Test report DSG-9190 · Rev A · tested October 30, 2026
Foundries & ManufacturingDevice under test
TSMC Weighs Up to 20% CoWoS Price Hike as AI Packaging Squeezes
TSMC may raise CoWoS packaging prices up to 20% as Blackwell testing takes 3-4x longer than Hopper and equipment lead times stretch to eight months.
- Read
- 3 min
- Words
- 527
- Node
- 10nm
- Operator
- Elena Vasquez
Spec summary
- TSMC reportedly weighs up to a 20% CoWoS packaging price increase, per Trendforce.
- Equipment lead times have stretched from 3-6 months to as long as 8 months.
- Blackwell takes 3-4x longer to test than Hopper; Rubin may need up to 7x longer.
- GlobalFoundries signed a $2 billion, five-year deal to make CoWoS interposers, ramping in H1 2028.
- TSMC targets 227,000 wafers per month in 2027, up from over 106,000 in 2026.

TSMC is weighing a price increase of up to 20% for its CoWoS advanced packaging, and equipment lead times have stretched from three to six months to as long as eight. Trendforce reported the move, citing supply chain sources and local media. TSMC hasn't confirmed it.
For practitioners, this lands on the tightest link in the AI supply chain. CoWoS — Chip on Wafer on Substrate — combines logic dies and high-bandwidth memory (HBM) stacks in a single package, which is how AI accelerators get the bandwidth their workloads demand. That packaging is getting harder as designers cram in more compute and memory.
Nvidia's Blackwell already pairs two large GPU dies with eight HBM3E stacks. The upcoming Rubin and Rubin Ultra are expected to push integration further, which will add even more pressure on packaging and the supply chain behind it.
Why is testing becoming a bottleneck?
Blackwell reportedly takes three to four times longer to test than Nvidia's previous-generation Hopper. Rubin could need as much as seven times longer. Longer test cycles keep equipment occupied for extended periods, which makes it harder to raise output as demand grows.
GPUs aren't alone in competing for capacity. Server CPUs and custom ASICs also need advanced packaging, adding strain to an already crowded chain.
What changed in TSMC's equipment buying?
Tools take longer to arrive, and TSMC has changed how it buys them. The company reportedly placed this year's equipment orders at existing prices rather than negotiating discounts or extra features — a break from its habit of using purchasing power for better terms. Getting tools on time now matters more to TSMC than getting them cheaply.
The squeeze echoes a September report that TSMC's projected equipment needs had almost doubled in six months. Cliff Hou, TSMC's deputy co-COO, said expected equipment requirements reached 1.9 times the original estimate by July as AI demand accelerated. "Compared with the pattern 30 years ago, we have never seen the demand increase that much, that frequency of changes," Hou said at the time.
Can the supply chain broaden fast enough?
TSMC is trying. Earlier this month, GlobalFoundries agreed to produce silicon interposers for TSMC's CoWoS in the U.S. under a five-year, $2 billion contract, with production ramping in the first half of 2028. But interposers are only one part of the packaging flow, so the deal alone won't eliminate shortages.
Industry estimates cited by Trendforce put TSMC's monthly capacity at over 106,000 wafers in 2026, targeting 227,000 wafers in 2027 — more than doubling output in a single year to keep pace with increasingly complex AI processors.
How far do price increases go beyond packaging?
TSMC is reportedly considering broader hikes for 2027:
- Advanced-node wafer prices up 5% to 10%.
- Customers on 2nm and 3nm-class nodes face the steepest increases.
- Buyers seeking capacity beyond original allocations may pay premiums that push some costs up by as much as 25%.
If both the packaging and wafer increases materialize, AI hardware costs will likely climb further, and some of that will inevitably be passed down the chain.
via Tom's Hardware (Source)
More from Elena Vasquez
Show full bio
Senior reporter covering people, launches and day-to-day practice in hardware layer of artificial intelligence at Die Signal.
498 articles
Same lot · LOT-C204
- DSG-90903nmNvidia Reportedly Eyes Intel Packaging, Easing TSMC's Backend Load
- DSG-97127nmTSMC Targets CoWoS Capacity Doubling by 2028 to Ease AI Chip Bottleneck
- DSG-686245nmTSMC reportedly plans up to 20% CoWoS packaging price rise
- DSG-682365nmHBM Prices Set to More Than Double Next Year Amid Ongoing Shortage
- DSG-44147nmHBM Shortage Shifts Power to Micron, Squeezes Nvidia Margins