New Street Research: Intel $INTC Target Price of $125, Server CPU Revenue Could Reach $50 Billion by 2030 — The CPU Abandoned by AI Is Now Being Rediscovered by Agents On Monday, Intel $INTC rose approximately 4%, closing near $90.50. The stock’s movement was driven by a buy rating and a $125 price target from renowned tech analyst Pierre Ferragu of New Street Research. This target is based on an expected EPS of $7.10 in 2030, valued at a 25x P/E multiple—implying roughly 38% upside from the current price. Meanwhile, Intel announced its acquisition of Israeli quantum software company Classiq and is set to report its third-quarter earnings this Thursday. These developments combined have made Intel one of the most discussed semiconductor stocks on X. Ferragu’s core argument is counterintuitive. Over the past three years, the market has widely assumed that the AI era belongs to GPUs, with CPUs sidelined—and Intel as the biggest loser. But he argues that AI agents are reshaping this dynamic. Unlike training large models, AI agents performing tasks require frequent tool invocation, database access, code execution, and complex logical reasoning—all strengths of CPUs. The more agents deployed, the greater the demand for server CPUs in data centers. According to his estimates, the server CPU market could exceed $220 billion by 2030; if Intel captures 20% to 25% of that market, its server CPU revenue could reach approximately $50 billion. Capacity is another key pillar. Ferragu expects Intel’s server CPU production capacity to increase by more than 50% by 2027, driven by ramp-up of Intel 3 process technology, mass production of the 18A node, and reallocation of some PC chip capacity toward server chips. This is critical: if CPU demand truly rises and capacity keeps pace, Intel stands to benefit on both pricing and market share. The acquisition of Classiq adds another dimension. Intel has not disclosed the deal price, but Israeli media Calcalist and Bloomberg report it is approximately $2.3 billion, paid in cash and stock, with completion expected in the first half of 2027. While commercial quantum computing remains distant, this move signals Intel’s strategic investment in software capabilities for next-generation computing. The bull and bear cases are clearly defined. Bulls argue that Intel’s current valuation still reflects the narrative of an “AI loser,” and if server CPU demand is reignited by AI agents, earnings recovery could be substantial. Bears counter that AMD $AMD has significantly increased its server CPU market share, and Arm-based custom chips are rapidly gaining traction among cloud providers—making it difficult for Intel to maintain over 20% market share. Moreover, the $7.10 EPS forecast for 2030 is over four years away, leaving ample room for uncertainty around process yields, capital expenditures, and execution. Intel has repeatedly delayed process node rollouts in recent years, and skepticism about its execution capability persists. Historically, the relationship between CPUs and accelerators has reversed multiple times. GPUs were once merely graphics cards before becoming primary compute engines. Now, as AI’s focus shifts from training to inference and agents, will CPUs return to center stage? Ferragu is betting exactly on this shift. The next three developments to watch closely: Intel’s Q3 earnings report this Thursday—particularly revenue and guidance for its data center business; progress on 18A process yields and external foundry customers; and whether cloud providers mention increased CPU demand driven by AI agents in their earnings calls. Will AI agents truly restore CPUs to relevance? Do you see Intel as a value opportunity in this cycle—or just another “cheap for a reason” trap?