Taiwanese chipmaker Nanya plans $6 billion in spending in 2027, riding AI boom

[Daniel Nenni]

- Projected 2027 spending is four times this year's total
- AI revolution driving up memory chip demand
- Nanya customers include Google, Nvidia, Qualcomm

TAIPEI, July 10 (Reuters) - Taiwanese memory chipmaker Nanya Technology (2408.TW), opens new tab said on Friday it plans capital spending ‌of more than T$200 billion ($6.2 billion) next year, roughly four times this year's figure, amid soaring demand for memory chips as it rides an AI boom.

President Pei-Ing Lee told an online press briefing that the preliminary expenditure plan aims to help ramp up spending on a new plant, although ⁠the budget has yet to receive board approval.

Lee was speaking after Nanya reported unaudited second-quarter revenue of T$82.55 billion, up 684% from a year earlier. The company's net income surged 1,324% to T$50.19 billion, while gross margin improved to 79.5% from a negative 20.6% a year earlier.

Nanya, whose customers include Nvidia (NVDA.O), opens new tab, Qualcomm (QCOM.O), opens new tab and Google (GOOGL.O), opens new tab, expects to spend more than T$50 billion this year, Lee said. Total investment in the new plant will reach about T$480 billion at full production capacity, he ‌added.

The first phase of the new plant is scheduled to reach capacity of 30,000 wafers per month in 2028, eventually expanding to 45,000 wafers per month.

Lee said structural changes driven by artificial intelligence were supporting a stronger long-term outlook for the ⁠memory industry, adding that the current supply shortage was expected to persist for several more quarters.

Global memory makers, including Samsung Electronics (005930.KS), opens new tab and SK Hynix (000660.KS), opens new tab, are ramping up investment to meet ⁠surging AI-driven memory demand.

Commenting on South Korea's push to expand semiconductor production, Lee said such efforts were positive for the industry's broader ecosystem and reflected confidence ⁠in market demand.

Shares in Nanya, which has a market value of around $47 billion, were not trading on Friday as Taiwan's stock market was closed due to a typhoon.

https://www.reuters.com/world/asia-pacific/taiwanese-chipmaker-nanya-plans-6-billion-spending-2027-riding-ai-boom-2026-07-10/

 

[jms_embedded]

I wish them luck, but the DRAM market is cutthroat, and historically weeds out those who can't survive the bottom of the cycle. Qimonda and Elpida died after the last big bust in 2006-2009. Nanya's been hanging on as a distant fourth to the Big Three Samsung/SK Hynix/Micron. (Huh. I guess Powerchip is still around too, as an odd-duck memory foundry.)

I quote Jeho Lee's article here (The Chicken Game and the Amplified Semiconductor Cycle: The Evolution of the DRAM Industry from 2006 to 2014):

Industry fluctuations in the supply of DRAM chips relative to demand have been characterized by what is called “the silicon cycle.” In the period between 2006 and 2008, the DRAM industry experienced an unusually sharp transition from a shortage of DRAM products to an extreme oversupply, culminating with the crash of DRAM prices in 2008. The industry’s overcapacity was preceded by a mad race to expand capacity; this race has been dubbed as the “chicken game” in the media. Even in the time of plunging DRAM prices, players preferred not to reduce their output. The amplified industry cycle accelerated the exit of financially vulnerable firms. I argue that the combination of the amplification of cycle and rising entry barriers fosters the transition of an industry to an oligopoly, in which cyclicality is curbed and the positions of market leaders are solidified.

 

[Daniel Nenni]

I wish them luck, but the DRAM market is cutthroat, and historically weeds out those who can't survive the bottom of the cycle. Qimonda and Elpida died after the last big bust in 2006-2009. Nanya's been hanging on as a distant fourth to the Big Three Samsung/SK Hynix/Micron. (Huh. I guess Powerchip is still around too, as an odd-duck memory foundry.)

I quote Jeho Lee's article here (The Chicken Game and the Amplified Semiconductor Cycle: The Evolution of the DRAM Industry from 2006 to 2014):

I agree but I feel there is value in having logic and memory chip production in country. Hopefully Taiwan can consume the capacity internally.

