AI infrastructure capex fueling data center rally

The race to build out AI compute capacity is driving unprecedented valuations across the data center and AI infrastructure stack. Cerebras Systems, a maker of AI chips designed for large language model training, has hiked its IPOInitial Public Offering - a company's first public sale of stock. price range to $150-$160 per share, up from an initial $115-$125 range, reflecting explosive institutional demand. The move signals that Wall Street sees AI infrastructure as a secular growth story with years of runway.

CoreWeave, a startup focused on GPU cloud services for AI training, has become a linchpin in the narrative. Nvidia has publicly stated it helped CoreWeave exist and backs its mission; Goldman Sachs and Wells Fargo have both reset price targets on CoreWeave higher after analyzing the scale of AI capex spending. The implication is that pure-play infrastructure providers may capture more durable value than chip makers alone, given the competitive pressures and commoditization risks in semiconductors.

Nvidia's own positioning is evolving: the company is investing heavily in adjacent infrastructure plays like CoreWeave and smaller vertically-integrated compute providers to ensure customer lock-in and defend against AMD's inroads. This creates a multi-layer ecosystem where Nvidia supplies chips, but third-party infrastructure firms monetize them. Traders are rotating between chip makers and infrastructure bets as they recalibrate exposure to the AI capex narrative.

Skeptics worry that the infrastructure spend is front-loading a capex supercycle that will eventually plateau. They also flag that Chinese competitors are building alternative stacks outside Nvidia's orbit, which could fragment the market. Nonetheless, the scale of announced capex commitments from hyperscalers and the backlog of GPU demand suggest the buildout phase is still in its early innings.