Home Creators Posts Import Register Favorites Logout

Content

The next episode of Broken Silicon will see us be joined by Jon from Asianometry. Jon creates video essays on YouTube about silicon companies, history, and geopolitics. He is also based in Taiwan, and so he has a firsthand account of how business operates there. At a minimum, we plan to discuss:

AMD's rise since Zen 1

What we expect from Zen 5 and Later

Intel's current situation and future

TSMC building 2nm Fabs in the United States with US Funding

Nvidia Blackwell and their AI Momentum

The recent Earthquake in Taiwan

Current Geopolitics as they relate to Semiconductors

You have till Saturday night US Central time to submit below! Be respectful, use good grammar, and be as concise as possible to have your submission considered!Check out Jon's YouTube Channel: https://www.youtube.com/@AsianometryLast time Jon was on: https://youtu.be/p1m0pc5TkNw?si=kB-6J3Osh-lX3uwWhttps://youtu.be/tB4q6Wp4XHc?si=bPC9SMQYgrJYGKwbhttps://youtu.be/Nk-lBok9_TU?si=f8zXjdeRbhpq7_tGhttps://www.anandtech.com/show/21337/tsmc-to-receive-6b-us-chips-act-set-to-build-2nm-fab-in-arizonahttps://youtu.be/TbWfywgY7sk?si=4gzq6gIcRD-qodGE

Comments

Intel FanBoy

There has been rumblings that Intel has snagged a big whale for its IFS, do you happen to know who it is?

Benjamin Cannon

What do you think will be the future of AMD’s CPU advantaged packaging regarding the chiplet to chiplet interconnects?

Benjamin Cannon

If someone told you where AMD was in 2024 before Zen’s launch, what would have been your reaction? Thinking they are insane is a valid opinion

Dig Wiggler

Why does Amd not stomp Intel with Zen? If I was Amd, I would release the 24 core 9970x3d. 8 vcache zen 5 cores, 16 zen 5c cores. This would undisputedly be better than any i9 KS chip. Why not have a special flagship chip like Intel? Why not develop some product in the low in that kills the i3 12100/12400 Intel products? Idk, maybe a 16 core Zen 5c chip for $150 and a 12 core for $120. Why does Amd not take the killing blows? Amd is frustrating sometimes, like they want the possibility of being in second place

Rentaro Matsukata

Why does he insist on pronouncing DRAM the way he does? Is he just trolling? What does he think about Rapidus and the upcoming GAAFET battle?

Christian Visconti

Hi Tom, hi Jon, I have 3 question for both the host and the guest: It's being reported that the first HNAEUV machines (High Numerical Aperture Extreme Ultraviolet Lithography machines) made by the Dutch company ASML are being bought by Intel as the company plans to get back on track with node quality and production. 1) What are these new machines capable of compared to normal EUV? (bigger wafers/better nodes/ both / other) 2) Is TSMC buying these machines too? 3) combining Roadmaps, leak and experience on the subject, how many years from now can we expect an intel rebound?

High Yield

As a person who will visit Computex for the first time, what are some tips and what do I have to see in Taipei outside of Computex? Tech question: Intel will be first with BSPD & GAA. Do you think 20A/18A have a real chance of competing (and maybe even surpassing) TSMCs 3nm nodes?

Dave Scholze

I have heard some rumors about a potential AM5+ platforms. Do you have any information on that or do you think it's just not true? What would you like to see if it were to be true?

Bran Muffin

Hello! Do you think RTX 5090 will need PCIe 5.0 to reach its full potential or will 4.0 be enough?

Bran Muffin

What is your prediction for RTX 5090 power usage?

Gergely

China can build out leading node - 2 or 3 gens of fab tech with domestic or smuggled tech, then effectively blockade Taiwan (to avoid having to go to war, or at least fire the first bullet against the US) or at least make the fabs unable the produce anything (drones, missiles etc.) What is the West's plan to stay competitive on Fab tech without TSMC?

10GHz Netburst

How much of Nvidia's dominance in HPC GPGPU and datacentre AI do you think just comes down to "nobody ever got fired for buying Nvidia"?

