Broadcom sounds alarm as TSMC capacity constraints threaten AI chip supply through 2026

TSMC Capacity Reaches Breaking Point

Broadcom has raised concerns about mounting supply constraints affecting the semiconductor supply chain, with manufacturing partner TSMC hitting production capacity limits that threaten to choke AI chip supply through 2026. Natarajan Ramachandran, director of product marketing in Broadcom's Physical Layer Products division, told reporters that TSMC's capacity—which he would have described as "infinite" just a few years ago—has become a critical bottleneck1

1

. The Taiwanese chipmaker has outlined capacity expansion plans extending through 2027, but the constraint is already affecting the broader tech industry in 20262

2

.

The world's largest contract chipmaker, whose major customers include Nvidia and Apple, acknowledged in January that capacity was tight as the boom in AI infrastructure buildout has absorbed much of its advanced production lines3

3

. TSMC stated it was working to narrow the gap between supply and demand, but the squeeze is particularly acute for cloud providers and AI developers dependent on leading-edge nodes for accelerators and custom silicon1

1

.

Ripple Effects Beyond Semiconductors

The supply constraints extend far beyond semiconductors, creating unexpected bottlenecks across multiple technology supply chains. Ramachandran revealed that despite multiple suppliers in the industry, there is a definite constraint in the laser space affecting production. Perhaps most surprising is the emergence of printed circuit boards as an unexpected bottleneck, with both Taiwanese and Chinese suppliers facing production capacity limits1

1

.

Lead times for PCBs used in optical transceivers have stretched dramatically from approximately six weeks to six months, illustrating how demand for artificial intelligence hardware is reshaping manufacturing timelines across the entire ecosystem2

2

. The growing scarcity affects wafers, lasers, and boards simultaneously, creating compounding challenges for companies attempting to scale AI infrastructure.

Industry Shifts to Multi-Year Supply Commitments

The soaring demand for AI chips is prompting fundamental changes in procurement strategies across the technology sector. Many customers are now entering long-term agreements with suppliers to secure guaranteed access to production, with commitments spanning three to four years3

3

. For buyers, such agreements effectively reserve portions of future output during a period when tightness affects multiple components. For suppliers, they provide clearer demand visibility, supporting heavy capital investment in new production lines and advanced processes1

1

.

This trend mirrors broader industry movements, with Samsung Electronics announcing last week that it is working with major clients to adopt three-to-five-year supply contracts2

2

. The shift reflects both customers' desire for longer-term supply security and chipmakers' efforts to guard against demand swings while justifying massive infrastructure investments. As the AI boom continues to strain global manufacturing capacity, these extended commitments may become standard practice for securing access to advanced AI chips and critical components through the remainder of the decade.