AI Data Center & Build-out

Investment in artificial intelligence is reaching unprecedented levels. According to data from the World Intellectual Property Organization, or WIPO, in its Global Innovation Index 2026 report, global venture capital investment rose 27.9% in 2025 to $510 billion, the strongest annual expansion since the peak in 2021. The number of deals, however, fell 1.4% to about 43,500, a fourth consecutive year of decline. AI accounted for 53% of global VC investment value in 2025, up from about 27% in late 2022, and in the first half of 2026 that share surged to 77%. North America attracted about $213 billion in AI investment, out of roughly $270 billion in AI-related VC worldwide in 2025, accounting for 86% of global AI VC value in early 2026. PwC estimates that global data center investment alone could total more than $30 trillion cumulatively by 2050, with a base-case estimate of $31.6 trillion produced jointly with Oxford Economics, within a projection range of roughly $22 trillion to $50 trillion. Annual data center spending could rise from about $800 billion in 2026 to roughly $1.8 trillion a year by 2050. But Bain & Company estimates that by 2030 the AI industry will need to generate about $6 trillion in annual revenue to support the computing power and infrastructure now being built, while existing AI services for consumers and businesses may generate only about $1.2 trillion to $1.8 trillion. That means the industry still faces a revenue gap of about $4.2 trillion that must be created from new markets. David Crawford, Bain's global head of technology, said AI infrastructure is being built far ahead of the demand curve, and making that investment sustainable may require raising global GDP growth by about 1 percentage point a year. Goldman Sachs estimates that major hyperscalers will need to generate about $300 billion in AI revenue within a few years just to earn back the money being poured into data centers, chips and networking systems. Under an assumption of a 15% annual return on investment, the six largest U.S. hyperscalers would need to generate combined revenue of about $1.42 trillion in 2028-2030 to support the investment being made in 2026-2027. Stijn Van Nieuwerburgh, an economist at Columbia Business School, estimates that AI investment in the United States could total about $9 trillion in 2025-2032, and the U.S. AI industry may need to generate about $3.55 trillion in annual revenue by 2032 to deliver a return on investment of about 10%. Diane Coyle, an economist at the University of Cambridge, notes, however, that major technologies of the past have typically taken about 10 to 50 years for their effects to be fully reflected in productivity figures. Meanwhile, HSBC's Global Entrepreneurial Wealth Report 2026 found that wealthy entrepreneurs worldwide plan to spend a combined total of about $367 billion on AI over the next 12 months, with four in five planning to increase spending on the technology. WIPO said R&D investment by the world's leading companies reached a record high of about $1.5 trillion in 2025, with software- and AI-related companies increasing investment fastest.

Oracle Corporation has issued a force majeure notice on Project Jupiter, its New Mexico AI data center campus, which includes Bloom Energy Corporation's largest single fuel-cell deployment covering up to 2.45 gigawatts within a project expected to involve up to $165 billion of investment over its life. The notice is a financial protection against regulatory delays intended to defer payments if the data center is not operational by 2028, not an indication that Oracle is leaving the project as a tenant, and Oracle told Bloomberg it remains fully committed to New Mexico. Project Jupiter has already faced permitting hurdles, including the New Mexico State Land Office's rejection of pipeline permits in March and again in July, which led Oracle to replace gas turbines with Bloom fuel cells earlier this year. Bloom Energy's exposure is significant and concentrated, as Jupiter represents its single largest deployment within roughly 25 gigawatts of total pipeline visibility, and debt linked to the project has been trading below 90 cents on the dollar. Bloom Energy said its equipment is fungible and can be deployed elsewhere if Jupiter is delayed, while Oracle carries about $89 billion more debt than cash and depends on projects like Jupiter opening on time to support growth that jumps to between 35% and 43% a year.

