The multi-billion dollar infrastructure agreements fueling the surge in artificial intelligence

The Race to Build AI Infrastructure

Running advanced AI systems demands enormous computational resources. As technology companies rush to harness the capabilities of AI, there’s a simultaneous scramble to construct the massive infrastructure required to support these models. During a Nvidia CEO Jensen Huang projected that global spending on AI infrastructure could reach between $3 trillion and $4 trillion by 2030, with much of this investment coming directly from AI-focused firms. This surge is putting unprecedented pressure on power grids and pushing the limits of current construction capacity.

Below is a comprehensive overview of the most significant AI infrastructure initiatives, highlighting major investments from industry leaders such as Meta, Oracle, Microsoft, Google, and OpenAI. We’ll continue to update this summary as the sector’s growth accelerates and spending rises.

Microsoft and OpenAI: The 2019 Game-Changing Investment

One of the pivotal moments in the current AI boom occurred in 2019, when Microsoft committed $1 billion to OpenAI, then a nonprofit organization best known for its ties to Elon Musk. A key aspect of this agreement was making Microsoft the exclusive cloud provider for OpenAI. As AI model training became more resource-intensive, Microsoft’s contributions shifted increasingly toward Azure cloud credits rather than direct cash allocations.

This partnership proved mutually beneficial: Microsoft boosted its Azure sales, while OpenAI secured essential resources for its most significant operational costs. Over time, Microsoft’s total investment approached $14 billion—a move poised to yield substantial returns as OpenAI transitions to a for-profit model.

However, the exclusive relationship has since evolved. In the past year, OpenAI announced it would no longer rely solely on Microsoft’s cloud services, instead granting Microsoft the right of first refusal for future infrastructure needs but seeking alternatives if Azure couldn’t deliver. Microsoft, in turn, has started exploring other foundational AI models, further diversifying its approach.

The success of the Microsoft-OpenAI arrangement set a precedent in the industry. Anthropic, for example, has secured $8 billion from Amazon and is customizing Amazon’s hardware for AI workloads. Google Cloud has also become the primary computing partner for startups like Lovable and Windsurf, though these deals did not include direct investments. OpenAI itself received a $100 billion investment from Nvidia in September, enabling further GPU acquisitions.

Oracle’s Rapid Ascent in AI Infrastructure

On June 30, 2025, Oracle disclosed a $30 billion cloud services agreement with an undisclosed partner—later revealed to be OpenAI. This deal alone surpassed Oracle’s entire cloud revenue from the previous fiscal year and positioned Oracle alongside Google as a key hosting provider for OpenAI after Microsoft. The announcement sent Oracle’s stock soaring.

Just a few months later, Oracle announced an even larger five-year, $300 billion compute power contract set to begin in 2027 on September 10. The news pushed Oracle’s valuation even higher, briefly making founder Larry Ellison the world’s richest individual according to reports. The scale of this agreement is staggering—OpenAI does not currently have $300 billion to spend, so the deal anticipates massive growth for both companies and a significant degree of optimism.

Even before any funds are spent, these deals have firmly established Oracle as a dominant force in the AI infrastructure landscape.

Nvidia’s Strategic Investments

As AI companies race to expand their infrastructure, most are sourcing GPUs from Nvidia. This demand has filled Nvidia’s coffers, and the company is reinvesting in the sector in creative ways. In September 2025, Nvidia acquired a 4% stake in Intel for $5 billion with plans for AI chip collaboration. Even more notable was Nvidia’s $100 billion investment in OpenAI, paid in the form of GPUs for new data center projects announced a week later. Nvidia has since made similar arrangements with Elon Musk’s xAI, and OpenAI has entered a GPU-for-equity deal with AMD as well.

These deals create a self-reinforcing cycle: Nvidia’s GPUs remain in high demand and short supply, while OpenAI’s private stock grows in value due to its exclusivity. For now, both companies are thriving, but if momentum slows, these unconventional arrangements may come under greater scrutiny.

Constructing the Next Generation of Hyperscale Data Centers

For established players like Meta, which already operate large-scale infrastructure, the challenge is both complex and costly. CEO Mark Zuckerberg has outlined plans to invest $600 billion in U.S. infrastructure by the end of 2028 according to reports.

In the first half of 2025, Meta’s infrastructure spending increased by $30 billion compared to the previous year, largely driven by AI expansion. While some of this budget is allocated to major cloud contracts—such as a $10 billion deal with Google Cloud for AI infrastructure—even more is being funneled into two enormous new data centers.

The Hyperion project in Louisiana will span 2,250 acres and is expected to cost $10 billion, delivering around 5 gigawatts of computing power with support from a local nuclear plant to meet energy needs. Another facility, Prometheus in Ohio, is set to open in 2026 and will be powered by natural gas.

These massive projects come with environmental consequences. For example, Elon Musk’s xAI constructed a hybrid data center and power plant in South Memphis, Tennessee, which has quickly become a major source of air pollution due to its natural gas turbines—raising concerns about Clean Air Act violations according to experts.

The Stargate Initiative: A $500 Billion Vision

Shortly after his second inauguration in January, President Trump announced a partnership between SoftBank, OpenAI, and Oracle to invest $500 billion in U.S. AI infrastructure. Dubbed “Stargate” after the 1994 film, the project was introduced with great fanfare, with Trump calling it “the largest AI infrastructure project in history.” OpenAI’s Sam Altman echoed this sentiment, describing it as “the most important project of this era.”

The plan called for SoftBank to provide the capital, Oracle to manage construction, and OpenAI to offer technical guidance, with Trump pledging to remove regulatory barriers. However, skepticism emerged early on, including from Elon Musk, who questioned the project’s financial backing.

As initial excitement faded, progress slowed. By August, reports indicated that the partners were struggling to align their visions. Nevertheless, construction has begun on eight data centers in Abilene, Texas, with the final building expected to be completed by the end of 2026 as reported locally.

The Capital Expenditure Surge

Capital expenditures, or capex, typically refer to investments in physical assets. But as tech giants revealed their 2026 capex plans, the scale of data center spending became both eye-catching and unprecedented.

• Amazon leads with a projected $200 billion in 2026 capex (up from $131 billion in 2025).
• Google follows closely, estimating $175–185 billion (up from $91 billion in 2025).
• Meta expects to spend $115–135 billion (up from $71 billion the previous year), though some data center investments are kept off the books.

Altogether, hyperscale cloud providers are on track to invest nearly $700 billion in data centers in 2026 alone.

This unprecedented spending has made some investors uneasy. Nevertheless, tech companies remain committed, emphasizing that robust AI infrastructure is essential for their future. This has created a unique tension: while technology leaders are optimistic about AI’s potential, financial analysts are increasingly cautious—especially as companies take on significant debt to fund these projects.

So far, these concerns haven’t slowed the pace of investment, but the pressure is mounting. The coming years will reveal whether these massive bets on AI infrastructure will deliver the returns that tech giants are counting on.

Originally published on September 22.