Speaker unverified: belief

Transcript context

…exity of the chip, etc., etc. the actual design of of the the architecture, the the RTL generation if you will, the mapping of the architecture is not the largest amount of time. The largest amount of time is in the verification, the validation, the debug, the documentation, etc., etc. AI is really good at that. Uh, and I would say we probably have 80 to 90% of engineers today in Sidearm who use it on a daily basis. And if we were to shut it off, uh, my analogy I give to people, it's like being in the 1990s, you've got internet and you're now saying, you know, only internet between the hours of two and four. >> After that, go to the library that we have down the hall that's got all the books that you can go up and look all this information up. People, there'd be anarchy. So, the genie's out of the bottle, right? There's and and there's no there's no stopping that. Now, there's certain things that the tools are still not that mature of. uh one of them is really around RTL generation and then physical design and implementation in best of class and that's simply because the models you know they're they're trained on on what's available publicly uh and a lot of that information is quite proprietary. That being said there's massive opportunity between the ecosystems and everyone in the industry to make that better. It's only going to get better. >> Have you been fine-tuning models to try and address that gap given the proprietary information that you've offer? >> We've been working with model makers uh around that. Absolutely. Uh and I think that's a big big opportunity and one of the things I'm proud of at ARM is given our business our core IP business we probably have the richest IP portfolio both in terms of not only the IP and this is the killer the documentation the test benches you know how how you build the IP you know I've worked for chip companies in the past that have said hey why don't we license this IP that we've got because it's really really valuable and then you get into wait a minute there's there's no documentation [laughter] there's no explanation No one's ever going to be able to use this. >> It's unusable and it's untestable, right? And and if it's unusable and untestable, it's actually untrainable. And if it's untrainable, it's not usable for AI. I think we have some some built-in advantages based on our business model that'll allow us to really be uh able to take advantage of the tools as they get better. >> Really exciting. How much do you think um if you were to extrapolate out this is a little bit of a uncertain question but if you extrapolate out 2 3 years and all the tooling is likely to come in AI and the ability to fine-tune models against you know some aspects of the design that you mentioned do you think that 24 to 36 month cycle shrinks to a year to 6 months do you think it stays roughly where it's at a little bit curious about how does that really impact these cycles and time to market because that has pretty dramatic ramifications >> in terms of the clock speed of the entire industry >> I don't know I don't know if it's uh in two to three years away, but five plus years, >> can you go from idea to a GDS2 file? tic ramifications >> in terms of the clock speed of the entire industry >> I don't know I don't know if it's uh in two to three years away, but five plus years, >> can you go from idea to a GDS2 file? GDS2 file being the the file that you actually send to the fab to to go get built for certain designs. Quite possible. So, it takes that whole design piece out of the way. It takes that whole piece out of the way relative to the verification. So, I think for the more straightforward designs, quite possible. Now if you go into the tool and say design me something that's 10% faster than Ver Rubin 20% cheaper and 30% more efficient on this model you're not going to be able to get press a button and have it happen right away. Uh but I think in 5 to 10 years you know our industry as well uh we're going to see some amazing differences relative to how chips are designed. How does it change um I'm sure you had some prediction of this but how does it change the way you look at the business given uh there's just a a big diversity of large players and new players that all you know want to have their own chip designs now and you know the the Vera and the gravitrons of the world they all use ARM um it's a it's a big step up for them but it's a it's a big diversification of the customer base right that can be only good >> oh absolutely uh I think what's going to matter back to the earlier discussion we had on supply chain. It's understanding the supply chain impacts uh how all of that gets built and and put into ultimate end products. Uh I think that's going to become a much more important muscle uh as we as we go forward because it's one thing to en said another way there's a lot of really great young companies today doing AI chips well-known companies getting tons of funding innovative designs etc etc selling into an industry where the capital requirements are just massive and the relationships with memory vendors is incredibly critical or the relationship with substrate vendors so companies are going to have to be much >> more access to a 3nometer line, a 16 line, advanced packaging line, all of it. Yeah. All of that. And I And I think that is um that's not going to stop in 12 months, it's not going to stop in 24 months. I think we're going to be in this constrained environment for 3 to 5 years at least. Uh so long as the transformer is the unit of energy relative to how you generate AI training and AI inference, by design, it is a it is it's very comput intensive. It's very memory intensive. So if you think about that, that's going to drive a lot of demand on having supply chain acumen, which then goes back to people have got great ideas on ship design. They're going to have to have need a lot of other things just to be able to get access to capital, wafers, everything you just talked about. We've just had a cascading series of things that have been the bottleneck to more compute for the AI industry. though you know 2 years ago or so I think it was like packaging and packaging related items and then eventually now people talk about how it's memory and things like that that are in some sense limiting to certain systems being built at sufficient scale. packaging and packaging related items and then eventually now people talk about how it's memory and things like that that are in some sense limiting to certain systems being built at sufficient scale. Uh do you have a view of what is the next sort of bottleneck that's coming? I think building out the data centers is is going to be a bottleneck. And and when I say building out, um if you look at all the projects that are being done today, um not a lot of them are ahead of schedule uh and needing less labor than they thought, right? And then when you when you layer on top of that a lot of buzz that's coming from different parts of the of the of the country in the United States here relative to slowing down data center development or putting restrictions around it. I think that uh infrastructure buildout uh could be a headwind just relative to to everything going on which may be you know end quote okay because if it was if if infrastructure buildout was not a headwind uh I think capacity for wafers capacity for memory that probably would be a would be a headwind so I think you're going to see a number of different governors if you will not governors of states but different things that are going to throttle the growth of this uh which just expounding for a second I've been on a bunch of panels and I get a lot a lot of questions about AI bubble and when's it going to stop and there's setting aside the valuation bubbles which is a a stock market index component the the bubble in terms of are we over um over supply to demand not even close and I think again that's because the demand is insatial just given the way these models work >> and infrastructure buildout access to wafers access to memory all of that's combining >> uh you mentioned that Um and I think a lot of companies are learning today that um strategic use of the cap table, access to capital in an era where you either need to consume a lot of compute or you need to put a lot of capex into the ground or you're just doing big technical projects like coming up with um uh CPU IP. You run SoftBank Group International. You have this one dominant shareholder. ARM itself as a business is just like a beautiful cash flow machine from the outside, right? I I'm sure you think a lot about like the the leverage of SoftBank and how to use that well like what advice do you have for entrepreneurs navigating these capex intensive industries from where you sit? >> Yeah. So, one of one of the benefits we have uh at ARM uh publicly traded, yes, uh but a very very large uh single shareholder. So I have lots of informal investor meetings with my chief shareholder you know all the time about this. Uh we have a big advantage in that uh there's a lot of things symbiotically we can do together that can help ARM advance its initiatives by having Soft Bank as our largest shareholder that we look to be very very um innovative around. uh to your point in terms of you know young companies I would say strategic partnerships incredibly early uh whether it's with people inside the supply chain uh people in private equity you the banks the banks themselves it's a different game you know now on one…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.