Investor focus in the artificial intelligence trade is moving away from hyperscale platforms and toward the companies supplying critical components, according to a new report from Jefferies.
The firm’s latest Greed & Fear analysis by Christopher Wood argues that the multiyear surge in AI spending has entered a phase where pricing power sits with memory producers rather than chip designers or cloud platforms. Memory suppliers such as SK Hynix (HXSC.F) and Micron (MU) have seen sharp gains as contract prices for advanced memory surged late last year.
Jefferies notes in the Jan. 22 report that AI-related capital spending continues to rise, even as broader non-AI investment shows signs of weakness. U.S. data show AI-linked spending on software equipment and data centers grew at a double-digit pace last year, while other forms of business investment declined. Markets have begun to price in a broader industrial and energy recovery, but the macro data have yet to fully support that view.
The report highlights a growing divergence in equity performance. Shares of major hyperscalers and AI platform leaders have lagged since late October, while memory producers and semiconductor manufacturers have advanced sharply. Jefferies estimates that memory prices jumped roughly 50% last quarter, reinforcing supplier leverage across the AI supply chain.
That leverage comes with rising costs. New fabrication plants now require investments measured in tens of billions of dollars, and Jefferies says some memory producers are pushing customers to share those costs in exchange for guaranteed supply. This marks a notable shift from earlier concerns that large chip buyers would pressure margins.
Despite the continued buildout, Jefferies cautions that the AI spending cycle is no longer in its early stages. The report compares the current environment to other capital-intensive industries where returns tend to normalize over time. The key risk is when investors begin to question whether AI spending will generate sufficient profits to justify the scale of investment.
That concern is already visible in equity markets. While chipmakers and memory suppliers have surged, several large cloud and internet companies have posted declines in recent months, even as their capital spending plans expand further into 2026. Jefferies estimates hyperscaler capex will continue to rise sharply this year, increasing pressure to show returns.
The report also flags rising energy requirements as a structural issue tied to AI growth. Some technology companies are moving to secure power supplies directly, signaling a shift toward more asset-heavy business models.
Overall, Jefferies says the AI trade remains intact, but leadership within it is changing. Memory suppliers are emerging as near-term winners, while investors are becoming more selective about companies committing vast sums to AI infrastructure without clear evidence of profitability.