 

[fatfat]

Nanya got 2.5 billions investment from Kioxia, SanDisk, SK Hynix, and Cisco from private placement at the end of March.
The proceeds would be ‌used to invest in factory facilities and production equipment for advanced memory manufacturing:

Nanya Technology shares surge 10% after $2.5 billion fundraising

Shares of Taiwanese memory chip maker Nanya Technology opened limit-up 10% ‌on Thursday after raising about $2.5 billion in a private placement ‌from SanDisk Technologies and others to expand advanced chip production. The fundraising comes as chipmakers boost production capacity and their...

finance.yahoo.com

Plus the profits from selling memory this year, they probably can afford the 6 billion spending in 2027.

 

[Fred Chen]

Their 1C/1D looks comparable to the G4 process from CXMT.

It looks like CXMT already surpassed them in volume at this point, but are serving mainland Chinese customers mainly.

Another question is whether Nanya will move toward a 4F2 architecture that it had worked on with Kioxia.

 

[Barnsley]

As a customer of ours , more power to them!

 

[coldsolder215]

Their whole schtick is legacy DRAM, technologies whose lines were shut down years ago with enough customers who don't want to redesign around their socket. I think their latest fab is their own DDR4, but it's not accurate to say they're playing the conventional memory game and trying to compete with the cartel. It's more like the supply crunch is making everyone desperate enough to regress their memory system.

 

[hist78]

View attachment 4858

- Projected 2027 spending is four times this year's total
- AI revolution driving up memory chip demand
- Nanya customers include Google, Nvidia, Qualcomm

TAIPEI, July 10 (Reuters) - Taiwanese memory chipmaker Nanya Technology (2408.TW), opens new tab said on Friday it plans capital spending ‌of more than T$200 billion ($6.2 billion) next year, roughly four times this year's figure, amid soaring demand for memory chips as it rides an AI boom.

President Pei-Ing Lee told an online press briefing that the preliminary expenditure plan aims to help ramp up spending on a new plant, although ⁠the budget has yet to receive board approval.

Lee was speaking after Nanya reported unaudited second-quarter revenue of T$82.55 billion, up 684% from a year earlier. The company's net income surged 1,324% to T$50.19 billion, while gross margin improved to 79.5% from a negative 20.6% a year earlier.

Nanya, whose customers include Nvidia (NVDA.O), opens new tab, Qualcomm (QCOM.O), opens new tab and Google (GOOGL.O), opens new tab, expects to spend more than T$50 billion this year, Lee said. Total investment in the new plant will reach about T$480 billion at full production capacity, he ‌added.

The first phase of the new plant is scheduled to reach capacity of 30,000 wafers per month in 2028, eventually expanding to 45,000 wafers per month.

Lee said structural changes driven by artificial intelligence were supporting a stronger long-term outlook for the ⁠memory industry, adding that the current supply shortage was expected to persist for several more quarters.

Global memory makers, including Samsung Electronics (005930.KS), opens new tab and SK Hynix (000660.KS), opens new tab, are ramping up investment to meet ⁠surging AI-driven memory demand.

Commenting on South Korea's push to expand semiconductor production, Lee said such efforts were positive for the industry's broader ecosystem and reflected confidence ⁠in market demand.

Shares in Nanya, which has a market value of around $47 billion, were not trading on Friday as Taiwan's stock market was closed due to a typhoon.

https://www.reuters.com/world/asia-pacific/taiwanese-chipmaker-nanya-plans-6-billion-spending-2027-riding-ai-boom-2026-07-10/

Nanya is in a very interesting, unexpected, and profitable position today.

1. 1. Due to delays in the development of its in‑house DDR5 process, Nanya’s DDR5 entered high volume production only in 2024, rather than through licensing from Micron. Additional DDR5 capacity is scheduled to come online in late 2027. With the rapid growth of AI exhausting the capacity of the top three memory makers, Samsung, SK Hynix, and Micron, the market urgently needs more supply. Apart from those top three, Nanya is the only company outside mainland China capable of providing additional DDR5 capacity.
2. 
   
3. 2. As Samsung, SK Hynix, and Micron have reduced or halted DDR4 or DDR5 production to focus on the more profitable HBM memory, Nanya’s DDR5 and DDR4 capacity has suddenly become critical.
4. 
   
5. 3. Many products used in today’s automotive, EV, storage (HDD, DDS), networking/communication, and consumer electronics sectors still rely on DDR4 because it remains sufficient (often more than sufficient) and cost effective. Products such as SK Hynix’s SSDs, Western Digital’s hard drives, SanDisk’s SSDs, Cisco's enterprise networking switches, and Tesla’s EVs all need DDR4. Technically, they could migrate to DDR5, but doing so may not be cost justifiable and could prove a dead end under current market constraints.
6. 
   
