AI Upstart Nebius (NBIS.US) Sees Cloud Revenue Surge 500%, Pre-Market Soars

Publicado a 2026-08-13Actualizado a 2026-08-13

Resumen

Driven by strong demand for AI computing power, Nebius (NBIS.US) saw its second-quarter cloud business sales surge 514% year-over-year.

Driven by robust demand for AI computing power, Nebius (NBIS.US) saw its second-quarter cloud services sales soar 514% year-over-year. The financial report shows that revenue from its AI cloud business, which constitutes the vast majority of its operations, reached $575 million for the quarter. Total revenue for the quarter exceeded $582 million, a 454% increase year-over-year, surpassing analyst expectations of $557 million. The loss per share was $0.68, exceeding the expected loss of $0.18.

In pre-market trading on Wednesday, the company's stock rose approximately 16%.

The company also reaffirmed its full-year 2026 outlook, stating that the accelerating demand for AI computing power is helping it secure larger, more profitable customer contracts.

In its first-quarter earnings report released in May, Nebius had reiterated its full-year 2026 expectations, forecasting annualized revenue of $7 to $9 billion, group revenue of $3 to $3.4 billion, and an adjusted EBITDA margin for the group of approximately 40%. The market consensus was for revenue of $3.38 billion.

CEO Arkady Volozh stated, "Everything we set out to do this quarter, we have done. In most cases, we have done more."

The announcement noted that the value of contracts secured this quarter quadrupled compared to the previous quarter, including four agreements with an average value each exceeding $1 billion.

The company indicated that pricing strengthened during the quarter, benefiting from demand for next-generation AI chips and higher rates for older generation GPUs. Approximately 70% of the contracts signed during this period included customer prepayments, covering 50% to 60% of the related capital expenditures.

Nebius stated that it finalized four landmark AI cloud deals during the quarter, each with an average total contract value exceeding $1 billion, resulting in a total contract value nearly four times higher than the previous quarter.

Volozh pointed out that new pricing initiatives launched at the beginning of the third quarter, such as initial auctions and short-term computing power trading, are showing promising prospects. He added that the company sees price opportunities in the range of $40 to $50 million per megawatt and signed its first such contract this week.

Competitor CoreWeave Inc., propelled by the ongoing AI spending boom, on Tuesday forecast its third-quarter sales to be above expectations.

Capital Expenditure Payback Period Shortens

Nebius is one of several so-called "new cloud" companies renting out computing power from data centers, capitalizing on the immense global demand for infrastructure capable of handling AI workloads. The company was spun off from Russian internet giant Yandex in 2024 and has established partnership agreements with Nvidia as well as major data center operators like Microsoft and Meta Platforms Inc.

Building infrastructure for AI workloads is an expensive business. Nebius stated in the announcement that in July, it secured $775 million in financing collateralized by assets including GPUs. In the second quarter, the company spent approximately $5.7 billion on purchasing chips, equipment, and expanding data centers.

Volozh said that, overall, the expected payback period for the capital expenditures and related operating costs associated with second-quarter deals is 1 year and 10 months, lower than the previously anticipated 2 to 3 years.

Nebius noted that it will continue to increase capital expenditures this year and plans to utilize diversified financing channels.

Lecturas Relacionadas

A 22-Year-Old Mathematical Puzzle, Solved by a Union Hospital Intern?

In an astonishing development, a longstanding mathematical problem known as the Crouzeix conjecture, which had challenged experts in numerical linear algebra for 22 years, appears to have been solved by Shanmu Jin, a neurosurgery resident and postdoctoral researcher at Peking Union Medical College Hospital. With no formal advanced mathematical training—his background is in geology and medicine—Jin relied on self-study and, crucially, the AI model GPT-5.6. The conjecture, proposed by mathematician Michel Crouzeix in 2004, concerns a fundamental constant (speculated to be 2) bounding the relationship between matrix norms and polynomial values over numerical ranges. It is critical for applications in matrix function analysis and numerical methods. Despite dedicated efforts by leading mathematicians, the best proven constant had only been reduced to 2.414. Jin approached the problem using sophisticated prompt engineering with GPT-5.6. He adapted a known prompting strategy, isolating the AI from external resources to force original reasoning, employing multiple divergent "sub-agents" to explore different paths, and enforcing rigorous adversarial review of proposed proof steps. After about 16 hours of autonomous, unsupervised operation, the AI produced a novel and elegant proof. The key insight involved a clever "sampling strategy" that reduced the problem to a simple positivity condition—a solution described as elegant and unexpected by experts. The proof was verified by the conjecture's originator, Michel Crouzeix, and other specialists like Alex Townsend and Anne Greenbaum, who expressed astonishment at its validity. Jin has made the entire process open-source, including the prompt, drafts, and formal verification code. Remarkably, just eight days after Jin's preprint appeared, mathematicians Emiel Lorist and Felix Schwenninger published an independent, concise 5-page proof using a different approach, also developed with the aid of ChatGPT 5.6. Jin welcomed this complementary work. This event marks a potential turning point, demonstrating how AI can enable experts from non-traditional backgrounds to solve deep theoretical problems and dramatically accelerate scientific discovery, heralding what some are calling a new "golden age" for interdisciplinary research.

marsbitHace 11 min(s)

A 22-Year-Old Mathematical Puzzle, Solved by a Union Hospital Intern?

marsbitHace 11 min(s)

Embodied AI Companies Have Yet to Learn How to Spend Money | TMTpost In-depth

Embodied AI companies in China are facing unprecedented challenges in capital management after a wave of massive funding. The industry, seen as the ultimate carrier for AI, attracted approximately 43.8 billion RMB in the first half of 2026 alone, creating a landscape where even small startups hold billions in cash. However, this influx has exposed a critical gap: many founders—often scientists and engineers—lack experience in deploying such large sums effectively. The article highlights contrasting and often problematic approaches to spending. Some companies practice extreme frugality, drastically limiting R&D, marketing, and even basic operational costs to extend their financial runway, sometimes resorting to living off investment income. This "wait-it-out" strategy, while conserving cash, risks stifling innovation, causing talent drain, and missing crucial product development windows. In one case, excessive cost-cutting led to catastrophic data loss. Conversely, other firms spend recklessly. Examples include a company secretly paying 100 million RMB for ineffective TV exposure, jeopardizing its IPO plans, and others funding multiple unproven product lines simultaneously or creating deceptive demos to attract further investment. The cautionary tale of Vicarious Surgical, which burned through over $100 million on an overly complex proprietary arm before failing, is cited. A core issue is the immense and often opaque cost structure. High salaries for scarce AI talent, exorbitant compute costs for training models (especially "embodied brains" or world models), and the colossal expense of acquiring high-quality robotic training data create financial black holes. Estimates suggest collecting 1 million hours of usable data could cost over 1.6 billion RMB, with most collected data being unusable. This lack of transparency extends to investors, who struggle to verify how funds are actually spent. There are reports of companies maintaining separate internal accounts, engaging in circular "data trading" to artificially boost revenue, and general obfuscation around R&D burn rates. In response, some investors are taking unprecedented control, embedding their own financial personnel to approve even minor expenses. The sector is at a crossroads. While capital continues to flow due to China's strategic advantage in supply chains and engineering, the fundamental question has shifted from securing funding to learning how to spend it wisely. The industry must now master the difficult discipline of allocating vast resources to drive genuine technological progress and sustainable business models, or risk a significant reckoning when the investment tide eventually recedes.

marsbitHace 27 min(s)

Embodied AI Companies Have Yet to Learn How to Spend Money | TMTpost In-depth

marsbitHace 27 min(s)

Trading

Spot
活动图片