Intel: Our balance sheet is a smoking ruin, but we think our new chips work
- Reference: 1723015510
- News link: https://www.theregister.co.uk/2024/08/07/intel_boots_18a_chips/
- Source link:
Chipzilla on Tuesday [1]announced what it described as "major milestones" for its foundry, which it now treats as a separate biz that supplies chipmaking services to both Intel and third-party customers.
The 18A process is Intel Foundry's attempt to better chipmaking tech employed by rivals like TSMC, and by doing so regain its status as Earth's most sophisticated chipmaker – thanks to its use of RibbonFET gate-all-around transistors and PowerVia backside power technology. The former is a new way of laying out chips that allows increased transistor density, and the latter is a means of delivering power to those denser collections of transistors.
[2]
Intel's plan is that tech like RibbonFET and PowerVia will return it to technical leadership that will see its own products improve, followed hopefully by sales. Fabless chipmakers will see 18A, swoon, and come-a-knocking to have Intel create their latest designs.
[3]
[4]
Order will be restored to the chipmaking universe – which means Intel on top and delivering rivers of profit on which shareholders will sail into early retirement as Intel scrip soars.
Intel badly needs that story to come true. The chip shop recently announced [5]further losses and a fresh round of mass layoffs. It's also dealing with [6]faulty products that look set to cost it plenty because lawyers are contemplating class actions.
[7]
Tuesday's announcement that Panther Lake client processors and Clearwater Forest server silicon made on 18A "are out of the fab and have powered-on and booted operating systems" is therefore welcome. So is news that this outcome was achieved "less than two quarters after tape-out" and that both products are "on track to start production in 2025."
But that's all the detail Intel provided.
We don't have detailed info about performance or yields – just an [8]assertion from head of foundry services Kevin O'Buckley that Panther Lake and Clearwater Forest "are being used inside the company, and are yielding and performing well."
[9]Intel to shed at least 15% of staff, will outsource more to TSMC, slash $10B in costs
[10]What AI bubble? Groq rakes in $640M to grow inference cloud
[11]UK court rules in Intel's favor in R2 Semi power patent case
[12]No love lost between Apple and Nvidia as iGiant chooses Google chips for AI training
More info would be appreciated so investors and buyers can understand if yields were strong. Did most of the dies Intel baked produce usable chips, or did just a few yield working products? That question matters – because if 18A has low yields, the process will struggle to make a dent in the market,
The announcement also omits mention of when in 2025 production might commence, or the potential volumes. That matters because Intel's rivals are not idle: Qualcomm in particular is keen to [13]move into desktop and workstation processors , after already seeing its Snapdragon Elite X silicon used as [14]Microsoft's poster child for Copilot+ PCs. AMD also has [15]new client silicon in the works.
[16]
Intel did at least offer a hint of third-party enthusiasm for 18A, revealing that a "first external customer is expected to tape out on Intel 18A in the first half of next year."
That's a little vindication of the plan to return to the foundry biz.
As The Register considered Intel's announcement, we couldn't help but notice it includes an unusually lengthy set of disclaimers about the forward-looking statements it contains. Out of curiosity, we ran a word count: Chipzilla produced 738 words of news, and 925 of legalese.
Which is very telling. This announcement was a tease, not a promise – and it remains to be seen if 18A is really on track to drive the turnaround Intel imagines. ®
Get our [17]Tech Resources
[1] https://www.intel.com/content/www/us/en/newsroom/news/intel-foundry-achieves-major-milestones.html#gs.d9livo
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZrNFxre24v-rEphUv-8-wQAAAMQ&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZrNFxre24v-rEphUv-8-wQAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZrNFxre24v-rEphUv-8-wQAAAMQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://www.theregister.com/2024/08/01/intel_to_ax_headcount/
[6] https://www.theregister.com/2024/08/05/another_law_firm_piles_on/
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZrNFxre24v-rEphUv-8-wQAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[8] https://www.intel.com/content/www/us/en/newsroom/news/kevin-obuckley-talks-progress-intel-18a.html#gs.d9qf4n
[9] https://www.theregister.com/2024/08/01/intel_to_ax_headcount/
[10] https://www.theregister.com/2024/08/05/groq_ai_funding/
[11] https://www.theregister.com/2024/08/01/uk_intel_patent/
[12] https://www.theregister.com/2024/07/30/apple_google_tpu_ai/
[13] https://www.theregister.com/2024/06/03/qualcomm_arm_pc_ambitions/
[14] https://www.theregister.com/2024/05/21/qualcomm_windows_microsoft/
[15] https://www.theregister.com/2024/06/03/amd_ryzen_300_ai/
[16] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZrNFxre24v-rEphUv-8-wQAAAMQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[17] https://whitepapers.theregister.com/
Re: Spin
They may not have them yet.
It's perfectly normal for the first samples of chips to have bugs that need to be fixed before you can proceed further. These chips are insanely complex.
"It boots an OS" is a step before "it runs reliably", and that's needed before you can run all the benchmarks. And after that, you look at optimizing things so the chip runs as fast as possible. Only then do you have performance numbers that will match the final chip.
Re: Spin
Fsck Wall Street. Telling investors what yields are before you even ramp up to near production levels is simply feeding their intensely greedy, selfish FOMO in hopes of jumping onto a bandwagon before its even finished.
Let them rot. When you have things sorted *then* make production announcements, before then tell them to suck an egg and wait until you get things right. You make an early announcement for yields, fails during roll-out...and the fscking greedy bastards sue you anyway.
Wall Street are not your friends. They are blood-sucking selfish bastards that deserve to rot on a pike as soon as possible.
The chip shop
They've certainly had a battering.
Let's hope the increased power density doesn't lead to "frying tonight",
"rivers of profit on which shareholders will sail into early retirement"
Shareholders can easily also have shares in AMD, thus playing the two big (only?) horses.
Not to mention that, said shareholders are that cash flush, they should have shares in TSMC as well, giving them revenue across the board.
Convenient timing for this announcement considering how most of the news stories regarding Intel of late have been negative. It sounds like PR spin as I think 18A production is going to end up slipping passed 2025 but who cares at that point if it makes the share price go up and the shareholders happy now.
New technology?
I think not, according to my other half I've been supplying "power via backside power technology" for years.
Spin
If the numbers for yield and performance were good, you can bet that they'd be releasing them.