Intel CEO Pat Gelsinger speaks at the World Economic Forum in Davos, Switzerland — 1/17/2024

18 Jan 2024 (10 months ago)

Introduction

The semiconductor industry can be categorized into four main types of chips: memory chips, specialty chips, mature chips, and Leading Edge chips.
Long lead time investments in semiconductor factories and the inability to accurately predict future demand contribute to the cyclicality of the industry.
Currently, there is a reasonable supply-demand balance in the semiconductor industry, but oversupply of memory chips and a potential undersupply of memory chips in the future.
AI is driving the demand for Leading Edge chips and will continue to do so in the coming decade.

No vote was taken in Brussels or DC to get rid of the semiconductor industry, but Asian countries such as Korea, Taiwan, and China put in long-term industrial policies and investments to attract the industry, resulting in the loss of factories in the US and Europe.
The need for balanced and resilient supply chains led to the US and the EU chips act.

Becoming a cutting-edge semiconductor supplier requires significant investment, both in terms of capital (billions of dollars per year) and R&D (billions of dollars per year).
There are only three companies in the world capable of producing Leading Edge chips: TSMC, Samsung, and Intel.
Cutting Edge chip production is capital-intensive, research-intensive, and requires significant intellectual property.
Building a new factory can cost around $30 billion, making it the largest construction project on Earth.

Intel plans to build an advanced manufacturing facility in Magur, Germany, that will produce 1.5nm devices, making it the most advanced manufacturing facility in the world.
Intel already operates advanced manufacturing in Ireland (4nm) and plans to bring manufacturing back to the US with the Silic and Heartland project.
Intel designs its own chips and has also started becoming a Foundry for chips.

AI has become an insatiable demand for compute power, driving the need for advanced logic requirements in the industry.
The next few years of AI will focus on deploying and utilizing AI models for inference and edge computing.
Research in AI is mainly driven by the industry rather than the scientific world, with AI experiencing a 50-year overnight success.
Long-term research investment is crucial for the continued advancement of AI.

The shift of semiconductor research and development to the industry from the scientific world has been a long-term trend.
Western democratic-oriented countries need to rebuild their long-term research capabilities to stay competitive.
The dependence on foreign supply chains, particularly China, poses risks to national security.
Renewable energy and sustainability are key considerations in the semiconductor industry.

The exponential growth in AI and compute demand will ensure a continued need for advanced logic capabilities.
Even with several investments in different countries, overcapacity is not a major concern due to the insatiable demand for compute power.
The future of the industry lies in the deployment and utilization of AI models rather than just training.

Rebuilding the semiconductor industry will take time and require a combination of government policies, investments in R&D, and capital.
Intel aims to regain unquestioned technology leadership through its manufacturing capabilities.
The industry needs long-term sustainable policies to avoid further dependency on foreign supply chains.
The Chips 2 Act may be necessary in the future to continue rebuilding the industry.

The semiconductor industry is complex and requires significant conceptual thinking and leadership.
Intel is committed to rebuilding the industry, focusing on advanced manufacturing, R&D, and sustainability.