Wednesday, July 2nd 2025

Intel Abandons In‑House Glass Substrate R&D, Leans on External Suppliers
Under its new CEO, Lip-Bu Tan, Intel is stepping back from building its glass substrate technology in favor of sourcing ready-made solutions from specialized vendors. Driven by Tan's strategy to concentrate resources on Intel's primary lines, CPU/GPU development, and foundry, Intel's operations are becoming leaner. By outsourcing glass substrates, Intel can significantly reduce development timelines and mitigate the financial risk associated with pioneering novel substrate manufacturing. Crucially, this model grants Intel the freedom to evaluate multiple suppliers, pivot quickly if performance or cost benchmarks change, and integrate advanced packaging more rapidly than if it continued to develop substrates internally. Just as the company is focusing on 14A and 18A-P(T), abandoning in-house glass substrate development is a step in the right direction to reduce costs and achieve profitability.
Meanwhile, the glass‑substrate sector is gathering momentum, particularly among South Korean firms. SK Hynix, in partnership with Applied Materials, has been trialing its Absolics pilot line for some time, while Samsung has recently scaled up its efforts in this niche. JNTC celebrated the opening of its first glass‑substrate plant in May, announcing sixteen prospective clients and projecting revenues to climb from $14.7 million this year to $147 million in 2026, and ultimately hitting $735 million by 2028. A complementary facility under construction in Vietnam will triple the original plant's output, targeting a combined annual production of roughly 500,000 substrates. Intel hasn't yet named which suppliers it will partner with. Still, the company plans to tap into this growing network, potentially joining JNTC's customer roster, to support its advanced packaging roadmap and accelerate product rollouts.
Sources:
Chips and Wafers, via ComputerBase
Meanwhile, the glass‑substrate sector is gathering momentum, particularly among South Korean firms. SK Hynix, in partnership with Applied Materials, has been trialing its Absolics pilot line for some time, while Samsung has recently scaled up its efforts in this niche. JNTC celebrated the opening of its first glass‑substrate plant in May, announcing sixteen prospective clients and projecting revenues to climb from $14.7 million this year to $147 million in 2026, and ultimately hitting $735 million by 2028. A complementary facility under construction in Vietnam will triple the original plant's output, targeting a combined annual production of roughly 500,000 substrates. Intel hasn't yet named which suppliers it will partner with. Still, the company plans to tap into this growing network, potentially joining JNTC's customer roster, to support its advanced packaging roadmap and accelerate product rollouts.
26 Comments on Intel Abandons In‑House Glass Substrate R&D, Leans on External Suppliers
Edit: The AMD and Intel test platform notes are hilarious!
Fixed that for ya.
Arm still got a very tough road ahead of them before we see them in profesionals/heavy duty computers. Thin and light for people who don't much beside browinsg and office stuff sure, but the other segement? AMD will just become the de facto replacement for intel without having to do anything beside existing. It's going to be a very unbalanced market for a long time if Intel kick the bucket without ARM windows becoming at least as good the Mac lineup both in formfactors, and software compatibility.
Their chips are more expensive to produce, inferior in terms of Perf/Watt and their only distinct benefit is slightly superior IPC/Single threaded performance which matter in certain areas but again is niche in comparison to everywhere else that is benefiting from AMDs superior MT advantages.
Some of you are STILL not getting it.
Nvidia is a monopoly. TSMC is a monopoly.
Obviously TSMC being a monopoly is not good. Unfortunately Samsung keeps rising the bar trying to beat TSMC not create a long lasting profitable node. They jump from one node to the next. Intel unfortunately under it's new CEO seems to have as a goal to try to do the same. Instead of focusing on 18A to start building competitive CPUs and GPUs and gaining experience, they try to rush to the next node to beat TSMC, hoping to secure Apple and Nvidia as customers. Because trillion-dollar companies like Nvidia and Apple will happily take the risk on investing in Samsung or Intel manufacturing, because both Samsung and Intel have perfect manufacturing records. This CEO that the morons in Intel board chose, is playing Intel's future the same way someone would be playing someone else's house in the roulete. He probably thinks "If I secure Apple or/and Nvidia I will gain the CEO title of the year, maybe even for the decade. If I don't and Intel goes bankrupt, who cares? I'll find work elsewhere."
