Goldman Sachs Raises U.S. Data Center Construction Forecast Despite Concerns Over Texas Moratorium

icon币界网
Share
AI summary iconSummary
Goldman Sachs raised its U.S. data center construction forecast, citing robust on-chain data and a full project pipeline. The firm now projects $80 billion in spending for 2026, $100 billion for 2027, and $125 billion for 2028. Despite a potential moratorium in Texas, the report states that delays will not halt the cycle. With 86 GW of active projects and $900 billion in pending work, the market remains strong. Inflation data shows no indication of slowing tech investment.
CoinDesk reports:

The boom in U.S. data center construction is facing political headwinds, but Goldman Sachs believes this cycle is far from over.

Market sentiment sharply deteriorated as Texas incoming Governor Gina Hinojosa called for an immediate moratorium on data centers. The Nasdaq Index fell 6% in a single week, the AI Momentum Index recorded its fifth-largest drop in history, and data center credit default swap premiums hit all-time highs.

However, Goldman Sachs raised its forecasts against the trend, increasing its projection for U.S. data center construction spending in 2026 from $67 billion to $80 billion, in 2027 from $87 billion to $100 billion, and providing its first estimate of $125 billion for 2028.

In two recent research reports, Goldman Sachs noted that the pipeline of U.S. data center projects far exceeds the vulnerability implied by market concerns—currently, 86 GW of projects are under construction, with an additional $900 billion in highly credible proposed projects, and construction spending growth has accelerated to 90% year-over-year ahead of the election cycle.

Goldman Sachs analysts Adam Bubes in Engineering & Construction and Michael Ng in IT Hardware reached the same conclusion: recent policy-related uncertainty has caused some project activities to be delayed, but it will not reduce the overall market size.

86 GW has broken ground, with the project scale far exceeding current spending levels.

The accelerated pace of this construction cycle was already evident before the election.

According to U.S. Census Bureau data cited by Goldman Sachs, the growth rate of U.S. data center construction spending accelerated from 55% year-over-year in June 2024 to 90% year-over-year in September, with the number of new projects starting in the third quarter "significantly increasing," reaching an annualized scale close to $100 billion.

Based on the Aterio project tracking database (data as of the end of October), Bubes categorizes ongoing projects into three tiers: projects that have broken ground, projects with a high probability of being announced but not yet started, and all other projects, and values each tier at approximately $3.5 billion per GW of construction cost (excluding servers, racks, GPUs, and networking equipment).

Among the approximately 86 GW of projects already broken ground, the corresponding construction expenditure is around $300 billion—six times the estimated $50 billion in data center construction spending for the full year of 2025—with over $200 billion of that amount yet to be invested.

Goldman Sachs believes that, after including projects with a high probability of announcement, the total project scale rises to approximately $900 billion; if all medium- and low-probability projects are included, the theoretical opportunity scale expands to approximately $1.9 trillion.

Capacity-side data strongly corroborates construction-side data.

From the perspective of capacity supply, Michael Ng arrives at a similar conclusion.

The United States currently has 57 GW of total active data center capacity, with an additional 9 GW under construction, projected to reach 65 GW by the end of 2026. Based on the 78 GW of projects under construction tracked by Aterio, net capacity additions between 2027 and 2030 are expected to double the U.S. total capacity to 143 GW.

Ng’s core assessment is that the under-construction capacity, even when considered solely as “low-risk,” is sufficient to support Goldman Sachs’ own projections. Converted to IT load at a power usage effectiveness (PUE) of 1.5, the under-construction projects correspond to net additions of 22/24/5 GW-IT in 2027/2028/2029, closely aligning with Goldman Sachs’ forecast of 20/24/23 GW-IT.

In other words, Goldman Sachs' capacity forecasts for 2027 and 2028 are already realized, while projections for 2029 and beyond rely on announced but尚未开工 projects.

Ng also highlights an important consideration: Aterio’s definition of “under construction” includes preliminary site activities such as land clearing, grading, and utility line installation, which developers often begin before signing tenant agreements. If a construction halt order is issued, projects that have already broken ground but have not yet obtained power service agreements or environmental permits may still be affected.

Texas Stay Order: Delayed, Not Terminated

The primary political risk currently focused on by the market comes from Texas.

Among the five states with a cautious stance toward data centers—Texas, Ohio, Nevada, Georgia, and Michigan—these states collectively account for approximately 42% of the capacity of projects under construction or with high probability, with Texas alone representing 26% of the net新增 capacity cumulatively added between 2026 and 2030.

However, Goldman Sachs made an important distinction regarding the nature of the suspension order. Ng noted in the report, “The suspension order is not a ban or cancellation”; its impact lies in causing delays in deployment (which in some cases may lead to contract cancellations), increased financing costs, reduced tax incentives, and higher infrastructure capital expenditures.

Bubes further presents an underappreciated market perspective: the construction timeline for 2028 was already overly aggressive before any voting took place—since, in theory, approximately 55 GW of capacity was slated to come online that year, far exceeding the historical pace of capacity additions.

Goldman Sachs originally anticipated that some projects would experience natural delays; the suspension order by newly elected Texas Governor Gina Hinojosa merely provided a political rationale for such delays, rather than adding further reductions.

Goldman Sachs economists also expect state suspension orders to be conditional and relatively short-lived, and any new federal legislation would still require bipartisan support in the Senate and faces a veto risk.

Regarding Texas, according to Wells Fargo analyst Jerry Revich, the Texas Legislature’s regular session has been extended from January 12, 2027, to May 31, which may also represent the approximate duration of Texas’s moratorium. The next key milestone is the Texas data center audit report, due on December 10.

Major cloud providers show no signs of reducing spending.

Demand side shows no signs of easing. According to Goldman Sachs data, capital expenditures by hyperscale cloud providers are projected to reach approximately $800 billion in 2026, with FactSet’s consensus estimate at around $1.3 trillion for 2027. The combined backlog of revenue for the three major public cloud providers has already exceeded $1.7 trillion.

According to data cited by Revich, H100 compute rental prices have increased by 26% year-over-year, while GB200 rental rates have risen by 50% year-to-date and an additional 12% in the past month, indicating sustained strong demand for compute power and continued market premiums for computational resources.

At the stock level, Ng highlighted IREN as having the highest Texas exposure within its coverage, with 73% to 87% of its data center capacity located in the state; Celestica (CLS) and Arista (ANET) may be impacted through servers and data center switches; Digital Realty (DLR) and Equinix (EQIX) have minimal exposure to development-intensive markets due to their asset positioning in primary and secondary metropolitan areas.

TAM may continue to exist, but highly leveraged developers face substantial risk.

Goldman Sachs' conclusion is relatively cautious; their original wording is:

The projects we track have sufficient scale to support sustained growth, but we currently lack the visibility to fully endorse their future trajectory.

This statement captures the dual nature of the current situation: the massive scale of 86 GW under construction, the substantial backlog of orders from hyperscale cloud providers, and the extreme pessimism emerging amid widespread market bearishness—all point to continued flexibility in the construction cycle.

However, Ng’s risk warning cannot be ignored: delays mean higher financing costs, loss of tax incentives, increased capital expenditures, and in some cases, contract termination.

For a hyperscale cloud provider with $1.7 trillion in backlog orders, time risk may be merely a rounding error; for a highly leveraged developer issuing bonds maturing in 2028 at a 12.5% interest rate, it could be a matter of life or death.

The overall market size may weather this period of policy turbulence, but some borrowers who entered with high leverage may not survive until the skies clear and the moon shines again.

Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.