1 hour ago
18
Ajit Manocha, President and CEO of SEMI (Semiconductor Equipment and Materials International), is making a bold call: this chip cycle doesn’t follow the old rules.
The numbers backing his confidence
Global semiconductor revenues hit $368 billion in Q2 2026, a 35% jump from the prior quarter and a 104% gain year-over-year.
Forecasts for full-year 2026 growth range from 90% to 112%, with some projections putting the total market near $1.6 trillion by year end.
AI-driven memory growth at major suppliers has been running at 50% to 76% quarter-over-quarter as of Q2 2026, with high-bandwidth memory (HBM) sitting at the center of that demand surge. HBM is essentially the plumbing that lets AI chips process enormous datasets at speed. Without it, frontier AI models don’t run. With it, every major hyperscaler on Earth is spending aggressively to secure supply.
Over 100 new fabrication plants are currently under construction or in planning stages worldwide. Building a leading-edge fab costs tens of billions of dollars and takes years to complete.
Why this cycle feels different
Manocha’s argument is that several structural forces are now dampening the amplitude of those swings.
First, hyperscaler capital expenditure. Amazon, Microsoft, Google, and Meta are collectively committing hundreds of billions of dollars annually to AI infrastructure. That spending flows directly into chip demand, and unlike consumer electronics cycles, enterprise infrastructure buildouts don’t reverse overnight.
Second, HBM requirements are creating a supply constraint that keeps prices elevated even as production scales. AI training runs are memory-bandwidth-limited, meaning the faster AI models get, the more HBM they consume per unit of compute.
Third, geographic diversification of manufacturing is adding resilience. Government-backed fab investments across the United States, Europe, Japan, and India are spreading capacity across jurisdictions. The CHIPS Act in the US, similar legislation in the EU, and Japan’s renewed domestic semiconductor push are all expressions of the same policy calculation: chip supply is now a national security issue, not just an industrial one.
Manocha’s career arc runs from Bell Labs through NXP/Philips to GlobalFoundries, covering research, design, and manufacturing across multiple technology generations. He has personally navigated the industry through four full boom-bust cycles.
What it means for the market and what could still go wrong
Risks exist and they are real. Supply chain vulnerabilities, particularly around rare earth materials and advanced lithography equipment, remain structural weak points. Export controls on advanced chips and equipment to certain geographies create both regulatory uncertainty and potential demand fragmentation. If hyperscaler capex moderates, whether due to a slower-than-expected AI adoption curve or a broader economic slowdown, the demand picture shifts faster than the capacity that has already been committed.
The surge in AI chip demand has attracted new entrants beyond Nvidia, with AMD, Intel, and a wave of custom silicon efforts from the hyperscalers themselves all vying for a share of the market.
Manocha’s call is a macro one: the cycle structure has changed, the floor is higher, and the amplitude of downturns is likely to be smaller than historical norms.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·