10GHz Netburst

I feel like in the past decades companies such as Synopsis and Cadence who have expanded from pure EDA to providing standard IP blocks for other companies to use in their designs have gone somewhat underappreciated. How much do you think these companies will grow in importance now that they can move to providing ready to fab chiplets with standard interfaces for use in more complex SoCs? Would you think it's a stretch to say these companies, rather than the likes of Intel, Nvidia, AMD, or other "end product provider" will be the big winners of the chiplet era?

10GHz Netburst

Having lived in many countries with big natural hazards (I was in Chile during the 2010 earthquake they had there), I feel like a lot of the commentary on the earthquake in Taiwan really is just yet another showing of how woefully inadequate and backwards disaster management in the US is, be it earthquakes, flooding or weather...

10GHz Netburst

Now that backside power delivery nodes are on the horizon and edging towards , what do you think the next step will be for improving CMOS logic?

10GHz Netburst

Do you think it is correct to say Nvidia is not really using chiplets for their B100 as opposed to AMD who has a more "legit" chiplet design on RDNA3 and MI300?

10GHz Netburst

AMD is often seen as the plucky underdog relative to Intel and after ATI also relative to Nvidia, to what extent do you think they themselves are stuck in a mindset of being an underdog company?

10GHz Netburst

Why are we building semiconductor fabs, a notoriously water intensive industry, in Arizona, a notoriously arid state?

10GHz Netburst

How does backside power delivery clash or synergise with advanced packaging technologies? Will it for example affect the interconnect densities or performance?

TechnoLadz

What are you most excited for with future Zen iterations?

AI-Soldier789

What is the Opinion of you two about Intel Gaudi 3 against Backwell. Is it just in time to get market share and will it be manufactured by IFS or by TSMC.

TechnoLadz

Do you think we will see a new Motherboard platform that is designed for the Halo SKUs; so it can have larger memory buses and bigger dies?

kjm015

Hello, Tom and Jon! Will Intel's planned transition from using their own fabs to using TSMC's fabs for their products cause more delay in TSMC's order queue? Will there be knock-on effects for other TSMC customers, and should we expect product delays or price increases?

TechnoLadz

What if AMD's current Halo being used for Strix is just the starting point, and we see bigger Halos from here on out (EG: Zen 6 could be 16 Z6 cores + 72 RDNA 5 CUs, Zen 7 could be 24 Z7 cores + 96 RDNA 6 CUs)

TechnoLadz

If the 5080 is $1300 for 4090 performance and 16GB VRAM, do you think people will buy that; or do you think it'll end up being a 4080 sales like situation where basically no one buys it.

TechnoLadz

If the world is about to enter WW3 (Biggest war in Middle East in over 50 years, wars starting in Asia, etc), what will we do with tech components. Do you think this could give Intel an advantage since their Fabs are mainly American based.

Dark Side of the Force

Hi y'all. My only question this time is would Nvidia ever stop "cheaping out" on Vram on their product? It's just too funny to me that they are scared of a probably slower MI350X, because it has more Vram.

Chris Rijk

If Nvidia tries to charge $1200 for the 5080, do you think there's a realistic chance it could sell well? Do you think they'll try to charge $900 again for the card just below it (the unlaunched "4080 12GB")? I find it difficult to imagine gamers accepting that with Blackwell when they clearly rejected it with Lovelace.

Donkey Kong Odyssey

With talks about Nvidia shortening Blackwell and Rubin launch times by using cheaper nodes, I think many people have lost the plot: AMD is challenging Nvidia. When you look at product launches between both companies over the years, you'll notice a particular trend: Nvidia will try to punch down AMD's momentum, or throw cold water at it, whenever AMD directly challenges Nvidia in any way. We saw this with the admittedly overhyped Vega vs 1080Ti (launched preemptively after Vega marketing hype), RTX 20 Super, the RTX 30 Ti launches (Save the 3060Ti), and even the recent 40 Super launches. We know Nvidia won't sit still, but I have to ask: did Radeon Instinct finally reach a level of performance, or price-to-performance, that put some kind of pressure on Nvidia? We know Nvidia doesn't make these kinds of moves when AMD's far enough behind, but I'd like to hear your perspective on the matter

TMCPayton

Howdy, Other than manufacturing silicon products and fasteners, what are Taiwan's biggest industries? Thanks