A Brief History of Cadence: the Present Day

In the early days, like all the larger EDA companies, Cadence grew through a mixture of acquisition (for example, the Tangent routing technology), internal development (such as the Virtuoso platform) and, most especially, acquiring technology that was gaining traction in the market and ripe for accelerated growth that additional investment and a larger sales force could provide (for example, Gateway's Verilog simulation technology). The reality was that in those days, complete flows were assembled by combining best-in-class point tools.The EDA market was still very fragmented with a large number of small companies with important pieces of technology.

In that era, a good point tool could be put together with a team of engineers who knew what they were doing and with some venture capital. I had some experience in this when I was VP of Engineering at Ambit before Cadence finally acquired them, although I was not there when the company was truly a small startup. I think it was Jim Hogan, when we were both at Cadence, who told me that with point tools, you could try and develop them internally and sometimes that would work, but sometimes the best solution turned out to be elsewhere and acquisition made sense. In EDA, almost all the profit goes to the #1 and #2 products in a space, so if you don't already have the best, then you need to go and get it. With a healthy flow of investment capital, those startups would come into existence, often several (there were at least a dozen companies created to address design for manufacturing, DFM, for example). One or two would be the winners and would get sold for a premium price. The others would close down or go at a fire-sale price.

That environment has all changed in the last decade, at least for EDA tools. IP is a little different since it is more of a portfolio strategy and adding a good-quality new interface adds to the breadth of the portfolio. The only real disadvantage of a broad portfolio is the cost of maintaining it, especially moving from one process node to the next (or occasionally in the opposite direction now that there is increasing interest in non-leading-edge nodes such as 28nm).

So Cadence took a breather in 2015 and did no acquisitions. This reflects two realities. Investors are no longer funding new startups in the EDA space and so there are few successful startups with complementary technologies.

More importantly, the changes that the relentless march of Moore’s Law requires impact broadly across the whole product line and so are not the kind of changes that can be driven by small companies. Double patterning and FinFETs are the most obvious of these sea changes. No matter how much investment capital you could raise  (and in practice you cannot), you cannot create "the FinFET startup" since you need to already have a full product line.

These big changes have required a re-architecting of the entire implementation flow and a suite of new products: the Tempus, Voltus, Quantus QRC, Innovus, Genus, Joules, and Modus solutions. These all are built on top of integrated engines and have been re-designed to use extensive parallelism to take advantage of the large datacenters that semiconductor companies have all installed over recent years. The result is an order of magnitude increase in the speed of implementation of a design and better quality results.

There have also been extensive changes and developments in the functional verification area, mostly under the Incisive name, pulling together simulation, emulation, formal approaches, virtual platforms, debug, verification management, and more. This has involved more horizontal integration of different technologies along with vertical integration of verification approaches at different levels.

So that is a brief history of Cadence from its earliest days as the startup SDA to where it is today. Cadence just announced their 2015 results, with revenues up to $1.7B. Cadence has a large portfolio of semiconductor IP, a strong position in custom layout and analog, a completely revamped digital implementation flow, and an powerful functional verification suite.

The technical requirements are unrelenting. As you would expect, Cadence is working with leading-edge companies on 10nm, 7nm, and even 5nm. The history is not slowing down.

The complete genealogy chart can be viewed by clicking on the thumbnail to the right. Oh, and thanks to Donn Ledwick for the photo of his 1991 Cadence T-shirt. Love that corporate branding.

Previous: Blue Gecko, Designed with Cadence

Next: Happy 25th Birthday, Virtuoso!