Artificial Intelligence has come a long way since the success of DeepMind over Go world champion Lee Sedol in 2016; the world is beginning to change according to the new possibilities afforded by AI. From the robust predictive abilities of OpenAI’s ChatGPT – where the AI chatbot can be used for all sorts of purposes, from creating scripts (including malware) to writing academic essays – to AI image generators that are so good that they can win Sony world photography awards, the complexity and capabilities of AI algorithms are growing at a startlingly fast pace.
Hardware is hardly divorced from this. As machine learning workloads become more complex and compute-hungry, hardware must scale appropriately to ensure cost-efficiency for end users without stalling progress in the software domain. Hardware and software are tightly linked, and the matter of control and ownership of AI hardware is emphasized in IDTechEx’s latest report, “AI Chips 2023-2033”.
The price of failure
2016 may have been the year that the world first took notice of the reality of AI, but IDTechEx believes that 2020 will be the year that is remembered as a turning point in technology initiatives across the globe. Chips for AI training – where training refers to providing AI algorithms with large datasets, such that the algorithm can adjust its weights to better fit the provided data – are typically at the most leading-edge nodes, given how computationally intensive the training stage of implementing an AI algorithm is.
Intel, Samsung and TSMC are the only companies that can produce 5 nm node chips. Of these, TSMC is currently the only company with any real success securing orders for 3 nm chips. TSMC is a Taiwanese company, Samsung South Korean. TSMC has a global market share for semiconductor production, currently hovering at around 60%.
For the more advanced nodes, this is closer to 90%. Of TSMC’s six 12-inch and six 8-inch fabs, only two are in China, and one is in the US. The rest are in Taiwan. Therefore, the semiconductor manufacturing part of the global supply chain is heavily concentrated in the APAC region, principally Taiwan.
Such a concentration comes with great risk should this part of the supply chain be threatened. This is precisely what occurred in 2020 when several complementing factors (such as Covid, the rise of data mining, a Taiwanese drought, fabrication facility fire outbreaks, and neon procurement difficulties due to the Russia-Ukraine war) led to a global chip shortage, which demand for semiconductor chips exceeded supply.
Since then, the largest stakeholders (excluding Taiwan) in the semiconductor value chain (the US, the EU, South Korea, Japan, and China) have sought to reduce their exposure to a manufacturing deficit should another set of circumstances arise that results in an even more exacerbated chip shortage.
But this is not the only reason national and regional government initiatives have been implemented to incentivize semiconductor manufacturing companies to expand operations or build new facilities. The manufacture of advanced semiconductor chips fuels national/regional AI capabilities.
These capabilities, in natural language processing (understanding of textual data, not just from a linguistic perspective but also a contextual one), speech recognition (being able to decipher a spoken language and convert it to text in the same language or convert to another language), recommendation (being able to send personalised adverts/suggestions to consumers based on their interactions with service items), reinforcement learning (being able to make predictions based on observations/exploration.
Such as is used when training agents to play a game), object detection, and image classification (being able to distinguish objects from an environment and decide on what that object is), are so significant to the efficacy of certain products (such as autonomous vehicles and industrial robots) and models of national governance and security, that the development of AI hardware and software should be at the top of the agenda for any government body that wishes to be at the technological forefront.
IDTechEx forecasts that the global AI chips market will grow to US$257.6 billion by 2033. How this pie is sliced will largely depend on how effective funding initiatives are. IDTechEx’s report breaks down each major national and regional funding initiative related to semiconductor manufacturing, analysing where funding is coming from, how this will be dispensed, and the relative effectiveness of each governing body’s initiatives.
The geopolitical background is given, difficulties and opportunities are presented, and each region’s use and expertise with AI are detailed. In addition, the major private investments made and announced since 2021 about semiconductor manufacture are surveyed and contextualized within overarching funding initiatives.
The report covers the global AI Chips market across eight industry verticals, with 10-year granular forecasts in seven categories (such as by geography, chip architecture, and application).
In addition to the revenue forecasts for AI chips, costs at each stage of the supply chain (design, manufacture, assembly, test & packaging, and operation) are quantified for a leading-edge AI chip. Rigorous calculations are provided, along with a customisable template for customer use and analyses of comparative costs between leading and trailing edge node chips.
IDTechEx’s latest report, “ AI Chips 2023-2033”, answers the major questions, challenges and opportunities faced by the AI chip value chain. Please refer to this report to understand the markets, players, technologies, opportunities, and challenges.
Leo Charlton is the technology analyst at IDTechEx.