TLDR: Pat Gelsinger, recently departed CEO of Intel, has candidly admitted that the company significantly underestimated the transformative power and scale of Artificial Intelligence, a misstep that allowed rivals like Nvidia to dominate the burgeoning AI chip market. His successor, Lip-Bu Tan, has acknowledged past shortcomings and is steering Intel towards an ‘AI Everywhere’ strategy to regain market leadership.
In a significant revelation on July 5, 2025, Pat Gelsinger, who recently stepped down as Intel’s Chief Executive Officer, openly acknowledged a critical strategic misjudgment regarding Artificial Intelligence (AI). Gelsinger confessed that both he and, by extension, Intel, ‘significantly underestimated the true scale and transformative power of Artificial Intelligence,’ a miscalculation that has profoundly impacted the company’s standing in the rapidly evolving semiconductor landscape . This admission comes as Intel grapples with intense competition, particularly from Nvidia, which has surged ahead by capitalizing on the AI boom .
Gelsinger’s tenure saw Intel’s traditional focus on Central Processing Units (CPUs) dominate, leading to what he termed an ‘AI blind spot.’ He notably remarked that Nvidia CEO Jensen Huang ‘got lucky’ with AI, highlighting that Nvidia’s Graphics Processing Unit (GPU) architecture proved inherently more suitable for the parallel processing demands of AI workloads . Intel had explored similar paths years prior with projects like Larrabee, an x86-based GPU attempt, but ultimately de-prioritized them . This strategic oversight meant Intel missed critical early opportunities in the burgeoning AI and High-Performance Computing (HPC) markets, allowing companies like Nvidia to achieve unprecedented heights .
According to Gelsinger, Intel’s strategy mistakenly prioritized AI inference over model training, a crucial area where competitors were already actively developing specialized accelerators . This focus left Intel out of the race, offering only a limited number of solutions like the Gaudi accelerator line, which did not see widespread adoption in data centers . Attempts to scale solutions, such as the Falcon Shores project, were eventually curtailed . The consequences of this underestimation have been substantial, with analysts estimating that the lost time in AI could cost Intel ‘tens of billions of dollars’ . The company has faced mounting pressure from investors and intensifying competition, reporting billions in losses and a significant stock drop in the past year .
In the wake of Gelsinger’s departure, Intel’s new CEO, Lip-Bu Tan, has taken the helm, acknowledging the company’s past struggles. Speaking at the Intel Vision 2025 summit, Tan admitted, ‘There are areas [where] we have fallen short of your expectations,’ and promised a ‘new era’ for the company . Tan, who joined Intel just 14 days prior to his keynote, emphasized a return to an ‘engineering-first’ approach and a revised IDM 2.0 strategy, focusing on chip architecture and design rather than solely on developing a foundry business . He also indicated an intention to spin off non-core business units .
Also Read:
- NVIDIA CEO Jensen Huang’s Vision for AI’s Future Draws Praise from Jim Cramer Amidst Strong Market Performance
- Nvidia-Backed CoreWeave’s Explosive Growth Disrupts AI Cloud Market
Intel is now aggressively pursuing an ‘AI Everywhere’ strategy, with a renewed focus on AI PCs, edge computing, and its foundry business, aiming to regain market share and re-establish itself as a leader in the silicon-driven future . However, competitors like NVIDIA and AMD continue to increase their share of the AI market, releasing increasingly powerful and energy-efficient solutions, while Intel continues to rely on its classic server CPU Xeon, struggling to offer a competitive AI accelerator . The company hopes to return to process leadership in 2025 with the formal launch of its 18A fabrication technologies . This radical change in course is expected in the coming years as Intel attempts to navigate the highly competitive AI landscape .
