Intel may scrap 18A node in favor of 14A to attract Apple and Nvidia

[DragonSlayer101]

TL;DR: Intel CEO Lip-Bu Tan is reportedly considering scrapping the company's 18A manufacturing process for external foundry customers. Instead, Tan wants Team Blue to focus on the newer and more efficient 14A node in hopes of securing major clients like Apple and Nvidia away from TSMC.

According to sources quoted by Reuters, Intel could face a major write-off – potentially amounting to billions of dollars – if it decides to scrap 18A. The 1.8-nanometer process was heavily promoted by Tan's predecessor, Pat Gelsinger, as the technology that would finally allow Intel to catch up to TSMC.

18A was a flagship project for Gelsinger, who invested billions in its development. However, since taking over as Intel's CEO earlier this year, Lip-Bu Tan is believed to have expressed dissatisfaction with both 18A and its variant, 18A-P. In internal meetings, he reportedly noted that ongoing delays have made the nodes less appealing to potential customers.

Intel dismissed the report as "hypothetical scenarios or market speculation," adding that it remains the largest customer for its own 18A process. The company noted that its Panther Lake processors, which it described as "the most advanced processors ever designed and manufactured in the US," will be the first to use the 18A node and are on track to roll out later in 2025.

Despite the denial, Intel confirmed it is making several major strategic shifts to return to profitability after reporting a staggering $18.8 billion loss in 2024. According to the company, it has "identified clear areas of focus" and is working with customers to strengthen its future product roadmap, build trust, and improve its financial standing.

Reuters notes that Intel already has agreements with Amazon and Microsoft to manufacture a limited number of processors using the 18A node. The proposed strategy shift reportedly won't affect those deals, as neither company is willing to wait for 14A deployment. Intel confirmed that it plans to "deliver on its customer commitments" using existing technology.

Following Gelsinger's retirement in December 2024, rumors began circulating that Intel might spin off its foundry business to focus on its core x86 chip design unit. Tan has neither confirmed nor denied these reports since taking over, but much will likely depend on whether Intel can convince high-profile customers like Apple and Nvidia to abandon TSMC in favor of its 14A platform.

Permalink to story:

Intel may scrap 18A node in favor of 14A to attract Apple and Nvidia

 

[HardReset]

Intel strike trash with 10nm process that was supposed to leave all behind. Now 8 years later Intel is still promising gold and delivering trash.

This is going better than I expected

 

[SRB]

Remember, Tan is paid by stock price, not performance. Why worry about making anything when you can hike the stock price by promising the next big thing. They've already farmed a good deal of their production to TSMC. Why spend all of that money tooling up the existing fabs when one or two of the test fabs can produce all you need for marketing purposes? Not to mention the labor costs you save by not making anything. Tan has already announced that they're axing the manufacturing and chip development people along with the rest of the company,. After all, you have to do more with less these days.

The board hired Tan to put the corpse on life support long enough to pump and dump the stock. Then sell off the parts for whatever they can get.

Private equity without the middleman!

 

[Kashim]

Ah yes, Intel will scrap 18A in favor of 14A, just like they scrapped 2nm in favor of 18A, and 4nm in favor of 2nm, etc. Who wants to bet they'll "scrap" 14A in favor of 10A by the end of the year? Then they'll scrap that in favor of whatever made up node they com up with next. Just keep promising better and better nodes without actually delivering anything. Smooth move.

 

[AnilD]

If Intel wants Apple and Nvidia, they better be ready to offer more than just node names that sound like WiFi standards. At this point, 18A might as well stand for "18 Apologies."

 

[redgarl]

The only way for Intel to survive is to get rid of their fabs.

If they don't, they will ultimately not be able to compete in any fields.

In the past, Intel was using their fabs to flood the market with cheap silicon to maintain a monopoly, now they can't do that anymore if they need to manufacture their CPUs at TSMC.

Once Intel lose the mobile CPU market to AMD, they will pass the point of no return where their business model is no longer viable. AMD will be the major player to gain out of this. TSMC also...

 

[J Oelschl]

Wait a minute. First Intel scrapped 20A to focus on 18A. They keep telling is that 18A has been testing very well and is set to go full production in 2H this year. Was that all a big lie?

