🔍 Read the full analysis: Why Did Altman Call PC Memory “Quite Terrible” Before Meeting Samsung Executives? on ThorstenMeyerAI.com
Get movie nights delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
OpenAI CEO Sam Altman met about 50 Samsung executives in Yongin on Oct. 31, two days after saying on a podcast that PC memory is “quite terrible” from a global allocation perspective. Samsung described the meeting as an AI leadership program; reports linked it to memory and AI infrastructure, but no new supply agreement was announced.
OpenAI chief executive Sam Altman met about 50 Samsung executives in Yongin, South Korea, on Oct. 31, two days after describing the amount of memory sitting in personal computers as “quite terrible” from a global allocation perspective. Samsung confirmed the meeting but presented it as an AI leadership program; neither company announced a new memory supply deal.
Samsung said the attendees were presidents and senior executives from its electronics, finance and biotech affiliates, and that they gathered at the company’s Human Resources Development Center in Yongin. The stated theme was “The Next 10 Years Shaped by AI: The Enterprise’s Choice.” Samsung’s account said participants discussed AI’s direction, its effects on decision-making and management, and the markets it could open. Yonhap, citing Samsung, reported that Altman told the group that businesses gain an advantage not merely by adopting AI but by using it to move faster, make better products and get ahead of competitors.
The meeting’s memory-chip significance comes from reporting, not an announced agreement. Bloomberg reported that Altman’s visit aimed to deepen cooperation on memory chips and AI infrastructure as shortages intensify. The Seoul Economic Daily reported that Richard Ho, OpenAI’s vice president leading its in-house chip work, traveled with Altman, and that OpenAI’s accelerator may use Samsung’s next-generation HBM4 memory. Those details have not been confirmed by OpenAI or Samsung. No contract terms, supply volumes or prices were disclosed.
Altman made the PC-memory remark on an episode of Stripe co-founder John Collison’s podcast, “Cheeky Pint,” published Oct. 29. In the discussion, Collison asked why high memory prices had not prompted more supply. Altman said the “invisible hand is failing right now” and argued that memory in personal computers is idle much of the time, while shared data centers could use scarce memory more intensively. His estimate that PC memory is being used 5% of the time was his characterization, not a cited measurement. The comment concerns how memory is allocated, not a claim that personal computers themselves perform poorly.
Infographic 01 · Confirmed, reported, not announced
What we actually know about Altman’s Samsung meeting
Samsung confirmed the meeting and framed it as AI leadership training. The memory angle comes from reporting. No deal was announced.
- Meeting with about 50 presidents and executivesConfirmed by Samsung
Electronics, finance and biotech affiliates; Samsung Human Resources Development Center. Theme: “The Next 10 Years Shaped by AI: The Enterprise’s Choice”.
- Altman: the edge comes from how you use AI, not from adopting itConfirmed by Samsung
As relayed by Samsung via Yonhap; quoted similarly by the Seoul Economic Daily.
- Aim: deeper cooperation on memory chips and AI infrastructureReported
Bloomberg, “as global shortages intensify”.
- OpenAI’s chip-hardware chief traveled along; HBM4 linkOne outlet
Seoul Economic Daily, citing industry sources. Not confirmed by either company.
- SK chairman Chey Tae-won met Altman the same dayOne outlet
Yonhap, citing industry sources. Content not disclosed.
- Volumes, prices, a new contractNot announced
Nothing announced. No statement from OpenAI on the visit.
How Samsung and OpenAI got here
- 1 Oct 2025Letter of intent: memory for Stargate, data centers, floating data centersSamsung and OpenAI announcement
- Jun 2026Samsung rolls out ChatGPT Enterprise and Codex to employeesOpenAI, as reported by the Korea Herald
- Jul 2026Samsung chairman Lee Jae-yong visits OpenAI in San FranciscoKorea Herald
- 29 Oct 2026Samsung’s record quarter; Altman’s “quite terrible” DRAM remarkSamsung results · Cheeky Pint podcast
- 31 Oct 2026Altman meets about 50 Samsung executives in YonginSamsung via Yonhap · Bloomberg
“Confirmed by Samsung” means Samsung’s own account, as relayed by Yonhap and the Korea Herald. A letter of intent is a statement of plans, not a purchase order.
Sources: Samsung via Yonhap, 31 Oct 2026, Korea Herald, 31 Oct 2026, Bloomberg via The Straits Times, 31 Oct 2026, Seoul Economic Daily, 1 Nov 2026, Samsung and OpenAI letter of intent, 1 Oct 2025.
Powered by Thorsten Meyer AI
AI Data Centers Compete for Memory
The dispute behind the remark is about who gets limited memory capacity: cloud providers building AI data centers, or manufacturers supplying PCs and other devices. High-bandwidth memory is a key component in advanced AI systems, while conventional DRAM is used across computers and consumer electronics. If production increasingly favors data-center demand, device makers could face higher costs or tighter supply. That could show up for buyers as price increases, lower memory configurations or slower product upgrades, though those outcomes are risks rather than confirmed effects of this meeting.
