If You’re Reading This Semiconductors & AI Are Taking Over

There's no summertime sadness in the semiconductor industry, thanks in part to chip demand from generative AI applications like OpenAI's ChatGPT.

In recent years, the semiconductor industry has experienced a notable departure from its traditional cyclical patterns, embracing a more secular behavior that defies the ups and downs tied closely to economic cycles. This transformation can be attributed to a range of factors, including the ever-expanding presence of technology in our daily lives and the surging demand for semiconductors across diverse sectors. Unlike its past volatility, the semiconductor market now exhibits remarkable resilience, driven by long-term trends such as artificial intelligence, Internet of Things, and cloud computing. This shift reflects the industry's capacity to adapt to evolving market dynamics, sustaining consistent demand and fostering innovation even amidst economic fluctuations.¹

As we continue to ride the wave of technological advancement, semiconductors, and artificial intelligence (AI)—stand at the forefront of this evolution. Together, they drive each other's growth, propelling us into an unprecedented era of technological revolution. As these two domains deepen their relationship, they're heralding a new surge of innovation. The future of AI is anchored in the creation of new, AI-optimized semiconductor chips—a dynamic we believe is poised to fuel significant growth in the next half-decade. This blog examines the interdependence of semiconductors and AI, the technological advancements driving their integration, and the future outlook of this thriving industry.

• How Are Semiconductors Behind ChatGPT’s Success?
• NVIDIA's Q1’23 Earnings Highlight AI and Semiconductor Market Capture
• Is AI Accelerating Semiconductor Design and Manufacturing?
• AI Design Tools’ Growth Rate Projected to Surpass Chip Sales by 3x?
• Semiconductor Industry Interlude: 4 Key Categories of Participants
• How Will US CHIPS Act Benefit Large Semiconductor Producers?

How Are Semiconductors Behind ChatGPT’s Success?

Machine learning and its insatiable appetite for data heavily rely on semiconductors for processing and storage. The growing complexity and demand of AI applications mean that semiconductors, particularly graphic processing units (GPUs), are increasingly important. With their multitasking capabilities, GPUs are crucial in managing the massive data required for AI model training, ensuring applications like ChatGPT operate efficiently.

As the applications grow increasingly complex, the demand for advanced semiconductor chips will escalate. This growing demand represents a significant and promising opportunity for companies in the semiconductor industry. For instance, graphic processing units (GPUs), are the powerful engines that drive companies like OpenAI and its applications, including ChatGPT, are in high demand. With their capability to handle multiple tasks simultaneously and more importantly proficiently, GPUs are the perfect fit for the heavy-duty processing required by machine learning. These processing engines manage the vast data needed for training generative AI models that use generative pretrained transformers (GPTs), enabling quicker responses and better language understanding. Simply put, semiconductors are the foundational support structure, ensuring the smooth operation of AI applications such as ChatGPT.

GPUs also play a pivotal role in a broad spectrum of AI applications, including:

• Natural language processing: GPUs decode the intricacies of human language, finding their place in applications such as voice assistants and machine translation.
• Facial recognition: GPUs identify individuals in images and videos, a tool with applications in sectors from security to marketing.
• Autonomous vehicles: GPUs adeptly manage sensor data, guiding vehicles safely through city streets.

NVIDIA's Q1’23 Earnings Highlight AI and Semiconductor Market Capture

NVIDIA (NASDAQ: NVDA) recently posted strong earnings, with both revenue and earnings per share surpassing analyst predictions.² The rise in demand for NVIDIA's GPUs, used extensively in AI applications, is driving the company's growth. As the demand for AI-powered applications rises, so will the demand for semiconductors and we believe NVIDIA is well-positioned to benefit from this growth. It is a leading GPU supplier for AI applications and is reaping benefits from partnerships with AI companies like OpenAI and Google AI.

Global Semiconductor Market Growth and Forecast

Source: World Semiconductor Trade Statistics and Mckinsey’s projected yearly growth. Past performance is no guarantee of future results. Not intended as a recommendation to buy or sell any securities mentioned herein, or as any call to action. For illustrative purposes only. Actual future semiconductor market growth is unknown.

The need for increased computing power, AI, and digital connectivity has resulted in significant semiconductor market growth. McKinsey estimates that semiconductor revenue could double from $550 million in 2022 to over $1 trillion by 2030.³

Is AI Accelerating Semiconductor Design and Manufacturing?

