Tackling EDA’s Broken Business Model: Blaze DFM CEO Jacob Jacobsson (Part 4)

SM: Can you talk about Blaze’s value proposition and business model?

JJ: We change the way a process works from an insider’s point of view. You have certain manufacturing processes with certain leakage, power, and performance characteristics. We change those to something more favorable. We do that by adding a software step at the far end of the design cycle. For example, the designer may have been designing towards a 90 nanometer process. We change that by adding software to make it like a 90 nanometer process on steroids. Then power performance is much better – it’s similar to actually going to a foundry that has better power performance, but you do it in software so you do not have to go to another foundry.

SM: You avoid the extra cost of an additional process.

JJ: We hope that you would pay something for it, but it is probably more cost effective to get there.

SM: How about from the foundries’ perspective?

JJ: That is what is attractive from a design as well as a foundry point of view. If you apply this software you can, in a cost-effective way, offer a better process node without having to buy or tune new equipment. Both sides win.

I think here is another fundamental difference from EDA. We affect the parametric yield because you will have some chips, even in the old process, that have the same favorable characteristics. But these are just a small percentage of all of them. In that particular case, if that is important to you then all of the other chips will be scrapped. We enable many of the chips to have favorable characteristics, and very few will be scrapped. We have taken inefficiency out of the system. The designer wins, the foundry wins, and we do that without having to steal away a piece of the market from someone. That is another way of looking at the problem with EDA. I can only be successful if I make someone else less successful.

On some level it is zero-sum, but the sum is big enough that there is some room for this. You basically tie yourself to this inefficiency instead of saying that anything that can be taken out of that inefficiency is for me to share in and participate in. You will see the amount of semiconductor material lost to parametric year losses is far bigger than the  EDA budget for the entire world.

SM: How big is that number?

JJ: Last year, conservative estimate were between $20 billion and $30 billion that was just thrown away. I am only talking about parametric loss. There are other losses as well, such as yield losses.

This segment is part 4 in the series : Tackling EDA’s Broken Business Model: Blaze DFM CEO Jacob Jacobsson
1 2 3 4 5 6 7