Samsung Foundry has begun raising the prices of certain chip fabrication services by up to 15%, taking advantage of demand that’s filling the industry’s most advanced production lines. The increases hit 4- and 5-nanometer processes and some products made at 8 nm. The timing is especially favorable for Samsung: advanced capacity in the market has become scarce, and the company itself acknowledges growing demand from the United States and China, largely driven by AI, high-performance computing, and HBM memory.
Samsung Foundry’s price hike in 20 seconds
- Samsung is applying increases of up to 15% on certain new chip manufacturing orders.
- The hikes affect 4 nm and 5 nm nodes and some 8 nm products.
- AI demand is filling capacity both directly and through the base dies used in HBM memory.
- Samsung recently confirmed Foundry revenue growth and expects double-digit sales growth in the second half of the year.
- The advanced-capacity shortage is a chance to close the still-large gap between Samsung and TSMC.
Samsung hasn’t officially announced the new rates. The information comes from sources familiar with the negotiations, since the company doesn’t comment on prices agreed with specific clients.
According to those sources, new orders placed in July on the SF4 4 nm process saw monthly increases of 10% to 15% for clients in China and the United States. In Taiwan, the increases run 5% to 10%.
The SF5 5 nm process reportedly rose 10% to 15% too, while certain 8 nm wafers are said to have gone up nearly 10%.
Beyond the specific figures, the change shows how the AI race is starting to shift the balance in the global chipmaking market.
Saturation of advanced capacity benefits Samsung
TSMC still dominates third-party chip manufacturing by a wide margin, especially in the most advanced processes used for CPUs, GPUs, and AI accelerators.
That position has a consequence: a growing share of its capacity is committed to the enormous demand tied to AI computing.
When clients can’t find enough capacity at the Taiwanese manufacturer, they have to look elsewhere. Samsung Foundry is one of the few companies that can compete in advanced processes, along with Intel Foundry.
This doesn’t mean TSMC has “collapsed.” It’s the opposite: extraordinarily high demand is limiting the availability of certain manufacturing technologies.
Samsung is starting to benefit.
In its Q2 2026 results, Samsung confirmed its foundry business saw revenue growth driven by demand for HBM base dies and strong orders from U.S. clients. For the second half of the year, it expects double-digit revenue growth on rising demand from the U.S. and China.
Samsung also says it keeps securing new designs from major clients, including HPC projects using its 2 nm technology.
That shift matters a lot for a division that has spent years trying to shrink the huge gap with TSMC.
AI is filling factories even when it isn’t making GPUs
AI’s pressure on chip factories is broader than it looks.
Not all AI-related capacity goes straight to making GPUs.
A good example is HBM4, the new generation of high-bandwidth memory for AI accelerators.
These memories stack layers of DRAM over a logic component called a base die. Samsung uses its 4 nm process to make its HBM4 base die, and it started commercial production of these memories in February 2026.
That creates an interesting situation inside Samsung itself.
Its memory division needs capacity from Samsung Foundry to produce HBM components. At the same time, the foundry has to fill orders from outside companies chasing capacity on the same processes.
The company forecasts its HBM sales will more than triple in 2026 versus 2025, and it’s already expanding HBM4 production capacity.
So AI’s growth is squeezing several parts of the supply chain at once: DRAM memory, HBM, advanced logic processes, packaging, and custom chip fabrication.
Samsung also has an advantage few rivals can match: it holds memory, foundry, and advanced packaging technologies within one group.
Samsung is trying to turn that integration into a key selling point for winning new AI infrastructure contracts.
From losing money to seizing an AI-chip opportunity
Fuller factories can also help with one of Samsung Foundry’s long-standing problems.
A semiconductor fab needs billions in investment and carries huge fixed costs. Running lines below capacity makes it much harder to recover that.
That changes when the factories are fully loaded.
Higher utilization spreads those costs over more wafers, and if the manufacturer can also raise prices, margins can improve further.
Samsung hasn’t confirmed the foundry division will return to profitability by a specific date. So any forecast pinning that to 2027 should be treated as an estimate, not an official target.
But there are signs of improvement.
In its Q2 results, Samsung highlighted more new designs with major clients and projected double-digit Foundry revenue growth for the second half of 2026.
The company also plans to invest more than 110 trillion won in facilities and R&D during 2026, as part of its strategy to strengthen its position in AI semiconductors.
Samsung is after more than higher prices
The real prize for Samsung isn’t necessarily that 15% increase.
The shortage gives the Korean manufacturer a shot at something more valuable: getting leading chip designers to test its manufacturing processes.
Switching foundries isn’t just sending the same design to another factory. Chips have to be adapted to each process’s specific technologies, libraries, and features.
Once that work is done and production validated, a client may decide to keep some of its future products with that manufacturer.
Samsung is trying to use this moment to widen its client base in AI and high-performance computing.
In July, for instance, it announced an agreement with Broadcom to expand collaboration in memory, foundry technologies, and advanced packaging for future AI infrastructure. The two estimate the collaboration could exceed $200 billion by 2030, though that’s based on a memorandum of understanding, not guaranteed revenue.
Samsung is also progressing on 2 nm processes for HPC and preparing new generations for the coming years.
Older chips see increases too
Another important point: the increases aren’t limited to the newest nodes.
Certain 8 nm products have reportedly risen close to 10%.
These mature processes are still used in many chips where the most advanced technology would be needlessly expensive.
That shows the current pressure on the industry can’t be explained by the race to 2 nm alone.
AI is reshaping investment and production priorities across the supply chain. Manufacturers reserve resources for higher-demand, more profitable products while customers compete for capacity across different process generations.
Samsung’s own strategy reflects this.
For the second half of 2026, the company plans to raise sales of HBM base dies, speed up its second-generation 2 nm process for mobile devices, and keep winning AI and HPC projects on advanced nodes.
Samsung is also preparing for even tighter integration between memory and foundry. For HBM5, it plans to use a base die made on a 2 nm process, versus the 4 nm used in HBM4.
So AI is letting Samsung Foundry recover ground just as its main rival faces demand it struggles to absorb.
TSMC still holds a huge lead in market share, technology, and customer base. A temporary price increase at Samsung doesn’t change that.
But fuller factories, higher rates, and new designs can improve the economics of its foundry business. For Samsung, which has spent years trying to convince major chip designers that there’s a viable alternative to TSMC, the capacity shortage could turn into an opportunity it could hardly have created on its own.
Frequently Asked Questions
How much is Samsung raising chip manufacturing prices?
The available information points to increases of up to 15% on certain new orders. The exact amount varies by process and region.
Which Samsung Foundry processes are seeing price hikes?
Increases have been noted on 4 nm and 5 nm processes, plus some wafers made with 8 nm technology.
How does AI relate to these price increases?
AI is driving higher demand for accelerators, custom chips, and HBM memory. HBM also uses advanced foundry processes to make components like its base dies, adding more pressure on available capacity.
Can Samsung catch up to TSMC through this?
In the short term, a big gap remains between the two. But the high demand gives Samsung a chance to raise factory utilization and win new clients in AI and high-performance computing.