There is also a business interest behind Altman’s argument. OpenAI sells access to AI through cloud services and is developing its own hardware, so more memory available to data centers could support its operations. That does not establish that Altman’s comments were intended to secure a particular Samsung allocation. The meeting’s agenda and any private discussions have not been made public.
Samsung’s financial results help explain why the issue matters to the company. The source report says Samsung’s chip division earned 106.5 trillion won of 107.4 trillion won in company operating profit in the third quarter of 2026. It also reports that the phone and network business recorded a 0.6 trillion won operating loss, partly amid cost pressures. Those figures put memory at the center of Samsung’s business, but they do not show that the Yongin meeting produced a commercial commitment.
Popular right now
Dyson Airwrap i.d. Multi-Styler for Straight and Wavy Hair
KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, Almond Cream
LG 85-Inch QNED73B Mini LED 4K Smart TV (2026)
As an affiliate, we earn on qualifying purchases.
OpenAI and Samsung’s Existing Ties
The Yongin meeting followed an existing relationship. In October 2025, OpenAI and four Samsung companies signed a letter of intent. Samsung said at the time that OpenAI’s memory demand was projected to reach up to 900,000 DRAM wafers per month. That was a projection in a letter of intent, not a confirmed monthly purchase or a supply contract. In June 2026, OpenAI said Samsung Electronics would introduce ChatGPT Enterprise and Codex to employees in Korea and its global device division. The Korea Herald reported that Samsung executive chairman Lee Jae-yong visited OpenAI in San Francisco in July.
Samsung is one of the world’s major memory producers, alongside South Korea’s SK hynix. Yonhap, citing industry sources, reported that SK Group chairman Chey Tae-won met Altman in Seoul on Oct. 31 with SK hynix executives; the subjects were not disclosed. Bloomberg has also reported visits to South Korea by Nvidia chief Jensen Huang and AMD chief Lisa Su to strengthen supply ties. These developments reflect competition for memory and other components as AI infrastructure expands, but they do not confirm the terms of any company’s arrangements.
Prices were still expected to rise, according to the industry forecast cited in the source report. TrendForce projected conventional DRAM contract prices would increase 10% to 15% in the fourth quarter of 2026 and warned that PC memory supply could decline in 2027 as more capacity shifts toward servers. These are forecasts, not finalized price changes or guaranteed shortages.
““Quite terrible””
— Sam Altman, on John Collison’s “Cheeky Pint” podcast
What the Meeting Did Not Confirm
Samsung has not released a detailed attendee list or a transcript of the meeting. Its public description focused on AI leadership and business strategy, while Bloomberg reported a memory and infrastructure purpose. The two accounts are not necessarily incompatible, but the companies have not publicly set out the meeting’s full agenda.
It remains unknown whether OpenAI and Samsung discussed a specific memory allocation, whether any commercial terms were agreed privately, or whether OpenAI’s future accelerator will use Samsung HBM4. The Seoul Economic Daily’s report about Ho’s attendance and prospective HBM4 use, as well as Bloomberg’s account of the visit’s purpose, should be treated as reporting rather than company-confirmed facts. Altman’s 5% estimate for PC memory use also has no measurement method or source stated in the material.
The cited price and supply outlook is likewise a forecast. It does not establish how much memory will be available to PC makers in 2027, what consumers will pay, or whether manufacturers will reduce specifications. Those effects depend on production decisions, demand and future agreements that have not been disclosed.
Watch for Supply and Chip Details
The next concrete developments would be a company announcement specifying a memory or infrastructure agreement, or further confirmation about OpenAI’s accelerator and its memory supplier. Until then, the Oct. 31 meeting is confirmed, but a new supply commitment is not. Samsung and OpenAI have not said when they might disclose additional details.
For buyers and device makers, the near-term indicators are DRAM contract-price updates and memory availability. TrendForce’s fourth-quarter forecast and its warning about possible 2027 pressure offer a market outlook, not a timetable for retail price changes. Any effects on PC configurations or consumer prices will need to be confirmed by manufacturers and subsequent market data.
Key Questions
Why did Altman call PC memory use “quite terrible”?
On the “Cheeky Pint” podcast, Altman argued that personal-computer memory often sits idle while shared data centers could put scarce memory to more intensive use. His estimate that it is used 5% of the time was his own characterization, not a cited measurement.
Did Samsung and OpenAI announce a new memory deal?
No. Samsung confirmed the Oct. 31 meeting, but neither company announced a new contract, supply volume or price.
What was the stated purpose of the Yongin meeting?
Samsung described it as a leadership program about AI’s effect on business and future markets. Bloomberg separately reported that the visit aimed to deepen cooperation on memory chips and AI infrastructure.
Will the meeting mean less memory for PCs?
That is not established. TrendForce forecast possible pressure on PC memory supply in 2027 as capacity shifts toward servers, but the meeting itself does not confirm a supply change or a specific effect on PC makers.
Has Samsung confirmed that OpenAI will use its HBM4?
No. The Seoul Economic Daily reported that OpenAI’s accelerator is expected to use Samsung’s next-generation HBM4, but neither company has confirmed that report.
Source: ThorstenMeyerAI.com
