I really doubt that the EDA software is really that much of an investment these days. It's the lithography tech that's the real crown jewel, and it is being protected.
> I really doubt that the EDA software is really that much of an investment these days
If EDA tools were so simple, why don't the big IC design companies -- e.g., Intel, Apple, Samsung -- just build their own in-house tools instead of paying millions in license fees?
> It's the lithography tech that's the real crown jewel, and it is being protected.
Sure, but lithography tech is useless unless you have access to solid tooling for design verification. In IC design, no one can afford to debug issues after the chip is fabricated. EDA toolchains include complex simulation software that, when combined with fab-provided models, allow you to simulate circuit behavior down to the lowest level.
Millions is not that much. TCO of six engineers for a year gets you into millions. If it's a solved problem, it's probably worth just buying off the street.
I hope someone can correct me in this thread, but I recall hearing that a single-seat license can cost ~$50k per year. I would imagine that any company with a large number of ASIC engineers would love to reduce that overhead.
Regardless, none of what you said disagrees with the point that EDA tools are complex and should therefore be protected from Chinese influence.
You always get screwed in niche B2B on a single seat license. There's no way the chip manufacturers are actually spending anywhere near that much.
And they really aren't that complex. China has more than enough capability to reproduce them.
I guess I'm not sure how to prove a negative here. What do you think is so special about EDA software that isn't fairly easy replicable? Other than the fab design rules, which if I'm making a litho vs EDA distinction in the first place aren't really a part of EDA.
And I’m not sure why you still think that EDA tools are so simple. I guess we’re at an impasse, then.
Going back to my first comment: I would be extremely impressed if China is able to build a home-grown, state-of-the-art EDA toolchain. I also still think that the US should work to protect its existing advantage in this area.
The two are quite linked. You can't make a good open EDA package without knowing current design rules and the like, and if you want design rules that aren't old enough to drink you'll be bound by NDAs, as fabs strive to 'protect' every bit of knowledge about their 'crown jewels'.
Sure, the implication though is that if you've got the fab design rules it's not that hard to make the EDA tools. It's not 1982 anymore where the DoD is spending obscene amounts of money to run this on a 10Mhz mainframe.