"Previously it was energy, now it's storage!" -- OpenAI COO on the "AI bottleneck"

```

The shortage of memory chips is replacing electricity supply as the primary limiting factor for the expansion of artificial intelligence infrastructure.

OpenAI Chief Operating Officer Brad Lightcap stated at the Hill and Valley Forum in Washington on Tuesday that the shortage of memory chips and the tightening of U.S. energy supply are currently the two main potential bottlenecks facing AI infrastructure expansion. "The current bottleneck is memory, previously it was power," he said bluntly on stage.

This view aligns closely with the judgment previously made by SK Hynix chairman Chey Tae-won, who anticipates that the global shortage of memory chips will last until around 2030, with the industry's wafer supply gap exceeding 20%.

For the market, this means that the core contradiction of the AI computing power arms race is undergoing a structural shift: from data center location and grid capacity to the supply chain security and capacity bottleneck of memory chips. The strategic position of AI accelerator suppliers such as Nvidia and manufacturers of HBM (high bandwidth memory) is thus further highlighted.

Memory Replaces Power, Becomes the New Bottleneck for AI Expansion

Lightcap's comments mark a significant change in the AI industry's understanding of infrastructure constraints. In the past two years, insufficient power supply for data centers has been the industry's most-discussed topic, but now the shortage of memory chips has become a more urgent practical obstacle.

The root cause of this shortage lies in explosive demand growth. AI companies such as OpenAI continue to purchase Nvidia AI accelerators on a large scale, and each accelerator is equipped with a large number of memory chips, thereby consuming a significant share of the world's memory production capacity. Lightcap pointed out that OpenAI is actively broadening its supplier base and expanding the geographic layout of its data centers to ensure that its infrastructure expansion plans are not constrained by a single supply chain.

According to Bloomberg, OpenAI has previously committed $1.4 trillion over the next few years for data center construction and chip procurement, to support the development of more advanced AI systems and the broader popularization of the technology.

Shortage May Continue Until 2030, Traditional DRAM Also Under Pressure

Previously, SK Hynix chairman Chey Tae-won stated at Nvidia's GTC Conference that the global memory chip shortage is expected to persist for another four to five years, as it takes at least as long to expand wafer production capacity, and major memory manufacturers will find it difficult to fully meet market demand before 2030.

Chey Tae-won also warned that the industry's excessive focus on high bandwidth memory (HBM) could lead to a shortage of traditional DRAM supply, thereby affecting the smartphone and PC markets. In recent years, SK Hynix, Samsung, and Micron have shifted a significant portion of their capacity to HBM for AI accelerators, leading to a decline in traditional DRAM output and a sharp rise in consumer electronics prices.

SK Hynix currently holds about 57% of the global HBM market and about 32% of the overall DRAM market. The company is currently constructing a $13 billion HBM packaging and testing facility in Cheongju, South Korea, scheduled to break ground next month, with the goal of completion by the end of 2027.

Unresolved Energy Issues, High Hopes for Nuclear Power and Government Support

Although memory has become the most pressing bottleneck at present, pressure on energy supply has not subsided. Lightcap stated that OpenAI is considering introducing diversified power sources, including nuclear energy, to meet its continuously rising energy demand, and revealed that the company is in talks with nuclear fusion startup Helion Energy for cooperation.

It is worth noting that OpenAI CEO Sam Altman was previously a supporter of Helion and announced on Monday that he would resign from Helion’s board of directors and recuse himself from negotiations between the two companies.

Lightcap emphasized that government investment in energy supply is "crucial" to the success of the AI industry, and gave high praise to the Trump administration's efforts to promote AI infrastructure construction and accelerate government adoption of AI technology.

On the government business front, Lightcap disclosed that OpenAI currently serves more than 1 million employees at local, state, and federal government agencies in the U.S., and has made accelerating product delivery to federal agencies a strategic priority. At the end of last month, OpenAI reached an agreement with the U.S. Department of Defense to deploy its AI models on classified Pentagon networks. Previously, the Pentagon had announced it would terminate its cooperation with competitor Anthropic PBC. Lightcap described serving government agencies as a "crucial" direction for the company.

Risk Warning and DisclaimerThe market has risks, investments require caution. This article does not constitute personal investment advice, nor does it take into account the special investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular situation. Investment based on this is at your own risk. ```