Tan and Intel's board of directors are right now the biggest threat to Intel's future, in my opinion. Indeed. A 500MHz Athlon? LOL. That's 1999, while Itanium 2 was 2002. AMD had 3-4 times faster CPUs than that in 2002. Not to mention the rest of the specs. The slowest IDE driver vs faster and bigger SCSI drives (in RAID I suppose). Is this real or a hoax?
I love that "Don't do at home" in the end.
I am also not surprised to see "Tom's Hardware" and "contract" in the same sentence.
It's looks like that massive HPC or phone revenue is pretty much mandatory to stay in the mainstream chip market wich sounds kind of scary. Even if Intel manage to make a good client CPU in the next few years it won't prevent them from having to leave the market if they can't get more money from HPC/datacenter. And look at zen 5, meager gains for the client, but great arch for the datacenter. So still a money maker either way.
It's insane to think that the company who still got over half of the marketshare in client CPU might not be able to stay in that market. Their client revenu alone was bigger than the total revenue of AMD the last that time I checked. Having the biggest client CPU revenu in the market isn't enough to stay in that market. Imagine still massivelly outselling your competitors, but having the futur looking dire because you don't do as well in other sectors. The client segment doesn't super healthy from a conssumer pov, since it doesn't seem to be a sector sustainable as a standalone.
Its turning into the digital camera market: none of the players who survived to this day in that segment have that market as their sole breadwinner. Its their other activities that pays for it. Wich explain how brands like Nikon managed to stay afloat even though they've been so complacent, people really thought that they gave up on competing in the camera sector at some point.
Its going to take a hell of a rework from the top down to get back to being competative IMO.
I wonder if Tan's scrapping of Intel marketing includes their IDF marketing slush fund. It's probably waht's keeping them alive with the OEM's.
Since what's a more profound and efficient way to burn through mountains of cash for naught, than to sell at cost or even at a loss, only to try at least maintaining their market-share? Yet it's still not working and their market-share is still steadily declining – Their Batlle Royale tilting at Windmills.
The funny thing is, Intel has done this from day 1!
As soon as anything Ryzen was launched, Intel immediately started to hand out huge rebates to fight AMD, the same sneaky and backhanded way Intel has always done and never really stopped. Intel even bragged about it before OEMs and ODMs at some event in Far East, that they could count on being seriously and richly compensated for sticking with Intel, as always.
Remember that slide from back then? Informed ones knew it from the get-go, that Intel was bribing outlets and OEMs to explicitly not use anything AMD.
TechSpot.com – Intel is using heavy discounts on Xeon CPUs to stop AMD from eating its server lunch (September 2021)
ServeTheHome.com – Intel is Serving Major Xeon Discounts to Combat AMD EPYC (September 2018)
They even paid OEMs massive slash-sums, to make them not release any AMD-boards for socket AM4, or at least stall adoption by intentionally delaying shipment for months. The whole reason forwhy there were basically slim to none AM4-boards for months (and those, which were in fact available, were of utterly inferior quality while being overtly expensive), is Intel and it always has been ever since … Who wants a product, which is at least as expensive as competitor-products, yet doesn't offer the same performance, while coming with shockingly worse efficiency?
And we're not even talking about the fact, that ARL was often slower than its direct predecessor. It wasn't even as fast as the last refresh. It was a regressing refresh. It was a Regresh!
Also, Arrow Lake came again with higher adoption costs for another new board which was already dead-end by the time it came to market – Let's f–ck over consumers even harder! I hope so and since years actually just still wait for the day, that Intel might actually acknowledge in a flash of sudden genius, that the days of their bribing are necessarily over, since they're just burning through mountains of money that way … and killing themselves financially.
Yet I also know, that the moment Intel ever stops their actual bribing for real for once, their market-share would be in free-fall from that point forward, and basically implode within months.
Especially in their current competitive situation (or rather the very lack thereof). So most definitely not happening …
I bet, Intel would rather go bankrupt than to stop bribing for pushing even their last shipped single CPU out the door, to be taken before anything AMD due to actual natural competition …
AMD also responded in kind to those rebate:
www.tomshardware.com/news/amd-starts-aggressive-partner-program
www.crn.com/news/components-peripherals/amd-brings-aggressive-cpu-rebates-to-vars-with-new-partner-program-exclusive I find it amusing that AMD is still perceived as a lost, naive puppy, despite having a strong portfolio and being very aggressive in business strategy and building customer loyalty. And in the end, the customers win. They get Epyc CPUs for cheaper because both companies decided to enter a price war (and one of them did it while still making a benefit, while the other is still bleeding money )