Intel and me for 25 years

The first Windows computer at my parents home was made by NEC, PC9821 architecture (different architecture to DOS/V, system BIOS and chipsets were NEC proprietary) consisting of Intel Pentium processor with around 100MHz, DSTN LCD monitor of 800 x 640, removable CD-ROM and FD drives, PS Card slot type II x 2, etc.

Since then, late 90's, I have followed the news of Intel and PC industry.

Now Intel faces the huge financial challenge, and I wanted to summarize what is the root causes of the difficulty of Intel.

Vertical integration and horizontal integration

Before going into the main topic, I'd like to say that the PC industry is my reference for analyzing an industry structure of any business.

At the dawn of personal computers in the 70's, startup started to build their devices by purchasing different components and designed their boards on their own. No standardized designs.

Eventually, CP/M software started to build a eco-system, which I believe became the foundation of so-called DOS/V architecture, with Intel processor and its compatible chips.

Huge companies entered the PC market later, often with the high degree of vertical integration of client computing devices, including proprietary software. NEC, which I mentioned is one of them, in this group of companies, there are a bunch of other names, IBM in the US, Siemens in Germany, etc.

When we look at the software landscape, 2 torrents + 1 are still worth mentioning today - UNIX and DOS, and Classical Macintosh OS.

For client computing device, originally limited hardware resources, they couldn't afford to have an OS with high degree of hardware abstraction which we see today. Direct access to hardware by application software were a common practice. And as the software start to grow its complexity, CP/M had a success to marketed to a number of hardware manufacturers, eventually replaced by DOS. As IBM decided to outsource a lot of development to Microsoft for its DOS variant, which Microsoft used as a base to sell to other manufacturers, too. Later, GUI (Graphical User Interface) was introduced widely to the market by Apple with Macintosh, and Microsoft followed the suite with Windows 1.0, GUI layer built on top of MS DOS.

While witnessing the rise of DOS variants in the market, UNIX development continued. With the launch of UNICS in 1969, various variants are made. Eventually, 3 descendants are still widely seen in our life today - Linux, BSD, and macOS.

In the process, software were eventually more standardized, for client computer, the Microsoft DOS and Windows provided softwares which can run on many hardware.

In the 90's, we saw the rise of Compaq and Dell, the company which took advantage of horizontal integration with improved supply chain management. They harnessed the economies of scale, and old-school companies like NEC, IBM Apple, who had tended to have still a high degree of vertical integration with their own chips (or with partners), switched their processors to Intel and compatible manufacturers like AMD, or continued by forming consortium like Power architecture by IBM, Motorola and Apple.

In 2002, Apple started to port OS X (called macOS now) to Intel x86 architecture, in 2005, it started to transition from PowerPC chip to Intel. The dominance of Intel in the client computing seemed so strong.

x86 chips - for client computing and data centers

Hardware exists to run softwares, or softwares define hardware in the field of computing without environmental sensors and actuators.

Today Intel and AMD are still the widely known names with their x86 and x86-64 architecture.

Since the birth of AMD, AMD have challenged Intel. In 1999, AMD launched K7, or Athlon, which threatened Pentium II and so forth.

Intel had always fought back successfully until, let's say' 10 years ago.

Intel has 2 major CPU design centers, Hillsboro Oregon in the US, Haifa in Israel. Tick Tack strategy, I still remember clearly the Pentium M processor, 130 nm process node highly energy efficient chip developed by the Haifa team. A few years ago before the launch of Pentium M in 2024, Transmeta challenged Intel and AMD with an energy efficient Crusoe chip with a unique VLIW based architecture. As more people wanted a longer battery life in their laptops, it snapped a huge interest. Intel in the end won the war of the energy efficient processor market that time.

There were many things happened in the client computing, Intel and AMD remained relevant.

When we look at the data center, Intel had a hard time to enter the market. Old school companies with main frame computers had had a long dominance with fragmented market shares.

For their high reliable scientific calculation and such, Intel released several products. Intel Pentium Pro, Itanium with IA64 architecture, etc. These efforts had never borne a huge fruit. Software developers and their employers never really fell in love with such architecture, Intel's relatively scarce resources assigned to support the software development environment

Eventually, x86 and x86-64 bit architectures had made into the servers, helped by the increasing use of Linux and Windows in the servers.

