How Intel went from iconic chimaker to AI laggard

[Daniel Nenni]

At the time, I thought that DEC management missed the boat of PC market and Compaq basically ruined DEC business (I was using the VAX system was very happy with it) and was surprised that HP swallowed the bad pill. But that was my opinion at the time. I don't know what would have happened if DEC would have gone the PC market way.

I worked for Data General right out of college. We had an old DG Nova system at school so I was ready for DG. Plus I loved the book Soul of a New Machine by Tracy Kidder. It was on the reading list for a CPU architecture class. Great book back then.

Unfortunately the Micro Vax really killed the minicomputer business then came Sun Micro Systems and that did it for DEC and DG. DG tried the PC market with a desktop and the DG1 laptop. I think it was one of the first laptops. The LCD screen was not great so it did not take off. We called it a luggable because it was heavy. DG was in no shape to compete with the likes of Compaq anyway. EMC bought DG for the storage business in 1999 and DG was officially dead shortly after.

 

[Michael Domnitei]

I worked for Data General right out of college. We had an old DG Nova system at school so I was ready for DG. Plus I loved the book Soul of a New Machine by Tracy Kidder. It was on the reading list for a CPU architecture class. Great book back then.

Unfortunately the Micro Vax really killed the minicomputer business then came Sun Micro Systems and that did it for DEC and DG. DG tried the PC market with a desktop and the DG1 laptop. I think it was one of the first laptops. The LCD screen was not great so it did not take off. We called it a luggable because it was heavy. DG was in no shape to compete with the likes of Compaq anyway. EMC bought DG for the storage business in 1999 and DG was officially dead shortly after.

We used tapes and optic disks to back-up data on the VAX, also e-mail and rn (read news) on the internet before the first windows showed-up. Regarding the laptops, I used special laptops in 1986/1988 at NEC America for testing boards (with IEEE 488 & 232). They were quite chunky (about 2.5 inches thick) but they were Japanese laptops specially designed for the board and equipment testing (very suitable for manufacturing floor of the mobile phone which was a 4 boards equipment - the size of 2 bricks) to be used in the car or in the field. The board and the measurement instruments were all connected to the laptop.

 

[coldsolder215]

All I know is they made the Optane design manager a VP, 'nuff said.

 

[hist78]

I lobbied for that before Intel hired BK. That would have been epic. Jensen Huang running Intel? This was in May of 2003. Intel would be a trillion dollar company!

If we look back, Intel has missed many incredible opportunities over the years. I would argue that this is the inevitable result of Intel’s DNA, the IDM business model. These mistakes cannot be explained by the 20/20 hindsight alone.

1. TSMC Founding Investment – Gordon Moore and Craig Barrett decided not to invest in TSMC at its founding. Instead, Philips stepped in, investing $58 million for a 27.5% stake at TSMC in 1987. As of the close on August 18, 2025, TSMC is valued at $1.219 trillion.
2. 
   
3. Apple iPhone SoCs – Paul Otellini declined Apple’s request for Intel to produce SoCs for the iPhone. Samsung, and later TSMC, took the deal. By 2024, TSMC’s revenue reached $90 billion, with Apple contributing 22% to 25% (up to $22.5 billion). In contrast, Intel’s 2024 revenue had shrunk to $53.1 billion, posting a loss.

Sold Intel’s ARM Division – Just before the debut of iPhones, Intel sold its XScale (StrongARM) ARM-based mobile processor division to Marvell in 2006 for approximately $600 million. Ironically, iPhones and nearly all smartphones today use ARM-based processors. Later, Intel tried to re-enter the smartphone market by acquiring Infineon’s wireless solutions business, including its cellular modem division, in 2011 for $1.4 billion. However, that effort also failed. Apple eventually acquired Intel’s smartphone modem division in 2019 for $1 billion. Intel missed the whole mobile/smartphone revolution.

Late Adoption of EUV – Intel was late to adopt ASML’s EUV lithography tools. While TSMC and Samsung were early movers, Intel only began using EUV at its Intel 4 process in 2023, four years after TSMC. This was despite Intel being the largest investors in ASML alongside TSMC and Samsung.

1. OpenAI Investment – Around 2017–2018, Intel declined opportunities to invest in OpenAI for a 15%–30% stake. Today, OpenAI is valued at around $500 billion.

https://www.fastcompany.com/91169548/intel-passed-chance-own-openai-years-ago-fallen-behind

 

[blueone]

All I know is they made the Optane design manager a VP, 'nuff said.

And the technical leader a senior fellow.

 

[tooLongInEDA]

I worked for Data General right out of college. We had an old DG Nova system at school so I was ready for DG. Plus I loved the book Soul of a New Machine by Tracy Kidder. It was on the reading list for a CPU architecture class. Great book back then.

Unfortunately the Micro Vax really killed the minicomputer business then came Sun Micro Systems and that did it for DEC and DG. DG tried the PC market with a desktop and the DG1 laptop. I think it was one of the first laptops. The LCD screen was not great so it did not take off. We called it a luggable because it was heavy. DG was in no shape to compete with the likes of Compaq anyway. EMC bought DG for the storage business in 1999 and DG was officially dead shortly after.

