Written by: Rita

The governors of Texas and Pennsylvania signed executive orders in August tightening regulations, energy, and environmental thresholds for data center development. Goldman Sachs utility analysts believe this has minimal impact on high-quality, large-scale projects, with the real victims being speculative projects with insufficient capital. SMCI delivered its F4Q26 financial report on August 11, with FY27 revenue guidance of $65 to $72 billion, and F4Q quarterly orders exceeding $60 billion, of which about 70% are related to pure AI deployments.

Demand for AI infrastructure has not slowed due to tighter regulations. Global announcements of AI projects in August remained intensive, from IBM's $240 million collaboration with Together AI to the launch of a sovereign AI platform in Saudi Arabia, and a multi-billion dollar agreement between CoreWeave and quantitative giant Hudson River Trading. In Goldman Sachs' AI project monthly tracking report released on August 31, key developments in the global AI infrastructure sector for the month were summarized. AI computing power is concentrating at the top, and the participation of sovereign nations is also accelerating.

Texas and Pennsylvania tighten regulations, Dell and Supermicro are most affected

On August 3, the governor of Texas signed an order tightening energy and environmental requirements for data center development. Pennsylvania followed suit on August 18. Goldman Sachs utility analyst Carly Davenport believes the policy changes have minimal actual impact on high-quality, large-scale projects, with the main effects felt by speculative projects with insufficient capital.

The Goldman Sachs U.S. IT hardware team noted that Dell and Supermicro are most exposed in greenfield projects serving emerging AI data centers. Supermicro's F4Q quarterly orders exceeded $60 billion, with FY27 revenue guidance of $65 to $72 billion, a 75% year-over-year increase. About 70% of this is related to GPU servers. Supermicro's performance also sends positive signals for Dell and HPE, with Dell benefiting from the sustained strong demand for GPU servers and HPE benefiting from the recovery in CPU server demand.

Sovereign AI projects launched globally

In August, sovereign AI infrastructure construction advanced simultaneously in multiple regions around the world. Saudi Arabia was the most active market that month.

On August 31, Saudi sovereign fund PIF-supported AI company HUMAIN, together with AMD and Cisco, launched an open large-scale AI platform in Riyadh, using AMD Instinct MI355X GPUs and Cisco Silicon One networking solutions. The next phase plans to deploy 250 megawatts, aiming for 1 gigawatt by 2030.

On August 24, French AI company Mistral signed a multi-million euro strategic cooperation agreement with HUMAIN to jointly develop sovereign AI infrastructure and build Arabic large models for regulated industries.

The Indian market was similarly active. On August 26, IntelliDB Enterprise partnered with Yotta Data Services to provide sovereign AI database infrastructure in India. Yotta previously announced the deployment of 20,736 NVIDIA Blackwell Ultra GPUs, with 10,000 allocated to the India AI Mission. Supermicro is one of Yotta's AI infrastructure partners.

On August 8, U.S. AI cloud infrastructure company Firebird announced the first phase of Armenia's sovereign AI infrastructure went live, with a first-phase investment of $500 million, planning for more than 400 megawatts, using Dell PowerEdge servers and NVIDIA Blackwell GPUs. In the UK, AI infrastructure startup Era4 partnered with HPE to deploy a modular data center network in the UK, using HPE’s AI modular solutions and NVIDIA Blackwell B300 GPUs, starting with an initial GPU inventory of 1,500, expanding to over 10,000 within 12 months.

New projects and cooperation agreements continue to be announced

IBM and Together AI signed a $240 million multi-year agreement on August 11 to deploy large-scale AI inference clusters on IBM Cloud, using NVIDIA HGX B300 systems and Spectrum-X Ethernet, expected to go live in the first quarter of 2027.

Fermi signed a binding lease agreement with AI cloud service provider TensorWave on August 10 to deliver a 222-megawatt data center solution in Carson County, Texas, with a total contract value of approximately $6.5 billion over 15 years, expected to start delivery in the second half of 2027. Fermi has the right to expand to 650 megawatts, using AMD Instinct GPUs and multiple power sources.

On August 25, Cisco and Supermicro reached a strategic cooperation agreement to expand the Secure AI Factory program, providing liquid-cooled and air-cooled AI infrastructure to support trillion-parameter model training, with Supermicro's solutions set to be sold through Cisco channels starting in October 2026.

On August 20, quantitative trading company Hudson River Trading signed a multi-year multi-billion dollar agreement with CoreWeave to deploy NVIDIA Vera Rubin NVL72 and HGX B200 GPU systems on the CoreWeave cloud platform for AI-driven trading research.

On August 1, Indian fintech company ZyBiSys completed the deployment of Dell unified infrastructure, achieving 99.99% availability across six data centers in India, supporting real-time trading for over 50 brokerage firms.

Goldman Sachs’ assessment is clear. The investment boom in AI infrastructure continues to accelerate, and tightened regulations have not substantially hindered the expansion pace of leading players. Demand for server hardware remains strong, and the participation of sovereign AI is spreading from Saudi Arabia and India to more regions. AI computing resources are concentrating among larger players with capital, technology, and compliance capabilities, while the space for small speculative projects is being squeezed.

Disclaimer

This article is a compilation and interpretation by Chao Xiang Research of a third-party brokerage report (Goldman Sachs, August 31, 2026), combined with publicly available market information. The ratings, target prices, profit forecasts, and related judgments cited in this article are the opinions of the analysts of that brokerage and represent only the position of their organization, not the views of Chao Xiang Research, and do not constitute any investment advice.

Markets involve risks, and decisions should be made independently. This article should not be used as the basis for buying or selling any securities.