AI is not just benefiting from semiconductor power; it has emerged as a significant driving force in the evolution of the semiconductor industry. Through its ability to redefine chip designs, identify defects, optimize processes, and predict chip failures, AI is enhancing efficiency and profitability within the sector. And there's more to the story—AI is catalyzing the creation of a new lineage of chips tailor-made for AI's distinctive requirements and needs. These aren't your regular chips, but one’s tailor-made for AI's unique demands. Think of Intel (NASDAQ: INTC) and its pioneering work crafting AI-optimized chips that excel in speed, efficiency, and power management.

AI Beats a Human at Chess and Go, Now Watch It Design Chips

Chips have exponentially more configurations than chess or Go⁴

Source: Deloitte Insights adapted from Synopsys, “What is Design Space Optimization?” as of July 2020.

AI has previously proven its capability by outperforming humans in complex board games like Chess and Go, but now it's stepping into an even more intricate field: chip design. With an astronomical number of potential configurations, far surpassing those in Chess or Go, the ability to optimize and innovate in chip design could be a game-changing application of AI technology. As the relationship of AI and semiconductors deepens, they're paving the way for a fresh surge of technological innovation. The forthcoming progression of AI is heavily dependent on the creation of new semiconductor chips that are specifically designed for AI applications. We believe this trend is likely to fuel significant growth in the next five years.

AI Design Tools’ Growth Rate Projected to Surpass Chip Sales by 3x?

5-year CAGR for chips, EDA tools, and advanced AI design tools (2023-2028)⁵

Sources: WSTS; Global Markets Insights; and Deloitte Global as of November 2022.

AI's imminent growth largely hinges on the development of semiconductor chips optimized explicitly for AI functionality. According to Deloitte, it is very plausible that this factor can trigger substantial growth over the next half-decade.

Semiconductor Industry Interlude: 4 Key Categories of Participants

The semiconductor industry is principally composed of four key categories of participants. The initial category comprises foundry operators, for instance, Taiwan Semiconductor Manufacturing Company (TSMC). These are businesses that undertake the production of semiconductor chips based on the specifications given to them by other companies. They can be thought of as factories crafting chips for different companies. The second category is filled with integrated device manufacturers (IDMs), with companies like Intel serving as examples. IDMs are unique because they handle the entire lifecycle of a semiconductor chip within their own facilities, from the initial design to the final manufacturing phase. The next category includes companies that only design chips also referred to as fabless companies. NVIDIA and Advanced Micro Devices (AMD) fall into this category. Fabless companies concentrate on designing and developing semiconductor chips but outsource the manufacturing to foundry operators since they do not have their own fabrication capabilities. The final and fourth category encompasses equipment manufacturers such as ASML. These businesses are responsible for creating the necessary machinery and equipment used in the manufacturing of semiconductor chips. They supply the crucial tools needed for both foundry operators and integrated device manufacturers to construct the chips.

How Will US CHIPS Act Benefit Large Semiconductor Producers?

Remember the US CHIPS Act? Since semiconductors are critical for the economy, the U.S. aims to become self-sufficient in semiconductor production and reduce dependence on foreign supply chains. The Covid-19 pandemic highlighted vulnerabilities in the supply chain, often including steps in over five countries and multiple global shipments. The Department of Commerce reports that in 2021, semiconductor shortages cost the U.S. economy nearly a quarter trillion dollars.

To increase domestic manufacturing of semiconductors, the US Chips and Science Act will channel $50 billion in investments into the industry over five years, including $11 billion for advanced semiconductor research and development, and $39 billion to accelerate and drive domestic chip production ($6 billion of which can cover direct loans and loan guarantees)⁶.

Budget To Expand Domestic Manufacturing (Advanced Semiconductor R&D and Chip Production)

Source: McKinsey⁷ as of October 2022.

Among the main beneficiaries of the US CHIPS Act are the U.S.’s large semiconductor producers. Instead of trying to pick individual stock winners in the sector, investors can gain exposure to the 25 largest most liquid, U.S. listed semiconductor companies in The VanEck Semiconductor UCITS ETF (SMH).

In closing, we believe semiconductors could be the 'picks and shovels' way to play the AI landscape and present a compelling way to capitalize on the growing AI sector, particularly when direct access to private AI companies is limited for many investors. The VanEck Semiconductor UCITS ETF (SMH) provides a way to invest in the entire value chain of the semiconductor industry, from chip design and fabrication to the machinery used in the manufacturing process. As semiconductors are the essential components that power AI innovation, we believe they are poised to gain value amid the potential deflationary impact of AI's efficiency—they also provide a unique opportunity to ride the wave of AI's transformative impact.