AI is the New Chips
-Of Gordon Moore and DeepSeek: Moore's Law Expands into AI
Dr. Gordon Moore, co-founder and chairman emeritus of Intel Corporation, was the undisputed king of semiconductors since the 1960s. His company was my archrival during my tenure at AMD and later became a key supplier when I served as a computer executive at Vtech, Compaq, and Silicon Graphics. Moore’s Law, his groundbreaking observation, stipulates that the number of transistors in an integrated circuit (IC) doubles every 18 to 24 months.
Remarkably, Moore’s Law now seems to extend its reach beyond semiconductors into the realm of artificial intelligence (AI).Since DeepSeek’s AI announcement in December 2024, headlines have been dominated by its breakthroughs, applications, and implications. What truly stands out, however, is not China’s entry into the AI race three years after ChatGPT’s global debut, nor the obscure company’s ability to create a large language model (LLM) - algorithm that applies neural-network techniques to understand,process, and generate human languages - with world-class mathematical and engineering capabilities. The real marvel lies in DeepSeek’s staggering cost reduction for training LLMs. While Meta spent US$6million on developing its Llama 3.1 LLM model, DeepSeek achieved similar results for just US$6. This is not merely a technical feat—it’s a paradigm shift, redefining efficiency and cost as the new battlegrounds in AI. DeepSeek’s achievements challenge industry giants, accelerate AI democratization, and force a re-evaluation of traditional strategies.
From my vantage point in Hong Kong, our AI ecosystem faces significant challenges, particularly a 49% talent shortage among firms. To bridge this gap, the HKSAR government, along with the localtech community, is investing strategically, such as the 2024-2025 High-Performance Computing (HPC) center and collaborations with Shenzhen.
The world is in the midst of the Fourth Industrial Revolution, which has brought forth the advent of a new "Imagination Age". Creativity and imagination have now become the primary drivers of economic value, as well as national power and confidence. As a global nexus and arguably China’s crown jewel, Hong Kong must balance high costs and East-West dynamics while mirroring the efficiency gains seen in semiconductors.
Drawing on my 40 years in semiconductors and computing, I propose a compelling hypothesis: AI is undergoing a transformation akin to the semiconductor revolution. Moore’s Law, once the guiding principle of technological progress, is now influencing AI. Just as it drove exponential growth in computing power, a similar dynamic is propelling AI forward-albeit with unique challenges.
Moore’s Law, articulated by Gordon Moore in 1965, states that the number of transistors on a microchip doubles every 18 to 24 months. This exponential growth has driven the technological revolution for six decades, making devices faster, lighter, more efficient, and increasingly sophisticated. From smartphones and drones to wearables like FitBits and iWatches, this relentless progress has transformed our world. The semiconductor industry’s ability to consistently improve performance and reduce costs has set a benchmark for other fields. AI isnow following a similar trajectory.
It’s evidentto me that “AI is the new chips,” with Moore’s Law beginning to shape its development. The parallels are striking: AI is experiencing rapid evolution, exponential growth, and cost reductions fueled by innovation and competition. However, AI diverges from Moore’s Law in a critical way. While semiconductor progress is constrained by the physical limits of transistor size, AI’s advancements rely on algorithmic breakthroughs, data availability, and computational power. Additionally, AI must navigate geopolitical tensions, legal challenges, and regulatory hurdles—factors the semiconductor industry largely avoided in its early days. These complexities could either accelerate or hinder AI’s progress, shaping its trajectory in ways distinct from the semiconductor revolution.
China is poised to play a pivotal role in driving down the costs associated with AI, leveraging its vast resources, talent pool, and strategic investments in technology. However, whether China will ascend to dominance in this field remains an open question. Even if it does, maintaining that position will require continuous innovation and the ability to outpace competitors. History offers a cautionary tale in the form of Intel, which reigned supreme in the semiconductor industry for over two decades, only to be recently challenged and surpassed by formidable contenders such as ARM, AMD, nVidia, Samsung, and TSMC. In a twist of fate, TSMC is now contemplating the acquisition of Intel's wafer fabrication operations, underscoring the dynamic and ever-shifting landscape of the industry. This serves as a reminder that no leader, no matter how dominant, can afford complacency in the face of relentless competition and technological change.
Ray Dalio’s seminal work on the rise and fall of empires applies equally to corporations. Giants like General Motors, U.S. Steel, and GE have faltered, reminding us that no entity is immune to change. The AI industry, too, must heed this lesson. As barriers to entry lower, smaller players and startups can challenge established giants. The winner-take-all playing field has been replaced as this democratization of AI is leading to a more diverse and innovative ecosystem. However, this also raises questions about sustainability, ethical considerations, and the potential for misuse. The role of regulation and international cooperation will be critical in ensuring that AI advancements benefit humanity as a whole, rather than exacerbating existing inequalities or creating new risks.
In conclusion, DeepSeek’s achievement heralds a new era defined by efficiency, cost-effectiveness, and rapid innovation. AI’s future will be shaped by technological breakthroughs, geopolitical dynamics, and regulatory frameworks. The lessons I learned working in semiconductors, as well as observing the rise and fall of corporate giants for decades, offer valuable insights into navigating this transformation. Whether AI will drastically change the world is no longer a debatable point. It already has. The important question is how we will steer this change toward a future that is equitable, sustainable, and prosperous for all.
About the author:
Philip Leung is Founding President of AAMA (Greater Bay Area). Technology Executive, highly experienced in international market entry. MBA from Harvard University. BSc in Electrical Engineering from University of Hong Kong, where he is Adjunct Professor. He has served as President of the Canadian Chamber of Commerce, Governor of the American Chamber of Commerce, and Adjunct Professor at the City University of Hong Kong. He teaches occasionally in HKU, CUHK and HK Polytechnic University. He has lived in various cities in Asia and the US, and worked in the international technology business community for more than 40 years.
Комментарии