The Seven-Gigawatt Gap: America's AI Ambitions Are Colliding With Its Power Map

A wave of delayed and canceled data center projects is pushing the contest to power American artificial intelligence toward a new class of developers that bring their own generation.

NEW YORK, June 11, 2026

The defining constraint on American artificial intelligence has shifted. It is no longer chips or capital but the ability to deliver electricity to the buildings where models are trained and run, which makes the AI race, increasingly, a national capacity problem, one measured in gigawatts and interconnection queues rather than benchmark scores.

On Thursday, Nixxy, Inc. (NASDAQ: NIXX) put a number on the problem. Citing recent industry reports, the company said nearly half of all U.S. AI data center projects planned for 2026 deployment have been delayed or canceled amid power constraints, equipment shortages, permitting challenges, and construction delays, and that industry analysts estimate more than 7 gigawatts of anticipated AI computing capacity may fail to come online as originally scheduled.

That is the company's framing, but it sits comfortably inside what independent trackers have been documenting all year. Sightline Climate's February "Data Center Outlook," which follows 190 gigawatts of announced capacity across 777 large data centers, found more than 16 gigawatts slated for 2026 across roughly 140 projects, with only 5 gigawatts actually under construction despite typical build times of 12 to 18 months. Given that track record, Sightline analyst Olivia Wang wrote, it "wouldn't be surprising" if 30 to 50 percent of the capacity slated for 2026 ends up delayed.

A demand curve that does not wait

The shortfall would matter less if demand were cooling, and every available indicator says it is not. The four largest hyperscalers have guided to roughly $725 billion in combined 2026 capital expenditure, up 77 percent from 2025's $410 billion, according to a Tom's Hardware aggregation of first-quarter earnings. The Stargate program, the $500 billion, four-year buildout announced by SoftBank, OpenAI, Oracle, and MGX, had grown to roughly 7 gigawatts of planned capacity and more than $400 billion in commitments by last September, according to OpenAI.

The macro forecasts point the same direction. The International Energy Agency projects global data center electricity consumption will more than double to roughly 945 terawatt-hours by 2030. Lawrence Berkeley National Laboratory and the Department of Energy project U.S. data centers will consume between 6.7 and 12 percent of American electricity by 2028, up from 4.4 percent in 2023. SemiAnalysis estimates U.S. AI power demand alone will exceed 28 gigawatts this year, up from about 3 gigawatts in 2023.

The people building the machine have been saying as much for months. "The biggest issue we are now having is not a compute glut, but it's power… you may actually have a bunch of chips sitting in inventory that I can't plug in," Microsoft CEO Satya Nadella said on the Bg2 Pod in November. OpenAI President Greg Brockman went further on CNBC last fall: "we're heading to this world where the economy is powered by compute, and it's going to be a compute scarce one… no matter how much gets built, we are still going to be in this world of compute scarcity."

For a country whose stated ambition is durable AI leadership, at a moment when governments worldwide are standing up sovereign AI programs to secure domestic compute, a multi-gigawatt slip in a single year carries weight well beyond the property market. It is the kind of shortfall that industrial policy normally exists to prevent.

Where the buildout jammed

The bottlenecks are well mapped. Lawrence Berkeley National Laboratory's "Queued Up" report counted more than 2,060 gigawatts of generation and storage waiting in U.S. interconnection queues at the end of 2025, with a median of roughly five years from request to operation. Data Center Knowledge reported in May that projects entering queues in 2025 face seven-plus years to operation in some regions, and PJM alone logged 95 large-load requests totaling about 54 gigawatts through November 2025.

The hardware pipeline is just as congested. Wood Mackenzie puts average power transformer lead times at roughly 120 weeks, up from about 50 weeks in 2021. GE Vernova disclosed in April that its gas turbine backlog had reached 100 gigawatts, with data centers accounting for roughly 20 gigawatts of it. "2026 and '27 are largely sold out; we are approaching filling out '28," GE Vernova CEO Scott Strazik said last July.

Washington has noticed. The Federal Energy Regulatory Commission has said it will act by this month on large-load interconnection reform, including fast-tracked studies for co-located load, and in December ordered PJM to create new co-location service options after finding its existing behind-the-meter rules "no longer just and reasonable." Texas's SB 6 is rewriting large-load rules in the nation's most active data center market. Speed to power, in other words, has become a policy priority as well as a commercial one.

