These are all backend tools, and winning in that space is hard without strong relationships with major foundries. But what about frontend tools? The push for left-shift is everywhere across the chip design flow; you don’t necessarily need foundry certification. So why aren’t we seeing significant growth there either?
Even within this oligopoly of three major players, there doesn’t seem to be much innovation, whether applying GPUs or AI to make EDA tools smarter and design flows faster for quicker tapeouts. It feels like they’ve been selling the same inefficient tools for years.
And even among these three, there doesn’t seem to be real competitive energy to innovate or capture more market share.
I've seen a ton of innovation in frontend tools, from RTL copilots like Primus to verification tools like Silogy to RTL simulators like Verilator!
They don't really compete with Cadence, Synopsys, or Mentor, though -- those big three companies make the vast majority of their EDA revenue from the backend tools.
I've also already written quite a bit about AI-powered EDA frontend tools, so I figured that I'd talk a bit about backend today :)
As much as the industry needs newer and shinier tools that can help companies do a lot better, getting past the big three evil axes of EDA companies is a near impossibility.
Yet I commend a lot of people for starting EDA companies recently.
One interesting point to add is that, given how crucial the relationship between EDA firms and foundries is—and how vital foundries are to the entire chip ecosystem—it’s notable that some emerging Chinese EDA players, despite lacking comprehensive ties with TSMC and Samsung due to a series of U.S. restrictions, are increasingly strengthening their relationships with domestic chipmakers like SMIC and Huahong, which primarily produce chips above 7nm.
On that point, adding recent EDA control, it actually give better-than-average chance for those CN EDA firm, admittedly the challenges will still be significant given wide range of IP portfolios, foreign EDA firms' ties with domestic firms, technical barriers, and etc.
Just want to say, this is a fantastic piece. Learn a ton despite not unfamiliar with EDA firms and business models. Love the angle from startups competition against "big 3."
My ex is a Ukrainian lady - physics degree, immigrated to US in 1994, got her Masters in Comp Sci at Texas State, started out at Intel in hardware verification, got an offer from Apple that Intel could not match a couple years ago to join Apple, now she has her own team at Apple using Cadence or Synopsis tools, our son is a software engineer (UT), he says that in 2024 Apple made her work so hard without help, she was pouring over files up till 10:00 at night 7 days a week, cleaning up problems in the M series CPU's, but that Apple rewarded her a cool $700,000 in stock for her efforts.
Hey,
These are all backend tools, and winning in that space is hard without strong relationships with major foundries. But what about frontend tools? The push for left-shift is everywhere across the chip design flow; you don’t necessarily need foundry certification. So why aren’t we seeing significant growth there either?
Even within this oligopoly of three major players, there doesn’t seem to be much innovation, whether applying GPUs or AI to make EDA tools smarter and design flows faster for quicker tapeouts. It feels like they’ve been selling the same inefficient tools for years.
And even among these three, there doesn’t seem to be real competitive energy to innovate or capture more market share.
What are your thoughts?
I've seen a ton of innovation in frontend tools, from RTL copilots like Primus to verification tools like Silogy to RTL simulators like Verilator!
They don't really compete with Cadence, Synopsys, or Mentor, though -- those big three companies make the vast majority of their EDA revenue from the backend tools.
I've also already written quite a bit about AI-powered EDA frontend tools, so I figured that I'd talk a bit about backend today :)
Nice article and I concur with your analysis.
As much as the industry needs newer and shinier tools that can help companies do a lot better, getting past the big three evil axes of EDA companies is a near impossibility.
Yet I commend a lot of people for starting EDA companies recently.
Just curious on your choice of the word "evil", why are the big three associated with the word "evil" here?
One interesting point to add is that, given how crucial the relationship between EDA firms and foundries is—and how vital foundries are to the entire chip ecosystem—it’s notable that some emerging Chinese EDA players, despite lacking comprehensive ties with TSMC and Samsung due to a series of U.S. restrictions, are increasingly strengthening their relationships with domestic chipmakers like SMIC and Huahong, which primarily produce chips above 7nm.
On that point, adding recent EDA control, it actually give better-than-average chance for those CN EDA firm, admittedly the challenges will still be significant given wide range of IP portfolios, foreign EDA firms' ties with domestic firms, technical barriers, and etc.
Just want to say, this is a fantastic piece. Learn a ton despite not unfamiliar with EDA firms and business models. Love the angle from startups competition against "big 3."
Insightful piece. Earliest date mentioned here is 2003, but Cadence has been perfecting their playbook for far longer.
Cadence Acquisitions Timeline:
https://en.wikipedia.org/wiki/Cadence_Design_Systems#Acquisitions_timeline
Great article.
My ex is a Ukrainian lady - physics degree, immigrated to US in 1994, got her Masters in Comp Sci at Texas State, started out at Intel in hardware verification, got an offer from Apple that Intel could not match a couple years ago to join Apple, now she has her own team at Apple using Cadence or Synopsis tools, our son is a software engineer (UT), he says that in 2024 Apple made her work so hard without help, she was pouring over files up till 10:00 at night 7 days a week, cleaning up problems in the M series CPU's, but that Apple rewarded her a cool $700,000 in stock for her efforts.
Western Semiconductor is US only EDA company that can compete with Cadence head on. In fact, they are more automated.