Tackling EDA’s Broken Business Model: Blaze DFM CEO Jacob Jacobsson (Part 6)

SM: How did you convince TSMC that your process works and that you should get paid for it?

JJ: TSMC would not endorse something like this without verifying it was sustainable. There was a lot of cajoling to get them to run the necessary semiconductor tests to validate that this really works. We negotiated this for a year and half. Once they were convinced of the technology the battle was over, they knew what the benefits were. They get to offer a better process without the usual capital investment for it.We also had to convince TSMC was that we understood enough about the semiconductor business, that we were not complete idiots in terms of implementation. The way a semiconductor company looks at the world is really different from the way an EDA does because there is a very specific margin goal. They can’t afford to drop prices because there is equipment to pay for. There needs to be a relationship between value delivered and reward.

SM: There is a floor below the price.

JJ: For me personally, on some smaller level the fact that I did have fabless background probably helped in that case. I remember what we did at Xilinx was clearly an understanding that our foundry needed to make money as well. I am not really sure that the people who buy EDA tools know that is the case.

SM: They are doing themselves a disservice because if EDA does not move with the innovation cycles they are not going to be able to design chips. You cannot design a 32 or 22 nanometer chip without EDA.

JJ: One kind of leadership here could come from the customer base to realize that this is, in the long run, in their interest. I think we are going to go for some big catastrophe before people realize it. On the other hand while we might be going first with something here we are probably not the only ones. Other businesses can probably develop similar business models which are more symbiotic. We have the advantage that we are relatively small so for us it is not so troubling that cash flow gets back loaded by doing something like this. For a big company to change to a similar business model, you are going to have to finance this.

SM: Just going back quickly to the TSMC situation … You have got the TSMC relationship, are there any semiconductor houses operating on this process?

JJ: Just two weeks ago we announced there are companies designing with this right now. We will announce when we have someone who is willing to talk about it. One of the funny things for us is that we do not necessarily care, in this steup, whether a customer is big or small. When you are selling EDA licenses you try to get big companies who can buy a lot of licenses. In our case if a small company makes a really big volume runner, that is just as good as one of the big fabless deals. The size of customers is potentially irrelevant because if a particular tape out only results in 10 wafers, it does not really help us. Our concern needs to be on volume runners.

SM: How is your incentive structure tied to TSMC

JJ: We are very well aligned in this respect. If this results in better yield, or for that matter bigger wafer volumes because you offer something you did not have before, then we win.

SM: Super. The only company who has a comparable business model is PDF Solutions. Having said that they are not a software company.

JJ: In that case I would say Artisan is probably closer to us.

SM: No competition for what you are doing?

JJ: There is no competition. There are two different levels. We obviously have patents but you can be ‘out lawyered’ easily. This is an exclusive arrangement in both directions. We stay away from other foundries, and TSMC stays away from others with similar technology if there were any. No other patent in the world is going to deter another company. However, if the 800lb gorilla is Locked in, that’s a bigger issue.

SM: Beyond that you are going to be working with some of the IDMs?

JJ: I spoke about the need in aligning interest. It would not be to TSMC’s interest that we went after their immediate competitors, so we do not do that. It does not matter to them if we go after IDMs because that is not production they would get anyhow. The more I look into IDMs they are growing fab-like and that conflict might arise in the future.
I don’t know how it will all play out but at 32 nanometers we will see very few IDMs and the merchant foundries will have taken up that business.

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