OpenAI CEO Acknowledges Short-Term AI Bubble Risk in Closed-Door Samsung Meeting
TREE NEWS reports: Sam Altman, CEO of OpenAI, told around 50 senior executives from Samsung Group that demand for AI memory and data center infrastructure may be overstated in the near term due to double-counting, while simultaneously proposing a joint development partnership for next-generation AI devices targeting the post-smartphone era. The remarks, made during a closed-door meeting in South Korea late last month, offer a rare candid assessment from a top AI industry leader about the potential for overheating in the AI infrastructure buildout.
Short-Term Bubble Risk and Demand Double-Counting
Responding to questions about whether AI and memory demand could saturate or slow over the next 5 to 10 years, Altman stated that there would be “fluctuations in the short term.” He specifically criticized the bubble debate surrounding emerging cloud service providers that rent out GPU-based AI compute infrastructure, noting that “there is double-counting of demand” and emphasizing that “only OpenAI and Anthropic can truly drive large-scale demand growth, and supply shocks and periods of instability may arrive.”
This admission carries direct implications for memory chip suppliers such as Samsung Electronics and SK Hynix, both of which announced participation in OpenAI’s Stargate project last October to advance advanced memory supply cooperation. The warning suggests that while long-term demand remains robust, the near-term order flow may not be as linear as some investors expect.
Long-Term Optimism Tempered by Supply Chain Bottlenecks
Despite short-term caution, Altman remains highly optimistic about AI’s long-term demand outlook. He cited explosive growth in token usage—from about 100,000 tokens per month for the top user in early 2020 to 100 billion tokens per month for the highest-usage user by early this year—and projected further acceleration as AI is applied to custom software, robotics, spacecraft, and scientific research.
When asked about the biggest bottlenecks to AI expansion, Altman listed supply chain, financing, AI safety and social acceptance, and corporate transformation, identifying “chip supply chain” and “data center infrastructure” as the most critical constraints in the coming years. He called for more investment in the supply chain, a signal that hardware and infrastructure providers remain essential to the AI buildout despite near-term demand volatility.
Low-Margin, High-Volume Business Model
On profitability concerns, Altman acknowledged that sustained large-scale investment creates pressure but outlined OpenAI’s strategy: building vertically integrated infrastructure at lower cost than any competitor, selling tokens at low margins, and winning through volume. He noted that investments in custom chips and compute efficiency are paying off, and long-term contracts enable higher operational efficiency to drive down cost per token. “The market is large enough that the rules of ordinary commodity industries don’t apply here,” he said. “Just meeting global token demand is enough to build an outstanding industry.”
Seeking Samsung Partnership for Post-Smartphone Devices
Altman proactively proposed close collaboration with Samsung to co-develop a joint device, describing a next-generation AI terminal that is not app-centric but instead has AI actively managing users’ physical and virtual lives with minimal intrusion. OpenAI is reportedly targeting a device for next year, most likely a screenless smart speaker smaller than a palm. Altman expressed reservations about AI glasses, stating discomfort with technology intervening in human relationships.
He also urged Samsung executives to go “all-in on AI,” arguing that using AI merely as a productivity tool yields 2-3x gains, but letting AI lead entire business processes can yield 100x. He warned that in 3-5 years, companies not built on AI will struggle to remain competitive.
Market Implications
For equity investors, Altman’s comments are a double-edged sword. On one hand, the acknowledgment of demand double-counting could temper enthusiasm for memory chip stocks and AI infrastructure plays that have rallied on expectations of uninterrupted demand growth. Samsung Electronics and SK Hynix may see near-term volatility as the market reassesses order sustainability. On the other hand, the long-term optimism and call for more supply chain investment support the secular growth narrative for AI hardware.
For the broader tech sector, the proposed Samsung-OpenAI partnership signals a potential shift in the competitive landscape of consumer devices, challenging Apple’s ecosystem dominance. This could weigh on Apple’s long-term positioning while benefiting Samsung’s device and component businesses.
In crypto markets, AI-related tokens and decentralized compute networks may face headwinds if the narrative of insatiable AI demand cools. However, the emphasis on token usage growth and cost efficiency could support projects focused on AI inference and data markets.
Bond markets are unlikely to see direct impact, but any slowdown in AI capex could affect corporate bond issuance from tech giants. Commodities, particularly those tied to semiconductor supply chains, may experience near-term pressure.
Key Takeaways for Investors
- Altman’s warning about demand double-counting suggests near-term volatility in AI infrastructure and memory chip stocks; investors should scrutinize order books for signs of overordering.
- Long-term AI demand remains intact, with chip supply chain and data centers as key bottlenecks; companies enabling these areas may offer resilient growth.
- The potential Samsung-OpenAI device partnership could disrupt the consumer electronics landscape, with implications for Apple and its suppliers.
- AI-related crypto tokens may face sentiment headwinds, but projects tied to token efficiency and decentralized compute could benefit from the long-term trend.
- Watch for further commentary from AI leaders on demand sustainability, as it could drive sector rotation.
