SKN | UBS Expands AI Infrastructure Coverage as Power Access Becomes the Key Constraint

Key Takeaways:

• UBS has initiated coverage of five AI data-center colocation companies, framing them as potential beneficiaries of constrained computing capacity and accelerating AI adoption.
• Analyst Ryan Gravett identified power access as the primary bottleneck limiting the speed at which new AI computing capacity can come online.
• UBS estimates the companies already have approximately 5 gigawatts of leased capacity, representing roughly $125 billion in contracted revenue, with another 12 gigawatts of potential expansion.
• The bank projects approximately 80% average annual EBITDA growth across the coverage group from 2027 through 2029 and sees an average 36% upside across the five stocks.

UBS has expanded its investment-research coverage into a specialized segment of the artificial intelligence infrastructure market, initiating coverage of five data-center colocation companies. The move reflects the bank’s view that the next stage of AI expansion is increasingly constrained not by demand for computing, but by the physical infrastructure required to deliver it.

UBS Identifies Power Access as the Critical Infrastructure Constraint

Analyst Ryan Gravett initiated coverage of Hut 8 Mining, TeraWulf, Applied Digital, Core Scientific and Cipher Digital with bullish ratings. His central argument is structural: AI adoption is increasing the requirement for data-center capacity while the availability of suitable power connections remains limited.

That distinction is important to UBS’s thesis. In its broader research, the bank has increasingly highlighted electricity availability as a determinant of where AI infrastructure can be developed and how quickly capacity can be delivered. For data-center operators, access to power can therefore represent a strategic asset rather than simply an operating cost.

UBS Sees Existing Grid Access as a Strategic Advantage

Gravett’s research highlights an unusual characteristic of the five companies: their previous cryptocurrency-mining operations helped establish access to electricity infrastructure in locations suitable for large-scale computing.

According to UBS, this legacy has already translated into approximately 5 gigawatts of leased capacity, corresponding to roughly $125 billion in contracted revenue. The group also has approximately 12 gigawatts of additional expansion potential, giving UBS a basis for viewing these operators as infrastructure platforms rather than simply former crypto-mining businesses.

What UBS Is Pricing Into the Infrastructure Model

UBS forecasts approximately 80% average annual EBITDA growth from 2027 to 2029 across the five companies, around 10% above the Street’s 2029 expectations. Despite the stocks having fallen roughly 45% from recent highs, UBS calculates average upside potential of approximately 36% from its initiation levels.

The more significant point for sophisticated investors is the infrastructure economics behind the forecast. UBS is effectively arguing that secured power, contracted capacity and the ability to convert existing infrastructure into AI-oriented facilities can become increasingly valuable as hyperscalers and AI developers compete for computing capacity.

For globally diversified wealth, UBS’s research also illustrates where AI infrastructure risk is moving. The constraint is becoming increasingly physical: electricity, transmission, permitting and data-center construction can determine whether enormous technology spending translates into usable computing capacity. UBS’s initiation therefore places greater emphasis on the infrastructure layer of the AI economy, while leaving execution, financing and tenant concentration as important variables to monitor. For a confidential discussion regarding your cross-border banking structure, technology exposure or international wealth strategy, contact our senior advisory team.