The AI Industry Faces Challenges Due to Scarcity of Powerful Chips
The ever-increasing demand for AI technology has exposed a significant challenge faced by the industry – a scarcity of high-performance chips essential for the development and implementation of AI models.
This ongoing chip crunch has impacted businesses of all sizes, including leading AI platforms, and industry analysts predict that it may persist for at least a year or longer.
Microsoft’s recent annual report highlighted a potential extended shortage in AI chips, specifically graphics processing units (GPUs), a crucial hardware type used to execute numerous calculations in AI algorithm training and deployment.
In the report, Microsoft emphasized its efforts to expand data center locations and increase server capacity to meet the growing demand for AI services. However, the availability of GPUs and other critical components remains a bottleneck for the industry, affecting companies building AI tools and products and indirectly impacting businesses and end-users seeking to leverage AI technology.
OpenAI CEO Sam Altman testified before the US Senate, revealing that the company’s chatbot tool was struggling to cope with the influx of user requests due to a shortage of GPUs.
“We’re so short on GPUs, the less people that use the tool, the better,” Altman stated.
An OpenAI spokesperson reassured CNN that the company is committed to ensuring sufficient capacity for users. This situation bears similarities to the scarcity of popular consumer electronics during the pandemic era, where gamers faced inflated prices for game consoles and PC graphics cards due to manufacturing delays, labor shortages, global shipping disruptions, and competing demands from cryptocurrency miners.
Diverging from past shortages, the current chip scarcity primarily results from the skyrocketing demand for ultra high-end GPUs designed for advanced AI work, such as AI model training and implementation.
Although production of these GPUs is at capacity, the overwhelming demand has outpaced the available supply sources.
Raj Joshi, a senior vice president at Moody’s Investors Service, who monitors the chips industry, describes a “huge sucking sound” caused by the insatiable demand for AI technology. The industry was ill-prepared for such an unprecedented surge in demand.
One company poised to benefit immensely from this AI boom is Nvidia, the trillion-dollar chipmaker, which reportedly controls 84% of the market for discrete GPUs. Industry estimates predict Nvidia to experience unparalleled revenue growth, with its data center business potentially surpassing Intel and AMD’s combined revenue.
In response to the rising demand for AI chips, Nvidia has procured a substantially higher supply for the second half of the year. However, the company declined to comment further during its pre-earnings quiet period.
AMD, another prominent player, expects to unveil its answer to Nvidia’s AI GPUs by the end of the year. AMD CEO Lisa Su expressed strong customer interest in their AI solutions and emphasized significant progress in their endeavors.
Complicating the chip shortage is the struggle of GPU-makers to secure a key input from their suppliers – the silicon interposer. This crucial technology enables the integration of standalone computing chips with high-bandwidth memory chips, a critical step in the completion of GPUs.
To address the global chip shortage, the Biden administration prioritized increasing US chip manufacturing capacity, as evident from the passage of the CHIPS Act, which allocated billions in funding for the domestic chip industry and research and development. However, the investments are broad and not specifically aimed at boosting GPU production.
While the chip shortage is expected to ease with increased manufacturing and expanded offerings from Nvidia’s competitors, some industry experts estimate it could still take two to three years for the situation to stabilize.
During this period of scarcity, companies may need to explore creative solutions. Efficient usage of available resources becomes crucial as businesses grapple with the limitations posed by the chip shortage.
AI startup d-Matrix’s founder and CEO, Sid Sheth, points out that necessity often breeds innovation. Companies may adopt smaller AI models that require less computational power or explore novel computation methods that rely less on traditional CPUs and GPUs.
Overall, this chip shortage, while challenging, could ultimately spur innovation and efficiency in the AI industry, leading to novel approaches and advancements in AI technologies.