NVIDIA is carrying $279 billion in supply and capacity commitments, up from $119 billion at the prior quarter-end, a schedule that now hinges on whether customer spending keeps pace. The filing allocates $92 billion of those commitments to the remainder of fiscal 2027, $87 billion to fiscal 2028 and $88 billion to fiscal 2029, with smaller amounts afterward, and NVIDIA said it may cancel, reschedule or adjust some agreements before firm orders. NVIDIA also disclosed August guarantees capped at $105 billion for an OpenAI-related SB Energy buildout, which generally begin as the relevant leases commence, with the first construction phase expected in fiscal 2029 and payments dependent on specified tenant defaults. Inventory rose to $31.58 billion from $21.40 billion at the fiscal year-end, while accounts receivable increased to $63.06 billion from $38.47 billion, and some investment-grade customers have payment terms from 90 days to one year. NVIDIA generated $96.22 billion of second-quarter revenue and $63.73 billion of operating income, an operating margin of about 66%, and management guided to $108 billion of next-quarter revenue, plus or minus 2%, excluding China data-center compute sales. At September 30's $228.38 close, NVIDIA's equity was worth about $5.51 trillion, and its trailing free cash flow of roughly $127 billion implies a cash yield of about 2.3%.

Alphabet reportedly won two separate US antitrust cases in late September, easing immediate legal pressure on Google. Google is helping form a cross industry coalition focused on AI data center development ahead of the 2026 US midterm elections, expected to address questions around power use, data sourcing and community impact from large scale AI computing hubs. The antitrust decisions support the view that Alphabet can keep rolling out AI powered features across Search, YouTube and Google Cloud without immediate structural remedies that break its distribution. The AI data center coalition cuts both ways, aligning with the potential to monetise AI infrastructure and long duration power contracts while spotlighting the risk that regulators and communities could still cap how far Alphabet can push energy hungry AI workloads. Analysts have already flagged regulatory pressure and high capital intensity as two of the biggest swing factors for the company.

Amazon Web Services CEO Matt Garman said the company has stopped using nondisclosure agreements in its dealings with government agencies as it seeks approval to build new data centers. The statement came in a blog post in which Garman pushed back against widespread suspicion of data centers, noting that more than 100 data center moratoriums are currently being considered across the United States after New York announced a one-year moratorium on permits for large data centers. Garman argued that direct data center water consumption accounts for only 0.5% of all industrial water usage in the United States, and that where energy rates are rising it is primarily because the grid is old and has not been invested in and expanded before the demand arrived. He also said data center generators are idle 99.9% of the time, running roughly 10 hours per year, and that Amazon has contributed more than $1 billion to communities across the U.S. in which it has a meaningful data center presence over the past three years. Critics remain unconvinced, with an independent watchdog recently blaming data centers for a 76% year-over-year price increase on America's largest electrical grid.

Meta Platforms has reportedly classified some of its artificial intelligence data centers as "pilot models" to claim federal research and experimentation tax credits, a move that reduced its tax bill by $3.9 billion in 2025, up from $2 billion in 2024 and $700 million in 2023, according to a New York Times report citing company filings. The classification, which describes certain AI data centers to the Internal Revenue Service as experimental work, remains a gray area, and Meta's accountants have raised concerns that the IRS could challenge the treatment, people familiar with the company's operations told the Times. The tax claims reportedly involve chips Meta purchases for its AI data centers, including Nvidia GPUs, and the Times also reported that Meta's auditor, EY, has promoted the tax strategy to other AI companies. Meta spokesperson Andy Stone told the publication that the company has invested $200 billion in research and development over the past five years, including $57 billion in the last year, using tax incentives established by Congress to support domestic investment in research, technology and jobs. The tax perk comes as Meta's AI investments weigh on its finances, with quarterly free cash flow of just $784 million, roughly $8 billion below the year-ago period, and the company is among several major technology companies, including Amazon, Microsoft and Alphabet, spending heavily on AI infrastructure.

EMCOR Group reported record second-quarter 2026 results, with revenues climbing 19.8% year over year to $5.15 billion and operating income surging 31.8% to $547.3 million. Operating margin expanded 100 basis points to 10.6%, while earnings per share soared 34.8% to $9.06. Remaining Performance Obligations reached a record $17.14 billion, up 43.9% year over year and 29% from December 2025, with Network & Communications, Water & Wastewater, and Institutional and Healthcare among the biggest contributors. The company is also widening its reach through M&A, as five recently announced electrical acquisitions bring roughly $625 million in trailing-12-month revenues, $105 million in EBITDA and about 1,500 employees. EMCOR said labor shortages, tariffs, supply-chain volatility and project-mix shifts remain risks, but its diversified demand base and acquisition strategy could make its broad market footprint a meaningful competitive advantage.