Market analysts see a gloomy outlook for the semiconductor industry in the short term due to economic headwinds and weak electronics demand, but the custom chip market, especially for the artificial intelligence and machine learning (AI&ML) segment, is showing great prospects.
While Gartner expects the overall sector to decline 11.2% this year, the AI chip market is forecast to grow significantly, with a compound annual growth rate (CAGR) of 29.72% over the decade to reach a market size of over US$227 billion by 2032, according to a Precedence Research study.
This is driven by the rapid development of the AI&ML segment amid the recent popularity of large language models such as ChatGPT, which was launched by San Francisco-based AI laboratory OpenAI in November 2022. Almost overnight, AI became the talk of the town, its enormous potential stoking investor interest as well as fears above its negative impacts on society.
With OpenAI finding solid support from Microsoft, other tech giants such as Alphabet, Amazon and Meta are also investing heavily in large language models to stay in the game.
Venture capital firms are also shifting their focus to AI&ML start-ups. In the first quarter of 2023, the segment recorded the highest number of deals amounting to 15, totalling US$844.2 million. This is second only in value to biotechnology, which attracted US$1.2 billion during the period, according to the Emerging Tech Indicator, which tracks the investment activity of the top 15 venture capital firms and is produced by financial data provider PitchBook.
Investment activity was dominated by the Web3 and DeFi sectors in the first three quarters of last year, but they were overtaken by AI&ML in Q4 2022. The trend is expected to continue at least for the rest of 2023, PitchBook says.
Meanwhile, the global market for graphics processing units (GPUs), a specialized type of computer processor that is designed to handle complex calculations, is experiencing accelerated growth, particularly in the field of deep learning and other AI applications.
Santa Clara, California-based Nvidia, a leading provider of hardware and software solutions for AI&ML, accounts for 60-70% of the global GPU market, according to industry data provider TrendForce. The company’s GPUs are well-known for supporting ChatGPT and other large language models.
US curbs prod China
In Asia-Pacific, the GPU market is seen expanding at a CAGR of 33.69% from 2021 to 2028, according to market research reports firm Inkwood Research.
China, the largest market in the region, is cultivating a bunch of GPU start-ups such as MetaX, Birentech, Enflame, and Horizon, as it seeks to be a global leader in AI and other cutting-edge technologies.
Nvidia founder and chief executive officer Jensen Huang says China appears intent on developing its chip sector, especially in the wake of US export restrictions. "Whatever the regulations are, of course, we will absolutely comply,” Huang tells a recent industry expo in Taipei. “But I think China will use the opportunity to foster their local entrepreneurs, and that's why there's so many GPU start-ups in China."
The competition in the GPU market is becoming more intense in the arena of application-specific integrated circuits (ASICs). These chips, which are designed for a particular use such as AI workloads, are capable of faster and more efficient processing as well as handling larger datasets.
Google, for example, has developed a custom chip called Tensor Processing Unit (TPU) while Amazon has developed the Trainium and Inferentia processors. While the big players are fighting it out in the AI chip arena, venture capital investors are supporting AI&ML start-ups that can carve out their own market share and, who knows, probably come up with the next big thing.
NeuReality, an Israeli start-up that is developing inferencing accelerator chips (used in the AI process that involves logic and decision making in dealing with available data), seeks to differentiate itself in the market by offering a suite of software and services to support its hardware. It raised US$35 million in a Series A funding round led by Samsung Ventures and other funds in December 2022.
The growth of AI technology is driving the demand for high-end chips, which in turn has led to a supply shortage, a stark contrast to the overabundance issue prevalent in the overall semiconductor sector, particularly the segments focused on consumer products.