NVIDIA and its partners: Can their AI chip dominance last?

NVIDIA and its partners: Can their AI chip dominance last?Since the advent of ChatGPT, NVIDIA and its key manufacturing partners SK Hynix, TSMC, and ASML have firmly grasped the AI supply chain, boasting a combined market capitalization exceeding $4 trillion. These four companies hold 80-100% market share in their respective sectors, a near-monopoly that has sparked widespread concern: How long can it last? How did this unprecedented market dominance emerge? And what potential challenges and threats exist? This article delves into the rise of this new generation of tech monopolies NVIDIA and its partners exploring their path to success, the challenges they face, and their future trajectory

NVIDIA and its partners: Can their AI chip dominance last?

Since the advent of ChatGPT, NVIDIA and its key manufacturing partners SK Hynix, TSMC, and ASML have firmly grasped the AI supply chain, boasting a combined market capitalization exceeding $4 trillion. These four companies hold 80-100% market share in their respective sectors, a near-monopoly that has sparked widespread concern: How long can it last? How did this unprecedented market dominance emerge? And what potential challenges and threats exist? This article delves into the rise of this new generation of tech monopolies NVIDIA and its partners exploring their path to success, the challenges they face, and their future trajectory.

NVIDIA's Rise: A Disruptive Shift from Gaming to AI

For decades, NVIDIA was renowned for its achievements in the gaming sector. Its GPUs (Graphics Processing Units), leveraging parallel computing, enabled the rapid generation of realistic game graphics, finding widespread application in games like Call of Duty. However, over a decade ago, forward-thinking researchers discovered that these GPUs were equally suitable for deep learning a computing method mimicking the human brain, which underpins today's large language models like ChatGPT.

NVIDIA keenly recognized this opportunity. Its CEO, Jensen Huang, delivered a $129,000 chip set to OpenAI, then a small research lab, in 2016, providing crucial support for its deep learning research. Harvard University economics professor Jason Furman commented, Initially, it was almost accidental, and then they cleverly seized the opportunity.

NVIDIA's success wasn't accidental. It built a vast code library around CUDA (Compute Unified Device Architecture), a programming language that made GPUs the only viable path for this new type of computation. With a large number of AI engineers already accustomed to using CUDA, alternative chips developed by well-funded startups and giants like Google have consistently failed to challenge its position, even outperforming former chip behemoth Intel. On March 18, 2024, Jensen Huang launched NVIDIA's latest generation of high-performance chips and accompanying software, Dynamo, calling it the operating system for AI factories, further solidifying its leadership in the AI field.

The Formation of a Monopoly Chain: Close Collaboration between NVIDIA and its Partners

To maximize GPU performance, NVIDIA relies on high-performance memory chips. Here, SK Hynix plays a crucial role, controlling approximately 80% of the global market for high-bandwidth memory (HBM) chips. SK Hynix's success wasn't overnight; its development of a novel packaging technology in 2019 solved overheating issues in data-intensive AI tasks, surpassing long-time rival Samsung Electronics.

NVIDIA itself doesn't own chip fabrication plants; all its GPUs are outsourced to TSMC (Taiwan Semiconductor Manufacturing Company). This Taiwanese company, with its advanced "fabless" model and superior manufacturing processes, dominates the global chip foundry industry. In 2013, Apple shifted iPhone and iPad component orders from Samsung to TSMC, a pivotal turning point in TSMC's history. Since then, giants like Intel and Samsung have repeatedly attempted to break TSMC's monopoly, all without success. TSMCs claim last year that it manufactured 99% of the worlds AI accelerators underscores its core position in the AI supply chain.

However, the true "monopoly king" in the AI field might be the Dutch company ASML. It exclusively manufactures the extreme ultraviolet lithography (EUV) machines required by TSMC and others to produce chips meeting NVIDIA and Apple's specifications. With a unit price of $380 million and incredibly high technological barriers, ASML's dominance is unmatched in the foreseeable future.

Monopoly Controversy: Market Share and Market Behavior

In terms of market share, NVIDIA and its partners undoubtedly hold a monopolistic position. Over 70% market share and high barriers to entry classify them as a monopoly. However, determining whether they abuse their market dominance like traditional monopolies requires further investigation.

Soaring chip prices are a clear indicator. In recent years, the price of a single NVIDIA GPU has skyrocketed to $90,000, and key component suppliers like SK Hynix have also raised prices due to buyers' lack of alternatives. This has directly resulted in NVIDIA's gross margin exceeding 70%, significantly higher than its competitor AMD's 50%. Theoretically, companies can pass costs onto consumers, but currently, major tech companies like Microsoft, Google, Amazon, and Meta, as NVIDIA's largest customers, willingly absorb the enormous computing costs to maintain their leading positions.

However, this situation isn't static. Large cloud computing companies are actively developing alternatives to reduce their reliance on NVIDIA. Amazon is developing custom chips, Microsoft is supporting AMD's development of AI accelerators, and OpenAI has even submitted chip design proposals to TSMC, attempting to circumvent NVIDIA.

The Persistence and Challenges of Monopoly: Historical Experience and Future Outlook

Tech monopolies often exhibit persistence. IBM, Microsoft, Intel, Google, and Apple have all maintained dominant positions in their respective fields for extended periods. These monopolies are typically difficult to dislodge, requiring proactive regulation and strong competitors to act as counterbalances.

Historically, antitrust actions, technological innovation, and new market demands have all led to the decline of monopolies. The US action against Microsoft, while not resulting in a breakup, diverted its attention and ultimately facilitated the rise of Apple and Google in the mobile market. The decline of Blackberry and Nokia also demonstrates how technological innovation can disrupt established companies.

While NVIDIA's current position is strong, challenges remain. DeepSeek's introduction of lower-cost AI models initially raised concerns about NVIDIA's future prospects. Although NVIDIA's stock price subsequently rebounded, DeepSeek's success suggests that building competitive AI applications doesn't necessarily require massive hardware investments.

Furthermore, NVIDIA faces regulatory risks. The US Department of Justice is investigating potential anti-competitive practices, such as prioritizing chip allocation to specific clients.

Conclusion: The Future of the AI Chip Landscape

NVIDIA and its partners currently have a firm grip on the AI chip market, but this doesn't guarantee their dominance will last forever. Technological innovation, the rise of competitors, and regulatory risks all pose challenges. The future will see even fiercer competition in the AI chip sector; those best adapted to technological advancements and market demands will prevail. The DeepSeek event, while causing a short-term impact on NVIDIA, also showcased market dynamism and innovation potential. For investors, the AI chip sector remains full of opportunities and challenges, requiring careful risk assessment for success in this rapidly evolving market. Whether it's NVIDIA and its partners or emerging AI companies, continuous efforts in technological innovation, market strategy, and risk management are essential for survival and growth in this uncertain field. The pace of AI development far exceeds expectations, and the future holds limitless possibilities. NVIDIA and its partners' ability to maintain their leading position will ultimately depend on their capabilities in technological innovation, market strategy, and navigating regulatory challenges.