AMD could be preparing a major foundry backup plan, and honestly, this one matters more than it sounds.
According to a report highlighted by Wccftech, AMD is in advanced discussions with Samsung to use the Korean giant’s 2nm manufacturing technology for future AI-focused CPUs and accelerators. The move comes at a time when demand for AI datacentre hardware is going absolutely gila, and chip supply is becoming one of the biggest bottlenecks in the tech industry.
Right now, AMD depends heavily on TSMC for its most advanced chips. That relationship has helped AMD push strong CPU and GPU products over the years, especially in high-performance computing and datacentres. But TSMC’s leading-edge production is extremely crowded, with reports saying its 2nm wafer supply is already booked out until 2028.
So yeah, if AMD wants to keep growing in AI, relying on one foundry alone is risky.
Why Samsung is suddenly in the picture
The report says Samsung’s foundry division has been advancing talks with AMD around 2nm chip orders. AMD CEO Dr Lisa Su reportedly visited Samsung’s Pyeongtaek foundry plant in Korea back in March, and that visit may have helped push the discussions forward.
For Samsung, landing more AMD work would be a big confidence boost. Samsung already has 2nm Gate-All-Around chip technology, but the market still sees TSMC as the safer and stronger option, especially when it comes to yield and advanced manufacturing consistency. If AMD gives Samsung serious orders, that would signal that Samsung is not just a backup name on paper — it can actually compete for high-end AI silicon.
That said, this does not mean AMD is ditching TSMC. Far from it. TSMC has already confirmed AMD’s Venice CCD as the first product on its N2 nanosheet process. The more realistic read is that AMD is trying to spread risk and secure more capacity wherever it can.
What chips are we talking about?
The report mentions AMD’s next-gen 2nm CPUs, including Venice and Verano. Venice is expected to be a compute-optimised Zen 6 product, while Verano is described as a Venice variant built for agentic AI workloads, including inferencing.
In plain gamer-bro language: these are not chips you’ll be buying for your next RM4,000 gaming PC build. These are datacentre-class parts aimed at the AI servers powering cloud platforms, enterprise tools, and future AI services.
But Malaysian and SEA users should still care. When AI hardware supply gets tight, it affects everything upstream — cloud pricing, availability of AI tools, datacentre expansion, enterprise services, and eventually even consumer products. More supply options could help reduce pressure across the whole ecosystem.
The bigger chip war is getting spicy
AMD looking beyond TSMC also shows how competitive advanced chipmaking has become. Samsung is trying to win more leading-edge customers, Tesla has also been linked to Samsung for AI5 and AI6 chip production, and Intel Foundry is pushing its own technologies like 14A, 18A-P, and EMIB to become another serious alternative.
For Malaysia and SEA, this matters because our region increasingly depends on global semiconductor flows. Malaysia already plays a major role in chip packaging, testing, and electronics manufacturing. When big players like AMD, Samsung, TSMC, and Intel fight for next-gen AI chip capacity, the ripple effects can influence investment, supply chains, device launches, and even local tech job opportunities.
The short version: AI is eating chip capacity faster than suppliers can comfortably deliver. AMD needs more room. Samsung wants more credibility. TSMC is fully loaded. And the whole semiconductor race is getting more aggressive.
If Samsung can prove its 2nm process is reliable enough for AMD’s future AI chips, this could be one of those behind-the-scenes deals that quietly shapes the next few years of computing.
Source: Wccftech Gaming