"Soul of a New Machine" will always be a great book. Really captures the excitement of what a tech startup can be and what technical leadership should be like. It will never date. Thanks for reminding me - just bought a copy for my son who's just starting out in computer science. Of course, the young don't seem to read books any more ... but we must try.

 

[Xebec]

They removed the graphics parts and it continued to be developed as the Xeon Phi family for HPC i.e. the 'Knights' chips. But it was a niche marginal market which did not pay for itself so it got eventually canceled.

I remember a few senior architects at Lockheed Martin were 100% convinced that Xeon Phi would take over ALL server duties in the future, lol. But, I've had quite a lot of bad predictions myself, so...

 

[Xebec]

And the technical leader a senior fellow.

I'm out of the joke loop here -- Optane seems to be good tech, but marketed poorly and poorly understood, so it never found it's market niche. But the technology seems pretty novel?

 

[Xebec]

1. TSMC Founding Investment – Gordon Moore and Craig Barrett decided not to invest in TSMC at its founding. Instead, Philips stepped in, investing $58 million for a 27.5% stake at TSMC in 1987. As of the close on August 18, 2025, TSMC is valued at $1.219 trillion.

Late Adoption of EUV – Intel was late to adopt ASML’s EUV lithography tools. While TSMC and Samsung were early movers, Intel only began using EUV at its Intel 4 process in 2023, four years after TSMC. This was despite Intel being the largest investors in ASML alongside TSMC and Samsung.

To be fair -- Re: TSMC - Intel was a much smaller company and made a huge loss in 1986, when the discussions likely happened. (Intel 1985 Profit = $2M, 1986 loss = $173M). They were still transitioning from being a memory business to CPUs, and trying to survive.

Re: EUV - that was a bit of a shocker, given how long and how often Intel had been talking about EUV. Was EUV never given serious consideration for Intel 10nm, even the earlier failed versions?

 

[blueone]

I remember a few senior architects at Lockheed Martin were 100% convinced that Xeon Phi would take over ALL server duties in the future, lol. But, I've had quite a lot of bad predictions myself, so...

Wow, I hope these guys weren't designing weapons systems...

 

[blueone]

I'm out of the joke loop here -- Optane seems to be good tech, but marketed poorly and poorly understood, so it never found it's market niche. But the technology seems pretty novel?

Over-sold by the team leadership, under performed compared to performance targets (latency was higher than projected), lack of projection of 3D NAND scaling capabilities and far lower cost, and very high R&D and production costs to scale Optane capacity by adding "decks".

 

[coldsolder215]

I'm out of the joke loop here -- Optane seems to be good tech, but marketed poorly and poorly understood, so it never found it's market niche. But the technology seems pretty novel?

Having worked on both the Intel/Micron and IBM variants of PCM, there was a common joke that PCM is good for precisely one thing: extracting R&D dollars from corporate coffers.

 

[Xebec]

Wow, I hope these guys weren't designing weapons systems...

One of them was an Aeronautics fellow...

 

[Michael Domnitei]

If we look back, Intel has missed many incredible opportunities over the years. I would argue that this is the inevitable result of Intel’s DNA, the IDM business model. These mistakes cannot be explained by the 20/20 hindsight alone.

1. TSMC Founding Investment – Gordon Moore and Craig Barrett decided not to invest in TSMC at its founding. Instead, Philips stepped in, investing $58 million for a 27.5% stake at TSMC in 1987. As of the close on August 18, 2025, TSMC is valued at $1.219 trillion.
2. 
   
3. Apple iPhone SoCs – Paul Otellini declined Apple’s request for Intel to produce SoCs for the iPhone. Samsung, and later TSMC, took the deal. By 2024, TSMC’s revenue reached $90 billion, with Apple contributing 22% to 25% (up to $22.5 billion). In contrast, Intel’s 2024 revenue had shrunk to $53.1 billion, posting a loss.

Sold Intel’s ARM Division – Just before the debut of iPhones, Intel sold its XScale (StrongARM) ARM-based mobile processor division to Marvell in 2006 for approximately $600 million. Ironically, iPhones and nearly all smartphones today use ARM-based processors. Later, Intel tried to re-enter the smartphone market by acquiring Infineon’s wireless solutions business, including its cellular modem division, in 2011 for $1.4 billion. However, that effort also failed. Apple eventually acquired Intel’s smartphone modem division in 2019 for $1 billion. Intel missed the whole mobile/smartphone revolution.

Late Adoption of EUV – Intel was late to adopt ASML’s EUV lithography tools. While TSMC and Samsung were early movers, Intel only began using EUV at its Intel 4 process in 2023, four years after TSMC. This was despite Intel being the largest investors in ASML alongside TSMC and Samsung.

