Nvidia and the era of 'Neoclouds': A $170 billion bet

The paradigm shift: Beyond the hyperscalers

The tech industry is undergoing an unprecedented structural reconfiguration. Nvidia, under the leadership of Jensen Huang, has injected over $170 billion in investments and direct financial commitments into so-called 'neoclouds.' This move, documented through an exhaustive analysis of financial reports filed with the SEC and analyzed by The Information and The Next Web, represents much more than a simple sales strategy: it is a forced verticalization of the artificial intelligence supply chain.

Historically, the cloud infrastructure market was dominated by the triopoly of AWS, Azure, and Google Cloud. These 'hyperscalers' built their empires on the foundation of general-purpose computing. However, the era of generative AI demands a radically different architecture, focused on low latency and massive parallel performance. 'Neoclouds' (such as CoreWeave, Lambda Labs, or Firmus) have emerged to fill this void, offering direct access to H100 and Blackwell GPU clusters without the abstraction layers or restrictions of traditional services. Nvidia is acting not only as a supplier but as the architect of this new infrastructure, ensuring that its hardware does not encounter bottlenecks in the procurement processes of the traditional giants.

Why is this bet critical now?

Jensen Huang's thesis is that the demand for compute is "off the charts." In the current context, hyperscalers prioritize their own models (such as Gemini, Claude, or Microsoft/OpenAI models), which often leaves startups and mid-sized companies with limited access to GPUs. By funding neoclouds, Nvidia guarantees an ecosystem where its hardware has a fast track to market. This strategy is reminiscent of Intel's vertical integration in the 90s, but with a key difference: Nvidia is financing its own customers so they can buy more chips from it.

The $170 billion investment is a figure that exceeds the GDP of many countries. It is not just capital; it is a transfer of risk. By backing companies like Firmus, Nvidia is acting as a tech development bank, assuming that the return on this investment will come from the industry's incessant consumption of compute cycles. If hyperscalers decide, at some point, to develop their own silicon on a massive scale—as they already do with their Trn or TPU chips—Nvidia will have already consolidated an alternative distribution network that allows it to maintain its market share.

Consequences for the ecosystem

• Decentralization of power: The absolute dependence on the three major cloud providers is fracturing. This allows AI companies to choose providers that optimize the performance of Nvidia GPUs rather than optimizing the profitability of the provider's software layers.
• Price competition and efficiency: The specialization of neoclouds allows for reduced operating costs by eliminating redundant services. This will force AWS and Azure to adjust their margins in high-performance computing, benefiting the end user.
• Systemic risk: This is the most vulnerable point. Nvidia is turning its balance sheet into a guarantee for the growth of the AI sector. If the bubble of investment in language models were to undergo a correction, Nvidia would not only lose future sales but would see the financial assets of its closest allies compromised.

Speculation and the future: Sustainability or bubble?

It is important to note that, although the data comes from official financial reports, the long-term viability of neoclouds is a subject of intense debate on Wall Street. There is speculation that Nvidia could be artificially inflating demand to maintain its stock valuation, although the company maintains that AI adoption in Fortune 500 companies is just beginning. Unlike the dot-com bubble in 2000, where companies lacked tangible assets, here there is real physical infrastructure: data centers are being built and GPUs are being installed.

The future will depend on whether the adoption rate of generative AI manages to monetize enough to sustain infrastructure costs. If neoclouds manage to democratize access to compute, we will have witnessed the birth of a new era of technological sovereignty. If, on the contrary, demand stalls, Nvidia could find itself with a portfolio of highly leveraged clients, which would force the company to make historic write-downs. For now, the market is watching with caution: the appetite for AI is insatiable, but capital, sooner or later, always demands tangible returns.