瑞银研报看好2027年AI数据中心基建需求，点名卡特彼勒、康明斯、Quanta为受益方

UBS expects demand for artificial intelligence infrastructure to accelerate in 2027, creating opportunities for industrial equipment manufacturers and engineering companies as data center developers expand capacity despite persistent supply chain constraints. In an Oct. 16 research report, UBS analyst Steven Fisher, CFA, said industry experts attending the Data Center World Expo in Singapore reported accelerating orders, growing multiyear capacity reservations and confidence that AI infrastructure investment will continue beyond 2027. The findings suggest that spending on data centers remains a significant growth driver for industrial companies even as the broader economy faces inflationary pressures. Power generation and construction companies stand to benefitUBS identified Caterpillar ( CAT ), Cummins ( CMI ) and Quanta Services ( PWR ) as beneficiaries of continued data center construction. Caterpillar and Cummins are positioned to capitalize on demand for power generation equipment, while Quanta stands to benefit from electrical grid investment. Caterpillar told the UBS team that demand for power generation equipment remains exceptionally strong and that it is expanding production capacity. The report also highlighted growing interest in behind-the-meter power generation, which allows data centers to generate electricity on-site rather than relying entirely on utility grids. UBS expects most projects through 2028 or 2029 to remain primarily grid-connected with traditional diesel backup systems. The eventual mix of on-site generation, energy storage and grid connections will determine how much equipment manufacturers benefit. Supply shortages remain a constraint Industry experts identified equipment availability and manufacturing capacity as major obstacles to data center expansion. Suppliers of cooling systems, power equipment and generation technology are competing to shorten delivery times and increase production. Customers are increasingly favoring prefabricated and modular systems to accelerate construction and reduce labor requirements. The industry is also moving toward 800-volt direct-current power distribution, with deployments expected between 2027 and 2029. UBS said the technology should reduce copper requirements, heat generation and the complexity of power distribution systems. Growing share of construction UBS estimated that data centers directly represent approximately $63 billion of the $1.3 trillion U.S. nonresidential construction market, or about 5%. However, including indirect spending on manufacturing facilities, power infrastructure and related projects, the firm estimated that data centers drive 20% to 30% of overall nonresidential construction activity and much of its growth. The strength of that spending contrasts with concerns about the broader industrial economy. UBS noted that inflation and higher interest rates associated with the Iran conflict have complicated expectations for an industrial recovery. The firm expects higher input costs to become more apparent in fourth-quarter results, particularly for companies exposed to fuel and materials. Caterpillar's power margins in focus Caterpillar's ( CAT ) Power & Energy segment exceeded second-quarter profit expectations, supported by incremental margins of 39%. UBS said consensus forecasts call for incremental margins to remain elevated in the third and fourth quarters, although they are expected to moderate from the second-quarter level. Caterpillar has cautioned that the unusually favorable incremental profitability recorded in the second quarter may not persist. Investors are also questioning whether industrial stocks can continue outperforming if AI-related order growth slows. Fisher argued that a material slowdown or flattening in Caterpillar's sequential backlog growth could make further share-price outperformance more difficult, even if absolute backlog levels remain high. AI investment risks UBS acknowledged that the AI infrastructure boom could eventually slow as technology advances reduce computing costs and smaller AI models become more widely adopted. Such developments could pressure AI companies' profitability and reduce demand for new data centers. The firm nevertheless expects technological adoption to create additional opportunities for industrial suppliers, while cautioning that investors should monitor the sustainability of infrastructure spending. The report also raised questions about the financial stability of newer cloud computing providers following a selloff in CoreWeave ( CRWV ) bonds. UBS said industrial suppliers generally don't disclose enough information to determine their exposure to these customers compared with established hyperscale cloud operators. The distinction could become increasingly important as investors assess the durability of equipment backlogs and future orders.