AI Chipsets: The FOMO Stock Rally

Artificial Intelligence (AI) is emerging as one of the most spoken-about novel technologies. It is evolving as a transformative force driving productive changes across industries. The boom in AI technology has developed a huge market opportunity for AI chipsets as they offer high-speed processing, low latency, and parallel computing capabilities to support the development and deployment of the AI infrastructure across various industries. AI chips offer energy efficiency, particularly in edge computing and IoT devices, and are suited for setups with power constraints. 

The AI chip market is anticipated to grow significantly at a CAGR of 40.5% by 2031.  The market is majorly driven by the emerging trends in the automotive industry. The growing adoption of technologies such as AI and Machine Learning (ML), is redefining information storage and computing in modern vehicles.  Autonomous vehicles are heavily dependent on a combination of sensors, cameras, radars, Lidar (Light Detection and Ranging), and other technologies that require the computational power of the AI chipset to handle complex perception tasks.

The market also sees opportunities in the rising demands of AI-based, Field Programmable Gate Array (FPGAs). This is because FPGAs offer increased flexibility and programmability compared to the fixed functions ASICs (Application Specific Integrated Circuits).  The AI models are growing rapidly and thus offer potential opportunities to FPGAs attributing to their tremendous capability to reprogram and reconfigure diverse AI workloads, hence, providing competitive advantage.    

The AI chip market is estimated to demonstrate the highest growth in the North American region. This can be primarily attributed to the presence of technology leaders such as Nvidia, Intel, Advanced Micro Devices (AMD), and more. These are launching new products to expand their product portfolios to stay competitive in the market. This has resulted in the emergence of Nvidia as a global chipset manufacturer, establishing it as a $2.0 trillion company. 

The AI chipmaker’s share price has surged by almost 450.0%, since January 2023. This has made Nvidia the third-most valuable firm in America.   The company reported full-year revenue of $60.9 billion, up by 126.0% in 2024. This made Nvidia’s stock surge which resulted in the AI FOMO (Fear-Of-Missing-Out) stock rally. This was seen as an eclipsing scenario for the competitors including Intel and AMD. Thus, analysts expect bolstering growth for Nvidia in the foreseeable future. 

However, the event witnessed benefits for other companies as well.  A 10.0% surge in the stock of AMD and a 2.5% increment in Palantir stock, was noted. Besides, the stocks of other chipmakers, such as Broadcom and Marvell Tech also surged in the AI FOMO. However, the Chief Financial Officer (CFO), Colette Kress, presented some concerns as the chips are majorly designed for inferencing (generation of content such as text, images, and more based on a specific input) only, thus, can be taken over by less powerful and more pocket-friendly chips. 

Nvidia majorly manufactures Graphics Processor Units (GPUs) or accelerators which are primarily used in video games. They use parallel processing, enabling faster calculations as against sequential task completion. GPUs are also useful for companies working on Large Language Models (LLMs), such as ChatGPT or Bard. GPUs would help filter information from the massive volumes of data. 

Thus, the business of competitors and CPU (Central Processing Unit) manufacturers such as Intel and AMD is likely to propel, as the customers will focus on cutting down the operating costs levied by the AI models. Further, the chips that are manufactured by Intel are already widely used in inferencing and do not require the computational power of Nvidia's cutting-edge and more expensive H100 AI chips, for accomplishing the purpose. 

Additionally, the major competitors have also started the in-house manufacturing of AI chips. For instance, in December 2023, AMD launched its MI300 accelerator. These are well-suited to power AI and HPC workloads, with improved computing, memory density, and bandwidth memory. They are also manufactured to support specialized data formats. Intel was yet another competitor to proactively respond to the situation and is preparing for the launch of its Gaudi3 accelerator in 2024. However, not many details, about the AI chip specifications, have yet been shared by the company.  

This was a defensive strategy adopted by these organizations, to counter the revenue decline. For instance, AMD recorded a net revenue of $22.7 billion, in 2023, marking a decrease of 4.0% compared to 2022 net revenue of $23.6 billion. The decline was primarily due to a 25.0% decrease in client segment revenue, resulting from lower processor sales, and a 9.0% decrease in gaming segment revenue primarily due to lower semi-custom product sales. 

Intel reported a revenue of $54.2 billion, in 2023, which was down $8.8 billion, or 14.0%, from 2022 revenue of $63.1 billion. This decrease was also attributed to the sales decline faced by the Datacenter and AI Group (DCAI) segment. DCAI revenue decreased by 20.0% due to lower server volume attributing to the softening CPU data center market.

Furthermore, other hyperscalers (large cloud service providers) such as Microsoft, Google, Amazon, and Meta are shifting their focus toward the production of their chips in the form of Application-Specific Integrated Circuits (ASICs).

ASICs are different from GPUs, as the former are designed to master the single trade they are designed for, whereas, the latter are manufactured to an array of AI-related tasks to cater to the diverse needs of the organization. However, the shift of hyperscalers towards the in-house ASICs would lead to a substantial decline in the need for Nvidia’s chips.

However, as inferencing becomes the primary use case for AI chips with widespread adoption of different AI models, the revenue of Nvidia is estimated to rise, even if it faces a minimal hit initially. In a nutshell, provided the market players keep a check on data security, safeguarding it from unauthorized access, breaches, and misuse, the market has enormous growth potential in the forecast period (2024-2031).