Sam Altman flies to Korea to meet Samsung as AI memory supply tightens
OpenAI's CEO met about 50 Samsung executives as chipmakers struggle to supply enough memory for AI data centers, PCs and phones all at once.
The short version
- Altman met about 50 senior Samsung executives in South Korea, per Bloomberg.
- Samsung and SK Hynix together hold about 83% of HBM revenue.
- Samsung said talks centered on a letter of intent signed in October 2025.
- AMD's Lisa Su also visited Korea this month to seek more HBM supply.
- TrendForce expects DRAM supply to fall 1% to 2% short of demand this year.
Why it matters
AI services depend on physical parts, and memory is one of the scarcest. When the biggest AI companies compete with makers of PCs and phones for the same chips, it can affect how much computing capacity is available. That is our inference, not a claim in the report, but it is a useful lens for any plan that relies on one AI vendor.
A CEO goes to the supplier
As technews.tw reported, citing Bloomberg, OpenAI CEO Sam Altman traveled to South Korea to meet roughly 50 senior Samsung executives. His goal was to expand capacity for AI and memory.
Samsung's Korean arm confirmed the meeting. It said the talks were about strengthening the existing relationship, especially a letter of intent (a non-binding statement of shared plans) signed in October 2025 covering global AI infrastructure cooperation.
Why memory is the choke point
HBM, or high-bandwidth memory, is a type of fast memory placed close to AI processors so they can be fed data quickly. Samsung and SK Hynix together take about 83% of HBM revenue, so a few companies decide how much is available.
Data centers are also competing with PCs, smartphones and other electronics for memory. Research firm TrendForce expects DRAM, the common type of computer memory, to be 1% to 2% short of demand this year, with the imbalance worsening in 2027.
Others are doing the same
Altman is not alone. AMD CEO Lisa Su visited Korea this month to discuss winning more HBM supply from Samsung. She said planning with partners 3 to 5 years ahead is important.
Altman had earlier said memory in personal computers often sits underused, and that scarce DRAM should go to data centers, where sharing across AI workloads raises usage. Critics noted that OpenAI's own services, such as ChatGPT, depend heavily on centralized computing, so the argument serves its interests.
What we do not know
The report does not say whether any supply deal was reached, or what volumes were discussed. Samsung described the talks in general terms: AI technology direction, possible uses and future market opportunities.
One small detail: Altman reportedly gave Samsung's AI center head a Codex Micro Keyboard, a device designed for use with OpenAI's coding agent tools.
What to watch
- Will Samsung and OpenAI announce a concrete supply agreement after the visit?
- Will the DRAM shortfall TrendForce expects for 2027 grow as projected?
Try it
If your business relies on one AI vendor, ask them how they would handle a capacity shortage and whether they have a backup plan for service limits or pricing.
Sources and links
- Reportingtechnews.tw: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by technews.tw · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.