Nvidia срывает куш: Big Tech выделил $650 миллиардов на «железо»

The stock market sometimes resembles a casino, but when $650 billion is at stake, this is no longer a gamble—it's a carefully calculated survival strategy. Nvidia shares are climbing again, and the reason here isn't just another hype cycle around chatbots, but the hard pragmatism of Big Tech. Microsoft, Google, Amazon, and Meta have essentially committed to the point of no return. They're willing to spend astronomical sums on infrastructure, and the lion's share of that money will inevitably end up in Jensen Huang's pockets. These aren't just investments; this is a massive restructuring of the global economy around the needs of neural networks.

Just six months ago, many self-proclaimed prophets predicted imminent "hangover" after the explosive growth of 2023. They talked about chip overproduction, about software lagging behind computational power, and about the bubble bursting any moment. However, fresh reports from tech giants revealed the exact opposite: computing power deficits are still being felt, and corporate appetites are only growing. These $650 billion—this isn't just a line item in the server upgrade budget. It's the price of admission to the club of those who will shape the face of technology for the next decade. If a company isn't investing in hardware today, tomorrow it simply won't be able to launch a next-generation model capable of competing on the market.

Why is this happening right now? Because the race has shifted from the "let's try" phase to industrial-scale deployment. Training models like GPT-5 or new iterations of Gemini requires exponentially more computational resources than their predecessors. Nvidia is playing the role of the sole supplier of quality "shovels" during the AI gold rush. And these shovels are golden themselves, with virtually no real alternatives. While AMD attempts to close the software ecosystem gap, and cloud providers are only designing their custom chips, Nvidia is already pushing out shipments of its new Blackwell architecture, which promises even greater performance dominance.

It's interesting to observe how investor rhetoric is shifting. If they once demanded immediate profits from AI implementation in end products, now they seem to have accepted that first you need to build a massive "factory," and only then wait for the output. These colossal capital expenditures (Capex) are a long-term bet. Corporations fear not that they'll overspend, but that they'll underspend and lose the arms race. In this context, Nvidia becomes not just a hardware manufacturer, but something like a tax authority for the entire AI industry. Want to build the future? Pay the Nvidia "tax" and get in line for a GPU.

For the entire industry, this is a powerful signal about ongoing consolidation. It will become even harder for small players to compete with giants that can afford purchases in the tens of billions of dollars. We're witnessing the formation of a new tech oligopoly, where computational power becomes as solid a currency as oil in the twentieth century or gold in the nineteenth. Whoever has more GPU clusters and a better-tuned CUDA ecosystem dictates the rules of the game. And as long as this trend continues, Nvidia will remain at the top of the food chain, ignoring any talk of market bubbles and temporary corrections.

Of course, a reasonable question arises: what happens when all data centers are built and packed with chips to capacity? But that moment is still very far away. Each new generation of large language models requires an order of magnitude more power for training and inference, which means the hardware refresh cycle will only accelerate. We're entering an era where software and hardware are linked as never before. Nvidia is successfully exploiting this connection, turning pure computational power into the most scarce and expensive resource on the planet. Investors understand this and continue to vote with their dollars for the "green" monopoly.

The bottom line: $650 billion is a powerful injection of confidence for the entire sector. As long as Big Tech is willing to pay for the right to stay in the game, Nvidia will grow, and skeptics will have to find new arguments. The only question is when these investments transform into real business value beyond chip sales, but the market seems ready to wait as long as it takes.