Earlier we had guest blog entries from Lauro Rizzatti of EVE and Chi-Ping Hsu of Cadence on whether innovation occurs in small or large companies. I’ve always maintained that the problem is a different one. I think it is clear that the engineering groups of large companies are capable of creating leading edge technology. Look at any franchise product like Design Compiler, Virtuoso or Verilog simulation and see how it has advanced over many generations spread over a decade or more in ways that involve large amounts of innovation.
Where large companies have a problem is that they are very poor at introducing new products into their channels. They have large efficient sales organizations but those organizations are geared up to closing deals with customers for products that the customer already knows it wants. Unfortunately, when a brand new product is introduced, there is an attitude among the salespeople that “he who goes first loses.” But just as the Luddites really were right that automatic looms would put them out of business, the first person in a large company to sell a new product really does lose. There will be problems with the product that will tie up their application engineering resources for months, and potentially a large multi-million dollar deal will be held hostage to problems in a single copy of a hundred-thousand dollar tool. Better simply not to sell the product until enough other sales have been made for it to be mature. But with every salesperson taking this attitude, no sales occur.
This can extend even to products that are acquired. When Cadence purchased Ambit’s synthesis product line, it was obviously very strategic for Cadence salespeople to sell it aggressively. If they were successful, it would start to cut off money flowing to Synopsys and even if they were less successful, they would force Synopsys to circle the wagons to protect its Design Compiler franchise and so have less effort available to put into threatening Cadence’s huge place and route franchise. But Cadence salespeople would not. They had big quotas at big semiconductor companies to close, and their focus was to let Synopsys have synthesis and try and close a deal to supply everything else. Selling synthesis against Synopsys required extra effort and the payback of a few experimental licenses would not move the needle on their quota.
Another product from my time at Cadence was called Heck (at least internally, I forget what unmemorable name it got given externally). It was a formal verification tool built on some technology developed at Cadence Berkeley Labs. To tell the truth, I’ve no idea whether it was any good or not, but since the salespeople refused to try and sell it we never found out. In the end Cadence acquired Verplex and the Conformal product line that customers were already starting to adopt.
Very few products have been successfully introduced by large EDA companies (once they have become large). By successful I mean built up into $100M per year businesses. And by product I mean a genuinely new product line, not a new version of an existing product. The only one I can think of is Calibre. This was developed over the years inside Mentor and somehow survived being canceled for almost a decade before coming to dominate physical verification. Cadence helped by making a huge misstep. They tried to protect their Dracula franchise by making their hierarchical DRC Vampire require incompatible rule decks. Mentor had no such qualms and as a result the obvious upgrade path from Dracula was to Calibre not Vampire.
Synopsys made PrimeTime a big success, but the story is complicated by the fact that they acquired Viewlogic and with it Motive, the market leader in static timing. They then shut down Motive and transferred all its customers to PrimeTime. But undeniably they did manage to get their salesforce to sell it.
So I think that it is not so much that large EDA companies are incapable of innovation. They do it all the time. But their salesforces are reluctant to sell any product for which there is not already strong market pull. Marketing in EDA is unable to create that demand either, which is a different story.
However, startups are different. The salesforce will sell new products because the salesforce typically has precisely one product to sell, and it is new. They are not really the same sort of salesperson either. Startup salespeople are more like hunters whereas large company salespeople are farmers. It seems to take that combination of single mindedness in the salesforce and an entire company whose success depends on getting those initial customers to adopt the product. Once customers start to clamor for the product, it is the moment for a large EDA company to acquire the startup and the huge machine of their salesforce can drive the bookings number up very rapidly.