7. 4. SK Hynix’s strategy differs markedly from Samsung’s. Rather than pursuing everything in‑house, SK Hynix has patiently built collaborations, alliances, and joint ventures with other companies. Its investments in Kioxia and Nanya, along with its early partnership with TSMC to integrate HBM memory into AI systems, are prime examples. Collaboration with TSMC is likely one of the key reasons SK Hynix has grown far faster than anyone expected three or four years ago.

At today’s close, SK Hynix’s market capitalization stands at US $874.9 billion. It is not far behind Samsung’s US $1.116 trillion.

 

[tim_b]

Nanya is in a very interesting, unexpected, and profitable position today.

1. 1. Due to delays in the development of its in‑house DDR5 process, Nanya’s DDR5 entered high volume production only in 2024, rather than through licensing from Micron. Additional DDR5 capacity is scheduled to come online in late 2027. With the rapid growth of AI exhausting the capacity of the top three memory makers, Samsung, SK Hynix, and Micron, the market urgently needs more supply. Apart from those top three, Nanya is the only company outside mainland China capable of providing additional DDR5 capacity.
2. 
   
3. 2. As Samsung, SK Hynix, and Micron have reduced or halted DDR4 or DDR5 production to focus on the more profitable HBM memory, Nanya’s DDR5 and DDR4 capacity has suddenly become critical.
4. 
   
5. 3. Many products used in today’s automotive, EV, storage (HDD, DDS), networking/communication, and consumer electronics sectors still rely on DDR4 because it remains sufficient (often more than sufficient) and cost effective. Products such as SK Hynix’s SSDs, Western Digital’s hard drives, SanDisk’s SSDs, Cisco's enterprise networking switches, and Tesla’s EVs all need DDR4. Technically, they could migrate to DDR5, but doing so may not be cost justifiable and could prove a dead end under current market constraints.
6. 
   
7. 4. SK Hynix’s strategy differs markedly from Samsung’s. Rather than pursuing everything in‑house, SK Hynix has patiently built collaborations, alliances, and joint ventures with other companies. Its investments in Kioxia and Nanya, along with its early partnership with TSMC to integrate HBM memory into AI systems, are prime examples. Collaboration with TSMC is likely one of the key reasons SK Hynix has grown far faster than anyone expected three or four years ago.

At today’s close, SK Hynix’s market capitalization stands at US $874.9 billion. It is not far behind Samsung’s US $1.116 trillion.

But they don’t have much room for growth, right? Taiwan is already a saturated labor market, and they don’t seem to have any plans to expand elsewhere.

 

[Barnsley]

But they don’t have much room for growth, right? Taiwan is already a saturated labor market, and they don’t seem to have any plans to expand elsewhere.

If they been in the doldrums , likely they looking to squeeze out what they can from the crumbs that have fallen from the top table.

The negative energy on this site for anyone but the industry leaders is quite something.

 

[hist78]

But they don’t have much room for growth, right? Taiwan is already a saturated labor market, and they don’t seem to have any plans to expand elsewhere.

It seems that land availability for fab expansion is not a major issue for Nanya Technology or its parent company, Formosa Plastics Group at this moment. Its Fab 5A is under construction and is expected to begin DDR4 and DDR5 production in late 2027. At the Fab 5A, Nanya has already built the shells for two additional phases of expansion at the private industrial park that Formosa Plastics Group has owned since the late 1960s. Beyond that, Formosa Plastics Group owns a substantial amount of land that is either idle or can be redeveloped for future fab projects.

 

[tim_b]

It seems that land availability for fab expansion is not a major issue for Nanya Technology or its parent company, Formosa Plastics Group at this moment. Its Fab 5A is under construction and is expected to begin DDR4 and DDR5 production in late 2027. At the Fab 5A, Nanya has already built the shells for two additional phases of expansion at the private industrial park that Formosa Plastics Group has owned since the late 1960s. Beyond that, Formosa Plastics Group owns a substantial amount of land that is either idle or can be redeveloped for future fab projects.

The biggest challenge is finding qualified engineers and technicians. TSMC’s growing presence in Arizona should help preserve some of the local talent base, but the overall talent shortage remains a concern.