Intel to Build GPUs Under New Leadership of Kechichian and Demmers
At the Cisco AI Summit, Intel CEO Lip-Bu Tan announced it will produce GPUs to support AI and gaming workloads, challenging Nvidia's lead. The effort is overseen by Data Center Group head Kevork Kechichian and bolstered by former Qualcomm SVP Eric Demmers, with development in early stages pending customer-driven planning.
1. Rating Downgrade Highlights Stalled Turnaround
Following Intel’s fourth-quarter report, a leading equity research firm downgraded the stock to sell, citing a 4% year-over-year revenue decline to $13.7 billion and a 350 basis-point contraction in gross margin. The analyst noted that the company’s restructuring charges and inventory write-downs exceeded prior estimates by $200 million, eroding operating profitability. Despite cost-cutting measures projected to save $4 billion annually by 2027, the firm concluded that momentum has stalled and reinstated concerns about competitive pressures from rival chipmakers.
2. Tepid Q1 Guidance Underscores Recovery Risks
Intel’s guidance for the first quarter forecasts flat revenue sequentially, undershooting the Street by approximately $300 million. Management attributed this softness to transitioning manufacturing lines from client to server products, which it expects will constrain supply by 5–7% in the period. With R&D spend rising 8% year-over-year and foundry capital expenditures slated to increase by 15% in 2026, the cash-flow outlook appears pressured, prompting the analyst to question whether capital allocation toward new node development will yield the anticipated return on invested capital.
3. GPU Push Signals Strategic Expansion Beyond CPUs
At the recent Cisco AI Summit, CEO Lip-Bu Tan announced Intel’s formal entry into graphics processing units, appointing veteran executives Kevork Kechichian and Eric Demmers to lead the initiative. Kechichian, who oversees the data-center group, and Demmers, with 13 years at Qualcomm, will develop GPU architecture to address burgeoning AI and machine-learning workloads. Although still early stage, this move represents Intel’s effort to capture a share of the $50 billion global GPU market currently dominated by a single competitor, with product sampling set to begin in late 2027.