June 2026 is witnessing a convergence of AI infrastructure investment announcements unprecedented in the history of technology. Four simultaneous developments — from OpenAI’s megacampus in Ohio to regulatory moratoriums in New York and Seattle — paint a picture of accelerated expansion and growing tension that will define the coming years of the sector.
Update, September 4, 2026. The Ohio lease, still under negotiation in June, was signed on August 17. The section below now carries the figures from the official announcement; the rest of the article describes June 2026 and is kept as such.
OpenAI and the Ohio Megacampus: Scale Shifts to Another Planet
In June, OpenAI was in advanced negotiations to lease a 10-gigawatt data center at the former site of the Portsmouth Gaseous Diffusion Plant in southern Ohio: Department of Energy land, built during the Cold War to enrich uranium and idle since 2001. Reuters put the total construction cost at at least $500 billion at then-current prices for chips, labor and energy.
The deal was signed on August 17, 2026. Per the official announcement, SB Energy — the SoftBank-backed energy company that had already broken ground in March — builds, owns and operates the PORTS-Pike Technology Campus, and OpenAI leases it for 20 years. The committed capacity is 8 IT-GW of compute, of which the first 4.25 IT-GW are covered now, with an option to extend. NVIDIA invests $1.5 billion in SB Energy, and the campus will host NVIDIA compute exclusively. SB Energy and SoftBank are also putting at least $4.2 billion into regional grid infrastructure, alongside an $80 million community benefits fund that OpenAI tops up by a further $40 million.
Capacity comes online in phases starting in 2028, not all at once. For scale: one gigawatt is roughly a large nuclear reactor, and ten would power some eight million American homes. It is, by a wide margin, the largest data center ever conceived.
One distinction the announcement makes and the coverage tends to lose: gigawatts of compute (IT-GW) are not the same as the site’s power gigawatts. And a signed contract is not a finished campus — OpenAI has walked away from a UK data center plan at a late stage before, over energy cost and regulation.
Broadcom, Blackstone, and Apollo: Private Capital Enters the Scene
On June 9, Broadcom, Apollo Global Management, and Blackstone jointly announced the creation of the “AI XPV Platform,” a strategic financing structure that launches with an initial transaction of $35 billion for more than 1 GW of AI compute capacity. The platform is designed to enable over 20 GW in total capacity by 2028, using Broadcom chips.
The model is innovative: it is not a traditional investment fund, but rather a “capital solution” based on private credit and structured debt, with participation from major global banks. Apollo leads the operation in partnership with Blackstone, and the destination for that compute capacity is frontier AI labs — with Anthropic mentioned as a possible lead customer, given it already uses Broadcom chips.
This platform reflects a broader trend: private capital is gradually replacing traditional corporate balance sheets as the dominant source of financing for AI infrastructure. The scale required — tens of billions per project — exceeds what even the largest tech companies can fund internally.
Oracle: Capex Surges and Debt Piles Up
Oracle reported its fiscal fourth-quarter 2026 results (ended May 31) on June 10, with record quarterly revenue of $19.2 billion (up 21% year-over-year) and annual revenue of $67.4 billion (up 17%). However, what captured the market’s attention was its capital expenditure: $55.7 billion in fiscal year 2026, a 162% increase from the $21.2 billion the previous year.
And the trend is not slowing down. The company guided for fiscal 2027 net capex of approximately $70 billion, with plans to raise around $40 billion in additional debt and equity — including a $20 billion stock offering already announced. Oracle’s total debt at fiscal year-end stood at roughly $117 billion, making it the largest issuer of corporate debt outside the financial sector.
The market reaction was immediate: Oracle shares fell more than 7% after hours, reflecting investor concern over rising leverage and the question of whether these investment levels will generate proportional returns. Oracle’s cloud division — which includes OCI (Oracle Cloud Infrastructure) — generated $34 billion in annual revenue, up 39%, but the cost of that expansion is extraordinarily high.
The Regulatory Counterweight: Moratoriums in New York and Seattle
While private investment in AI infrastructure accelerates, local governments are beginning to push back. Two key jurisdictions have approved restrictive measures in June:
The New York State Legislature passed the “Responsible Data Centers Act” (Senate Bill S9144A), which imposes a one-year moratorium on issuing permits for new data centers over 20 MW. It is the first state-level moratorium of its kind in the United States. The law requires the Public Service Commission to issue orders minimizing the impact of data centers on electric and gas rates for residential, commercial, and industrial users. Governor Kathy Hochul has yet to sign or veto the bill.
Meanwhile, the Seattle City Council — home city of Amazon and Microsoft — approved a one-year ban on new AI data centers. The measure reflects growing public unease over the massive energy consumption of these facilities, at a time when power grids in many regions are already strained by the electrification of transportation and heating.
These moratoriums are not isolated cases. Several states are considering similar legislation, and the debate over the environmental and energy impact of AI is intensifying as projects reach scales that were previously unimaginable. A 10 GW data center like the one planned by OpenAI would, on its own, consume the equivalent output of several nuclear power plants.
The Clash of Trends
What makes this moment unique is the simultaneity of these movements. On one hand, investment in AI infrastructure is approaching near-trillion-dollar scales when adding up OpenAI’s project ($500 billion), Oracle’s capital expenditure ($55.7 billion annually), the Broadcom/Apollo/Blackstone platform ($35 billion initial), and the investment plans of other players like Microsoft, Google, Amazon, and Meta, which will collectively spend hundreds of billions in the coming years.
On the other hand, the regulatory reaction is just beginning. The moratoriums in New York and Seattle could be only the start. The challenge for the industry will be to demonstrate that the massive deployment of AI compute capacity can be reconciled with climate goals, grid stability, and equitable access to energy.
AI infrastructure is reaching a scale that rivals entire industries — power generation, construction, and the financial sector. The open question is whether regulation will manage to keep pace, or whether the regulatory brake will arrive too late or too early for a sector moving at exponential speed.
Primary source: NVIDIA Newsroom — PORTS-Pike Technology Campus