The power-first developers

Into the gap has stepped a class of developer that declines to wait in line. Forty-six data centers totaling roughly 56 gigawatts now plan to generate their own behind-the-meter power, according to figures reported by Grist and Marketplace in February. The appeal is timeline arbitrage: behind-the-meter gas generation can reach first power in roughly 18 months, against grid connection waits of four to seven years in major markets, according to analyses from Data Center Knowledge, SemiAnalysis, and Orrick. Crusoe's Stargate campus in Abilene, the up-to-4.5-gigawatt Homer City redevelopment in Pennsylvania, and ExxonMobil's planned 1,500-plus-megawatt data center plant all follow the pattern. "Speed-to-market is the most critical aspect of powering AI development in the United States," Engine No. 1 founder Christopher James has argued.

Thursday's announcement places the proposed Nixxy-Tachyon9 combination squarely in that cohort. The combined platform, the company said, is being designed to integrate power, infrastructure, financing, and compute deployment into a single execution framework, a deliberate contrast with projects dependent on third-party power providers and fragmented construction partners. Through its development pipeline, Tachyon9 is pursuing approximately $1 billion of planned capital investment, largely through debt facilities and construction loans intended to be backed by projected offtake agreements, with a Phase 1 buildout of 120 MW slated for Q2 2027, according to the announcement. The company's stated route around the interconnection queue is behind-the-meter gas turbines.

"We recognized early that AI's greatest bottleneck would not be models or GPUs - it would be infrastructure," said Shahal Khan, chief executive officer of Tachyon9, in the release. "We are building behind the meter gas turbines, so we avoid these delays."

Per the announcement, Tachyon9 contributes approximately $64 million in equipment, land option rights for the Nakota project, and a signed LOI for the entire 1 GW development, while Nixxy brings its public market platform, telecommunications infrastructure, AI technologies, and capital markets capabilities. Nixxy describes its strategy as supporting global demand for AI compute, sovereign AI initiatives, and energy-backed digital infrastructure.

The financing model the companies describe tracks a template that has become standard across the sector, with lenders sizing construction debt against contracted cash flows rather than developer balance sheets, a structure documented by Orrick and S&P Global Market Intelligence as data center debt issuance nearly doubled to $182 billion in 2025. That template carries a gating requirement: projected offtake has to become signed, creditworthy contracts before the debt can price. The conversion is the milestone to watch as the Nixxy-Tachyon9 plan advances.

Whether the seven-gigawatt gap becomes a ceiling on American AI capacity or an opening for the power-first builders will be settled project by project over the next 24 months. The market has already decided where the leverage sits. "Power has become the new real estate," as JLL's head of U.S. data center research, Andrew Batson, has put it, and the companies announcing combinations like Thursday's are betting that the observation holds for the rest of the decade.

About Nixxy, Inc.

Nixxy, Inc. (NASDAQ: NIXX) is an AI communications and data infrastructure company focused on next-generation digital infrastructure platforms positioned at the intersection of artificial intelligence, high-performance compute, energy, and data center infrastructure. The Company is pursuing large-scale opportunities supporting the rapidly growing global demand for AI compute capacity, sovereign AI initiatives, and next-generation energy-backed digital infrastructure. For more information, visit www.nixxy.com.

About Tachyon9

Tachyon9 is a private operating company specializing in energy infrastructure, transmission equipment, and data center assets. Tachyon9 serves as the primary asset and revenue contributor in the proposed transaction, contributing approximately $64 million in equipment, land option rights for the Nakota project, and a signed LOI for the entire 1 GW development.

Contacts

Investor Relations: Nixxy, Inc., [email protected] Media: John Arundel, Managing Director, Perdicus Global Communications, Washington, DC, [email protected], (703) 963-4191

Forward-Looking Statements

This communication contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995, including statements regarding the proposed strategic combination between Nixxy, Inc. and Tachyon9 Corporation, planned capital investment, development timelines, projected offtake agreements, and anticipated market conditions. Forward-looking statements are based on management's current expectations and involve risks and uncertainties that could cause actual results to differ materially from those expressed or implied. Investors should review the risk factors described in the Company's Annual Report on Form 10-K filed with the Securities and Exchange Commission on April 15, 2026, and other filings with the Securities and Exchange Commission. Nixxy undertakes no obligation to update forward-looking statements except as required by law.