 

[BadThad]

They should carve off the fab business and stick to design as the priority. I hope they're working on quantum gate technology very hard. We are about to have a major revolution in CPU design with the elimination of transistors and the transition to entanglement on a chip. That would render this node race far less critical to increasing performance.

 

[wiyosaya]

Promises, promises. The current US Admin canned the chips act, of which Intel was planning on grabbing a piece of the pi-e, even though said Admin claimed it wanted to bring manufacturing back to the US.

Said admin and Intel: Promises, Promises. Where's the delivery of the promises?

 

[wiyosaya]

The only way for Intel to survive is to get rid of their fabs.

If they don't, they will ultimately not be able to compete in any fields.

In the past, Intel was using their fabs to flood the market with cheap silicon to maintain a monopoly, now they can't do that anymore if they need to manufacture their CPUs at TSMC.

Once Intel lose the mobile CPU market to AMD, they will pass the point of no return where their business model is no longer viable. AMD will be the major player to gain out of this. TSMC also...

IMO, Intel lost its way a long, long, long time ago. Long before they claimed AMD was in their rear-view mirror. They've been resorting to BS for so long now, they have no clue how to run a legitimate business. Maybe their epitaph will read "Once a leading Technology company, their own Hubris killed them."

I once worked for a company that succumbed to its own hubris. It was very painful for over 70,000 employees.

 

[Kashim]

Wait a minute. First Intel scrapped 20A to focus on 18A. They keep telling is that 18A has been testing very well and is set to go full production in 2H this year. Was that all a big lie?

Yes.

 

[wiyosaya]

If Intel wants Apple and Nvidia, they better be ready to offer more than just node names that sound like WiFi standards. At this point, 18A might as well stand for "18 Apologies."

It's Intel's Wet Dream. I doubt they will be able to make it come true. It would be too much of a change for them.

 

[SRB]

Promises, promises. The current US Admin canned the chips act, of which Intel was planning on grabbing a piece of the pi-e, even though said Admin claimed it wanted to bring manufacturing back to the US.

Said admin and Intel: Promises, Promises. Where's the delivery of the promises?

You mean they should have helped Intel, after they canned a German fab? After Ohio became 203? for opening? After shelving node after node, years mired in 14++++++++++? Making all of their latest chips at TSMC? Firing the guy hired to get the fabs profitable? How much did Intel already get from CHIPs before they blew all of this up?

Even if the CHIPS act survived, I wouldn't give Intel a dime until they actually make something better than warmed over renamed 10nm from days gone by.

 

[Tantor]

It's mind boggling that Intel would scrap a node that cost them so much. But there's another possibility. A couple of weeks ago Iran hit Israel's Kiryat Gat microprocessor manufacturing complex with a hypersonic missile.

Intel has a $25 billion plant there. I would not be surprised if they got hit. Intel is expected to lay off hundreds of employees at the Kiryat Gat plant.

https://www.calcalistech.com/ctechnews/article/bjuzok00mxe

 

[Watzupken]

Wait a minute. First Intel scrapped 20A to focus on 18A. They keep telling is that 18A has been testing very well and is set to go full production in 2H this year. Was that all a big lie?

The writing was on the wall when 20A was reported to have very poor yield, and this was something that Broadcom mentioned if I did not recall wrongly. Doubling down a refined 20A, also called 18A, I am not sure if it materially improves the situation. The node may be better, but poor yield = losses due to no customer. Until Intel finally say they are scrapping the foundry business, I doubt there will be any turnaround. But in the process, they will take a substantial loss which I don’t think they afford now.

 

[Jack77]

They said they think about not offering the process to "external customers" anymore. That does not include the internal customer - Intel CPU division - that already has a lot of designs made or in preparation for that process.

 

[erickmendes]

Intel has a $25 billion plant there. I would not be surprised if they got hit. Intel is expected to lay off hundreds of employees at the Kiryat Gat plant.

I wouldn't be surprised if Intel itself paid to get hit by a missile lol... So they can ask for more gov funding...

Feels like Intel can't shrink bellow 10nm... Meanwhile even Huawei that's having a hard time getting it's hands on the needed tools is already at 6nm.