By watching the competitions of companies, I understood the the importance of having an ecosystem of working with other companies, industrial standards, and open source activities

What's gone wrong with Intel

I wrote already a lengthly article, I try to keep the main topic concise. The problems and perspectives are following;

Client computing: Intel continue to suffer in the competitive market situation

• x86-64: Intel Core (Lunar Lake) is not differentiated enough against AMD Ryzen (ZEN5), unless Intel improves its design issues and manufacturing issues, it is hard to see any significant sales and profitability improvements
• Competition vs ARM: It's clear that Microsoft has a plan to promote more and more Qualcomm Snapdragon, I'm almost certain it will be followed by Mediatek and such. Microsoft is aware of the challenge posed by Apple with Apple Silicon, and it will not change its strategy for a foreseeable future

Conventional Data centers（CPU): Intel continue to suffer in the competitive market situation

• x86-64: AMD leads the battle, I don't see fundamental change in sight. The reason is same as in the client computing segment
• Competition vs ARM: Amazon Graviton was released in 2018 for AWS, Google develops its ARM processors for Google Cloud. These cloud service providers will continue to use x86-64 but by having their own chips, they have good burgeoning powers
• RISC-V: At the moment, this competition is a small topic in terms of revenues. However, let's say in 5 years, it will be interesting to revisit this topic, as many companies struggle with the licensing policy of ARM.

Data centers for machine learning applications (GPU/TPU/NPU, etc.): Intel won't be successful in a near future

• Nvidia dominance will continues for various applications, thanks to the software developer assets Nvidia have patiently nurtured
• Other architectures: Fragmented landscape might accelerate depending on the applications - GPUs were partly replaced for blockchain application with ASIC, for its power efficiency. I have no clear picture in this market, there are so many things, but I don't really see the presence of Intel

Automotive and Embedded applications: Intel is facing more and more issues

• Intel Mobileye: While Mobileye business for driver assistance functions are still a big success in the market, Nvidia and Qualcomm threatens Mobileye with their excellent software suites. Mercedes 'recovery' in the automated driving cannot be discussed without the presence of Nvidia, I'm not well informed but I bet Qualcomm will help BMW in the same field, for example. And a bunch of car manufacturers, starting with Tesla, have their own chips deployed or in their pipelines.
• (It's off the topic, but I'm really surprised by Honda CES2025 announcement of 2000 TOPS chip with Renesas. I really wish they can prove their relevance. )
• Embedded: x86, a popular architecture for Windows-based embedded application, will face the competition same as the client computing segment.

Foundry service: Not sure but we will see in a few years, what the ousted CEO Pat Gelsinger of Intel envisioned will be successful or not

Lisa Su on Pat Gelsinger - People make the world go round

Lisa Su, is the CEO of AMD, relative of Jensen Huang, Nvidia CEO. The manager with engineering background is well credited for turning around the business of AMD.

Then Jacobs brought up her competitors, Nvidia and Intel. Intel’s CEO Pat Gelsinger recently stepped down, and when Su was asked who should replace him, she did not name names but said she has “a lot of respect for Pat as a colleague” and emphasized that the role is “a tough job.” When Jacobs asked whether President Joe Biden’s administration has encouraged her to merge AMD with Intel, she said no.

Source: CEO of the Year Lisa Su Talks Competition and Collaboration in the Semiconductor Industry https://time.com/7199761/year-in-time-ceo-lisa-su-2024/

I felt that she meant what she said, when I read this comment.

Pat Gelsinger was a known as a great engineers back in the time before early 2000's, she must know all his achievements, as she herself is a great engineers credited for Silicon-on-Insulator design and such.

I like reading about the semiconductor industry, it's a full of talented people working hard to make things happen.

I have worked in different industries and now in the electronics industry with my current employer, and I truly believe that the semiconductor industry has the best engineering talents in the manufacturing world.

I love tech companies and I'm frustrated with the financial industry and the consulting companies, which attract highly intelligent people with high salary and other financial incentives.

I am not against the finance per se, I see a great value in it. Project financing, islamic finance, microcredit service and such, especially when they are combined with the power of IT, can significantly improve our social well beings. But I feel like these talented people are either not given a chance, or pessimistically, not motivated to bring social impacts which their organisations could bring. And if these talents are given an incentives to work to improve our societies, I'm sure we could solve many severe issues that we face today. In other world, today's democratic market economies of the most countries have bad incentive schemes.

I'm a big fan of technology, which is an enabler for a better world. I want to see that we will see the world where labour is not a must. The automation technologies should drastically reduce our boring jobs while the income guarantee should come from other wealth distribution mechanism than the work. The lack of time and of money make people act stupidly, that is what I see everyday and this is a root cause of many, if not most, social issues. I often see myself acting not wisely, and I can imagine many other people might have the same issue. A wise use of technology can change it.

The work of future, it will always have to do with the advancement of technology to support the humanity. I'm not a big contributor for this until now, but this is the passion I have and I'm convinced that I'm on the right side of history. Technology is great, but the greatest wisdom of human being is the ability to harness the use of technology in business and in private life.

Learning about the people like Pat Gelsinger, Lisa Su are so inspiring. In the semiconductor industry, Jim Keller is another figure who strongly attract my interest. Learning about their successes and failures have given me a lot of ideas, how I want to lead my business activities.

Thank you, Intel, for giving me a lot of different thoughts over 25+ years.

And I wish the company a good luck, it is too important to fail in our world for business and international security of 2030's. In whichever forms it can have in future, its people and assets will help us advance our society.