1. OpenAI Investment – Around 2017–2018, Intel declined opportunities to invest in OpenAI for a 15%–30% stake. Today, OpenAI is valued at around $500 billion.

https://www.fastcompany.com/91169548/intel-passed-chance-own-openai-years-ago-fallen-behind

In 2010 I have worked on an Intel secret project which was about a CPU with an ARM processor hidden inside about which the customers not supposed to have access or know about it. I asked my manager how is Intel going to get away with that secret ARM processor and he said that after the project is completed all the documentation will be destroyed and "that project never existed". The work was done in a lab with multiple video cameras and a camera outside above the entrance door. I made progress very fast on this project (a few weeks ahead of schedule) that they came up with a list of other projects for me and I ended moving out of that claustrophobic environment (I had to sign for a every component or cable I need to do the validation & compliance tests, besides being watched by cameras).

 

[Markwrob]

"Soul of a New Machine" will always be a great book. Really captures the excitement of what a tech startup can be and what technical leadership should be like. It will never date. Thanks for reminding me - just bought a copy for my son who's just starting out in computer science. Of course, the young don't seem to read books any more ... but we must try.

They are good for setting laptops or displays on to get the viewing angle right. And they are useful when the internet is out of reach.

 

[tomatoma]

If we look back, Intel has missed many incredible opportunities over the years. I would argue that this is the inevitable result of Intel’s DNA, the IDM business model. These mistakes cannot be explained by the 20/20 hindsight alone.

1. TSMC Founding Investment – Gordon Moore and Craig Barrett decided not to invest in TSMC at its founding. Instead, Philips stepped in, investing $58 million for a 27.5% stake at TSMC in 1987. As of the close on August 18, 2025, TSMC is valued at $1.219 trillion.
2. 
   
3. Apple iPhone SoCs – Paul Otellini declined Apple’s request for Intel to produce SoCs for the iPhone. Samsung, and later TSMC, took the deal. By 2024, TSMC’s revenue reached $90 billion, with Apple contributing 22% to 25% (up to $22.5 billion). In contrast, Intel’s 2024 revenue had shrunk to $53.1 billion, posting a loss.

Sold Intel’s ARM Division – Just before the debut of iPhones, Intel sold its XScale (StrongARM) ARM-based mobile processor division to Marvell in 2006 for approximately $600 million. Ironically, iPhones and nearly all smartphones today use ARM-based processors. Later, Intel tried to re-enter the smartphone market by acquiring Infineon’s wireless solutions business, including its cellular modem division, in 2011 for $1.4 billion. However, that effort also failed. Apple eventually acquired Intel’s smartphone modem division in 2019 for $1 billion. Intel missed the whole mobile/smartphone revolution.

Late Adoption of EUV – Intel was late to adopt ASML’s EUV lithography tools. While TSMC and Samsung were early movers, Intel only began using EUV at its Intel 4 process in 2023, four years after TSMC. This was despite Intel being the largest investors in ASML alongside TSMC and Samsung.

1. OpenAI Investment – Around 2017–2018, Intel declined opportunities to invest in OpenAI for a 15%–30% stake. Today, OpenAI is valued at around $500 billion.

https://www.fastcompany.com/91169548/intel-passed-chance-own-openai-years-ago-fallen-behind

It's kind of shallow to link with anything that it's because of IDM.
This is not what IDM has caused
It's just the wrong choice...

 

[Paul2]

The stock market can be irrational for longer than you can be solvent. In this case we see zero usage of all those "AI" chips by the consumer, yet they attribute all their misfortunes to it.

I met a team of financial reporters interviewing me during CIVID chip crisis, who thought that Intel's chips are actually manufactured in PRC, rather than just assembled.

 

[tomatoma]

The stock market can be irrational for longer than you can be solvent. In this case we see zero usage of all those "AI" chips by the consumer, yet they attribute all their misfortunes to it.

I met a team of financial reporters interviewing me during CIVID chip crisis, who thought that Intel's chips are actually manufactured in PRC, rather than just assembled.

It's a terrible cognitive ability...
I bought the i7 12700K in the middle of the corona misfortune... The i7 assembly was labeled as China
The comprehension of analysts and financial companies is the worst.
They don't understand anything at all, but they loudly proclaim their foolish ideas.

 

[Xebec]

Over-sold by the team leadership, under performed compared to performance targets (latency was higher than projected), lack of projection of 3D NAND scaling capabilities and far lower cost, and very high R&D and production costs to scale Optane capacity by adding "decks".

Do you know - was the latency problem at the silicon level was it the interfaces and ("software") protocols? I always wondered if it was just held back by the interface..

 

[Paul2]

It's a terrible cognitive ability...
I bought the i7 12700K in the middle of the corona misfortune... The i7 assembly was labeled as China
The comprehension of analysts and financial companies is the worst.
They don't understand anything at all, but they loudly proclaim their foolish ideas.

And the "market can be irrational for longer than you can be solvent" applies both ways. Very well doing companies outside of the hype circle might actually getting bad coverage, and correspondingly bad stock market life. The wider public is only getting input from the media, rather than the research on the ground.

Now the news that nanasheet FET has terrible thermals has surfaced in mass media, and Samsung is in hot water again, despite previously being lauded for getting RibbonFET early.