ARR compared to CoreWeave's 6.4x price-to-sales ratio [14]. Bulls argue this discount reflects the market's lingering uncertainty about Nebius's Yandex heritage and execution risk, while bears worry that the neocloud model is structurally dependent on continued hyperscaler spending that could slow if the AI investment cycle turns.

Source: Company filings and press releases

Data as of Mar 17, 2026CSV

The Macro Backdrop: War, Rates, and AI Exceptionalism

Nebius's capital raise comes at a fraught moment for global markets. The U.S.-Israeli war on Iran, now in its third week, has sent oil prices surging past $95 per barrel and pushed 10-year Treasury yields above 4.28% as inflation fears mount [15]. Mortgage rates have hit 2026 highs, and the Federal Reserve is expected to hold rates steady at its March meeting.

Yet AI infrastructure spending appears to exist in its own economic universe. Goldman Sachs projects that AI companies will invest more than $500 billion in 2026, with hyperscalers alone planning nearly $700 billion in data center projects [9]. Amazon projects $200 billion in 2026 capex, while Google estimates $175-185 billion [9]. The sheer momentum of these commitments — driven by competitive pressure to deploy AI models rather than traditional return-on-investment calculations — has created a capital spending supercycle that appears, at least for now, immune to the broader economic uncertainty.

For Nebius, this dynamic is both opportunity and risk. The company's contracted backlog provides unusual revenue visibility, but the infrastructure it is building is expensive, illiquid, and specialized. If AI spending slows or the technology fails to generate the returns hyperscalers expect, neoclouds could find themselves holding billions in GPU hardware with declining resale value.

What the Debt Markets Are Saying

The fact that Nebius can raise $3.75 billion in convertible debt days after its Meta announcement speaks to the depth of investor appetite for AI infrastructure exposure. Convertible notes are particularly popular with hedge funds and institutional investors who can hedge the equity component while collecting the coupon — essentially a way to bet on the AI buildout with a floor on losses.

But the pace of issuance warrants scrutiny. Nebius will have roughly $6.5 billion in convertible debt outstanding against trailing twelve-month revenue of approximately $700-800 million. Even with the contracted backlog, the company must execute flawlessly on data center construction, GPU procurement, and customer delivery timelines to justify these obligations.

The 2031 and 2033 maturities give Nebius time, but the conversion feature means existing shareholders are implicitly betting that the stock will be substantially higher in five to seven years — or accepting significant dilution.

The Volozh Factor

At the center of this transformation is Arkady Volozh, whose career arc from building Russia's most important technology company to leading an Amsterdam-based AI infrastructure startup is unlike anything in the industry [7]. Volozh successfully navigated Nebius through the most perilous corporate restructuring in recent tech history — splitting a $30 billion company across geopolitical lines while retaining its engineering core.

His credibility with partners like Microsoft, Meta, and Nvidia has been essential. These are not arms-length vendor relationships; they are multi-year, multi-billion-dollar commitments that require deep trust in the counterparty's ability to deliver. That three of the world's largest technology companies have simultaneously placed massive bets on Nebius reflects their confidence in Volozh's team as much as the neocloud model itself.

The Road Ahead

Nebius's 2026 guidance of $7-9 billion in ARR, if achieved, would place it firmly in the top tier of AI infrastructure providers globally. But the gap between contracted commitments and delivered revenue remains enormous. The company must secure power supply, build or lease data center capacity, procure millions of dollars in GPU hardware, and manage the complex logistics of deploying infrastructure across multiple geographies — all while servicing billions in debt.

The broader neocloud sector, projected to nearly quadruple global data center capacity by 2030, faces its own structural questions: whether GPU-dense facilities built for today's AI architectures will remain relevant as models evolve, whether power grid constraints will bottleneck expansion, and whether the hyperscalers will eventually bring this capacity in-house once their own construction catches up.

For now, though, the market's verdict is clear. In eighteen months, Nebius has gone from a post-sanctions orphan trading at $20 to a company with $50 billion in contracted revenue, backing from three of tech's most powerful players, and the ability to raise billions in debt on a single day's notice. Whether that trajectory represents the emergence of a genuinely new category of infrastructure company — or the peak of an AI spending cycle that has yet to face a real downturn — is the multi-billion-dollar question investors are now being asked to answer.