Intel is struggling... we all know it, Samsung is struggling... we all know it... but now rumors are swirling that Intel Foundry and Samsung Foundry could combine forces to compete against the semiconductor leader: TSMC.
In a new video from leaker Moore's Law is Dead, his sources have said that Intel and Samsung could combine resources in their foundries, to compete with TSMC. As I was writing this story, I noticed that South Korea's Daily Economic News that Intel and Samsung are in discussions to form a "wafer foundry alliance" to fight TSMC.
UDN reports that the industry is concerned that Intel could shift its processor outsourcing orders as a "conditioin of exchange" with Samsung, which will affect TSMC (but not too much, they're too busy fabbing chips for NVIDIA, AMD, Qualcomm, MediaTek, Apple, and more).
However, with Intel's continued struggles in the processor market (including the new Core Ultra 200 series "Arrow Lake" processors) are one thing, but the US chip giant has also been spending billions of dollars acquiring the best High-NA EUV lithography machines in the world from ASML, so once Intel has its High-NA EUV lithography machines setup in the months and years ahead, we'll see things massively improved.
- Read more: Intel trying to poach South Korean chip design firms away from Samsung
- Read more: Intel CEO wants to make chips for AMD, Google, NVIDIA, and Qualcomm
- Read more: Samsung Foundry wins first 2nm AI chip order, stole TSMC client
- Read more: Samsung Foundry will tease its 1nm process node in July, coming in 2026
- Read more: Samsung Foundry renames SF3 node to SF2
- Read more: TSMC wins Google, Qualcomm 3nm orders away from Samsung Foundry
- Read more: Meta switches from TSMC to Samsung Foundry for AI chips, 'uncertainty and volatility' at TSMC
A partnership with Samsung makes sense, as NVIDIA has teamed heavily with South Korean memory rival SK hynix, and Taiwanese semiconductor manufacturing leader TSMC on its market-leading Hopper H100 and H200 AI GPUs, and now its Blackwell B200 and GB200 chips powered by new HBM3E memory from SK hynix. NVIDIA's future-gen Rubin R100 AI GPU will use HBM4 and will see the light of day in Q4 2025.
NVIDIA, TSMC, and SK hynix also recently formed the "triangular alliance" for next-gen AI GPUs and HBM4 memory... so Samsung and Intel need to do something, right?