Summary of Key Points
Intel has laid off nearly 47,000 employees over the past three years (a 35.5% reduction), and even its recently growing “Data Center and Artificial Intelligence Group (DCAI)” has started to cut staff. This is not simply a matter of having too many people; it reflects a lack of strategic clarity. As a giant in the “Integrated Device Manufacturing (IDM)” chip industry, Intel operates three segments: traditional x86 products, wafer foundry services, and new businesses such as GPUs and AI. However, it lacks a cohesive strategy, leading to inefficiencies between its product development and manufacturing divisions. The company faces a dilemma: should it completely transition to a “fabless” model (designing chips without manufacturing them) or continue with the IDM approach? The real challenge is not the number of layoffs but finding a clear strategic direction.
I. Behind the Layoffs: It’s About Strategic Prioritization, Not Just Overstaffing
Many believe Intel is laying off due to having too many employees, but the truth is that it is assessing which businesses are worth sustaining:
- Growth segments are also being affected: DCAI, one of Intel’s few growing divisions (with revenue of $5.1 billion in Q1 2026, a 22% increase year-over-year), has seen layoffs. This indicates that the adjustments are not about eliminating redundancy but cutting non-core business lines or teams, such as unprofitable AI chip projects.
- The reality of employee numbers: Intel currently has 85,000 employees, more than the combined total of AMD’s 31,000 and Nvidia’s 42,000. However, this is not solely due to bureaucracy. As an IDM company, it needs to manage its own wafer factories, including construction, operation, and equipment maintenance. In contrast, AMD and Nvidia are fabless companies that outsource manufacturing to third parties like TSMC.
- Strategic dispersion is the issue: Intel manages three separate entities: a product development company, a foundry company, and a new business division. Each requires significant investment without a unified strategy, resulting in wasted resources. For example, its foundry division relies heavily on internal orders (17.8 billion in 2025, with 17.5 billion coming from internal customers), meaning the product development division’s profits are used to cover the foundry’s losses.
II. The IDM Model: A Double-edged Sword
The IDM model was once Intel’s strength, but it has become a liability:
- Manufacturing losses: In 2025, the foundry division incurred a loss of $10.3 billion with an operating margin of -58%. This is because its production capacity is primarily used to support its own products, without the cost-sharing benefits from external customers like Apple and Nvidia. TSMC’s profitability (with 90,000 employees) comes from a global customer base that supports its factories.
- Conflict between product and manufacturing: The product development division wants to use advanced manufacturing processes (like TSMC’s) to compete in the market, while the foundry division prefers to use internal processes to maintain its capacity. This leads to inefficiencies, as products using internal processes may fall behind those from AMD, and using external processes leaves foundry capacity underutilized.
III. Industry Comparison: Intel Like a Retreatant IBM, AMD Like the Intel of the Past?
- Intel vs. IBM: Both companies once defined their respective eras (IBM with computers, Intel with x86 chips), but they no longer lead the industry. IBM survived by transitioning to software services after selling its hardware business. Without a high-profit software segment and if it loses its CPU competitiveness and abandons manufacturing, it could become another ordinary chip designer.
- AMD vs. the Past Intel: AMD is now similar to Intel in 2005—with a clear product strategy and customer confidence that drives competitors (like Intel) to follow its lead. AMD’s success stems from separating its manufacturing division in 2009, freeing it from competing in both product development and manufacturing.
IV. The Future Path: The Appeal of Fabless and Its Challenges
There is speculation about whether Intel will fully transition to a fabless model. The pros include:
- Benefits of Fabless: Using the best manufacturing processes (e.g., TSMC’s 3nm) to quickly enter new markets; reducing capital expenditure on building factories; streamlining operations and focusing on product competitiveness.
- Challenges of Fabless: Losing the unique asset of in-house manufacturing (only TSMC, Samsung, and Intel can produce advanced processes); relying on TSMC for capacity when demand is high (AMD and Nvidia are already major customers); and facing political resistance from the U.S. government, which views Intel as a strategic player for maintaining domestic advanced manufacturing.
V. The Most Probable Path: Hybrid IDM (IDM Lite)
A hybrid IDM model, where Intel retains some advanced manufacturing capabilities (e.g., 18A processes) while outsourcing certain products to TSMC, could be a viable solution. However, avoiding the temptation to try to do everything internally is crucial. If the foundry division still lacks external customers and continues to use internal processes for products, it will only perpetuate inefficiencies.
VI. The Real Challenges: Organizational Habits Hindering Strategy
Layoffs can reduce costs, but they do not address fundamental strategic issues. Intel needs to overcome three persistent problems:
1. The obsession with trying to dominate all markets: It cannot lead in every segment (client CPUs, server CPUs, GPUs, AI, foundry services) simultaneously and must focus on areas where it has a competitive advantage.
2. The blame-game between product and manufacturing: The product development division should have the freedom to choose manufacturing processes, and the foundry division needs external customers to validate its capabilities.
3. Relying on past success: While the x86 ecosystem and customer base provide stability, they cannot guarantee future growth. In the AI era, Nvidia sets the standards, and Intel cannot rely on past achievements.
Conclusion
Whether 80,000 employees are too many depends on Intel’s vision for its future role. For a pure design company, it might be excessive; for an integrated company with manufacturing and new businesses, it may not be enough. The key is whether these employees are working towards a shared strategic goal. If Intel aims to be a product development company, it should outsource manufacturing and compete directly with AMD and Nvidia. If it wants to be a foundry company, it must secure external customers for its foundry division. Regardless of the number of layoffs, the real challenge is to align resources around a clear strategy.
Layoffs can alleviate staffing issues, but they cannot solve strategic uncertainties. Intel’s future depends on whether it can overcome these organizational barriers and find a sustainable path forward.