Intel's CEO warns that the global memory shortage will worsen, leading to project delays and price hikes, particularly impacting consumer electronics.
Pat Gelsinger, CEO of Intel, has issued a stark warning about the escalating global memory shortage, predicting it will lead to significant project delays and a price surge of 5 to 7 times. This announcement was made during the AI Infra Summit in Santa Clara, highlighting a critical issue for the tech industry.
“When I mentioned at the start of last year that memory could become a significant bottleneck, many seemed unaware of it. It has indeed materialized, and the situation is set to worsen,” he stated.
Gelsinger pointed out that manufacturing capacities are severely limited, forcing many companies, especially those in the budget and mid-range smartphone and laptop sectors, to delay projects due to difficulties in sourcing components.
While he did not specify which types of chips are most affected, the memory shortage is evident. According to TrendForce, DRAM manufacturers reported a staggering 59.5% revenue increase in the second quarter compared to the previous quarter, driven by rising contract prices and increased demand from AI servers.
However, TrendForce anticipates a slowdown in contract price growth for standard DRAM to between 13% and 18% in the third quarter, a stark contrast to the first quarter's increases of up to 98%.
With supplier inventories at historic lows, any additional production capacity is primarily directed towards the server market, exacerbating the shortage for consumer electronics. Smaller manufacturers are adapting by redesigning devices, reserving components ahead of time, or using memory from older models, as the primary issue they face is not just price but the actual availability of chips.
Previously, SK Hynix's CEO had warned that 2027 might be a particularly challenging year for the memory industry, with demand potentially outstripping production capabilities even beyond 2030.
AI Demand Is Driving Memory Needs
The rise of AI technology has significantly boosted the demand for various memory types, including High Bandwidth Memory (HBM) for accelerators and large volumes of DDR5 and high-capacity RDIMM for servers. As manufacturers shift their production focus to meet this demand, growth in memory supplies for personal computers and smartphones is likely to slow.
This shift is already impacting the consumer electronics market, with reports indicating that manufacturers are modifying designs and sourcing components in advance to cope with the supply constraints.
In addition to memory shortages, Gelsinger identified energy access and cooling systems as other significant barriers to scaling AI infrastructure. He noted the industry's transition from selling individual processors to offering integrated systems, combining GPUs, central processors, and networking equipment into comprehensive solutions.
“Another area is cooling. Beyond air cooling, there are liquid and microfluidic cooling systems, and we are investing in several of these directions,” Gelsinger added.
He reaffirmed Intel's ongoing collaboration with Nvidia, emphasizing the need for a holistic approach to infrastructure that encompasses energy use, inter-component connections, and heat management. Recently, companies like Huawei and MetaX have raised prices for AI accelerators, attributing these increases to the global shortage of HBM.
