Test report DSG-4647 · Rev D · tested October 10, 2026
Foundries & ManufacturingDevice under test
Samsung Leverages HBM4 Supply Position in AMD Foundry Talks
Samsung used its HBM4 memory supply position to push AMD into foundry negotiations, Tech Times reports, linking scarce AI memory access to contract chipmaking talks.
- Read
- 3 min
- Words
- 647
- Node
- 28nm
- Operator
- Amara Osei
Spec summary
- Samsung used its HBM4 memory supply position to bring AMD into foundry negotiations, Tech Times reports.
- HBM4 is a critical component for AI accelerators, with few qualified suppliers.
- AMD currently relies on TSMC to fabricate its flagship processors.
- No contract values, volumes, or timelines for the foundry talks have been confirmed.
- Neither Samsung nor AMD has publicly confirmed the reported arrangement.

Samsung used its position in HBM4 high-bandwidth memory supply to push AMD into foundry negotiations, according to a report published by Tech Times. The claim, if accurate, describes a supplier converting control of a critical AI-accelerator component into bargaining power in a separate business line: contract chip manufacturing.
The story matters because of where HBM4 sits in the supply chain. High-bandwidth memory is a gating component for AI accelerators and data-center GPUs, and any manufacturer that secures preferential access to it gains leverage over companies that design those accelerators. AMD designs high-performance GPUs and, like most fabless chipmakers, depends on external manufacturers to fabricate its silicon.
What does the reported arrangement involve?
According to the Tech Times report, Samsung's strength in HBM4 supply became a factor that brought AMD to the table for foundry discussions. The report frames the memory business not as a standalone product line but as an instrument of cross-business negotiation.
The three elements of the reported dynamic:
- Samsung controls meaningful capacity in HBM4, the memory generation that current and upcoming AI accelerators require.
- AMD needs HBM4 supply for its GPU roadmap and needs advanced-node manufacturing capacity for its chips.
- Samsung operates a foundry business that competes for customers against the market leader, TSMC, which currently fabricates AMD's flagship processors.
Read together, these elements describe a classic bundling play: access to one scarce input becomes conditional on, or at least connected to, conversations about another service.
Why is HBM4 supply leverage significant?
HBM has become one of the tightest components in the AI hardware stack. Accelerator vendors cannot ship data-center GPUs without stacked high-bandwidth memory, and the pool of qualified suppliers is small. SK Hynix currently leads the segment, with Samsung and Micron competing for qualification and share.
In that environment, a credible HBM4 supply commitment from Samsung carries direct commercial value to AMD. The Tech Times report suggests Samsung recognized that value and used it as an opening into foundry discussions — a segment where Samsung has trailed TSMC in advanced-node share and has sought major external customers to fill capacity.
For AMD, the calculus cuts both ways. Diversifying foundry supply away from a single dominant manufacturer would reduce concentration risk, but shifting flagship silicon to a foundry with a weaker record on leading-edge yields carries execution risk. Memory supply terms could tip that balance.
What does this mean for the foundry market?
The reported tactic highlights how memory scarcity is reshaping negotiating dynamics across the semiconductor industry. When one company holds positions in both a bottleneck component and contract manufacturing, it can link the two in ways that pure-play foundries cannot.
Potential consequences industry watchers will track:
- Foundry pricing and allocation negotiations increasingly tied to memory supply agreements.
- Accelerator vendors weighing single-supplier convenience against multi-supplier leverage.
- Regulators examining whether cross-product bundling by large memory suppliers distorts competition.
Samsung has publicly pursued major foundry customers for years, and AMD — which buys wafers at leading-edge nodes — sits high on any prospective-customer list. The Tech Times report does not state that a foundry deal has been signed, and neither company has confirmed the terms under discussion.
What remains unconfirmed?
The report rests on the claim that HBM4 supply position was used as leverage, not merely offered alongside standard foundry pitches. No contract values, volume commitments, process nodes, or timelines appear in the reporting, and neither Samsung nor AMD has issued a public statement confirming the characterization.
Absent official comment, the story stands as a reported account of supplier behavior in a market where memory allocation has become a strategic currency. If the account is accurate, it signals that HBM4 access now functions as more than a component sale — it operates as a negotiating instrument that can open doors in adjacent businesses, including the fiercely contested foundry market.
via Google News: HBM memory (Source)
More from Amara Osei
Same lot · LOT-C1C6
- DSG-102920nmAMD CEO Visits Samsung to Negotiate HBM Supply, Foundry Deal on the Table
- DSG-59533nmAnalysts: HBM Memory Cut Samsung Losses by 41% in a Year
- DSG-87827nmSamsung Passes HBM4E Qualification with Nvidia and Major Hyperscalers
- DSG-502510nmSamsung Exec: HBM to Reach Nearly 30% of Global DRAM Capacity
- DSG-84123nmHBM Prices Set to Double as Memory Shortage Fuels Samsung Profits