AI growth meets energy limits: Nvidia leads, nuclear poised for revival
Nvidia is once again driving the American stock market. The chipmaker’s securities have reached a historic high this week, and the company’s capitalization has exceeded $4.5 trillion. The 49% increase since the beginning of the year reflects not only the steady demand for accelerators, but also the avalanche of new deals that increasingly link Nvidia with OpenAI. Investors are also focused on the giant Stargate project – a consortium of data centers worth hundreds of billions of dollars, where the infrastructure will be built on Nvidia GPUs.
According to Jensen Huang, Nvidia products account for about 70% of the cost of new AI data centers. This indicator makes the company an infrastructure monopolist in the new technological cycle. Add to that CoreWeave’s $14.2 billion cloud service deal with Meta and announcements from Google and Microsoft, and it’s clear why Nvidia continues to almost single-handedly lift market indexes and boost ES futures.
OpenAI also remains at the forefront of the AI market, but its business model is yet to achieve financial stability. In the first half of the year, the startup earned $4.3 billion, with operating expenses of $2.5 billion and additional $6.7 billion invested in R&D. By the end of 2025, OpenAI projects $13 billion in revenue and $8.5 billion in expenses, which confirms the company’s ability to maintain a high pace of development, although the long-term return on investment remains uncertain.
The key factor remains the Nvidia deal. Formally, the company announced a $100 billion partnership, but a significant portion of these funds will flow back to Nvidia through rental payments for its accelerators. In fact, OpenAI is becoming not only Nvidia’s largest customer, but also a catalyst for its further market growth. Every new dollar OpenAI or Meta invests in development directly translates into higher demand for Nvidia accelerators.
But accelerators require energy – creating not only new risks but also significant opportunities. It is expected that the increase in electricity consumption by AI centers will lead to a boom in investments in the US nuclear energy industry of approximately $350 billion by 2050. The total capacity of nuclear power plants can grow by 63%, mainly due to small modular reactors (SMRs).
However, there are several limitations to consider here, such as the high cost of construction, the shortage of personnel, and the lack of commercial SMRs in the next ten years. Capacity growth will be limited until 2035, potentially creating an imbalance between the pace of data center construction and available generation. The US authorities have already slowed down the closure of coal-fired power plants in order to maintain the demand for AI data centers.
The existing Nvidia-OpenAI-energy partnership outlines a new framework for global growth. For the US stock market this means a continuation of the AI rally, with Nvidia leading the way – as reflected in the heatmap, where it holds the top spot by market capitalization. But at the same time, the load on the energy infrastructure is growing, which means that nuclear companies and suppliers of generation equipment will gradually take a stronger role.
The current cycle is not only a race of chips, but also a race of resources, capital, energy, and infrastructure. Nvidia is winning here and now, OpenAI is still building scale without guaranteed payback, and the US nuclear industry is getting a chance to be reborn after decades of stagnation. Together, this triad will shape market dynamics for the upcoming years.