Those who thought that EDA360 was the musing of an executive new to EDA should begin to rethink their positions. Denali is the perfect model for the expansion of Cadence's business toward a mix of software, IP, and design tools. I find it interesting that during all the years in which Denali had been held up as the model of the "small" EDA vendor, no one had offered an explanation for their success, given that they certainly did not fit the "traditional" EDA vendor mold.

Denali started by addressing an area no one was paying attention to: memory models. It was not considered important. Now, Denali dominates that market, and, according to Cadence's John Bruggerman, memory models are the primary reason for his company decision to acquire Denali.

I will make a prediction today about both Cadence and the other EDA vendors that will be leaders five years from now. In 2015 it will be hard to say whether Cadence is an EDA vendor that also sells both hardware and software IP, or a "fabless foundry" that also sells EDA tools. Not long ago I argued in print and on panels in favor of counting IP as part of the EDA industry, as EDAC had started to do. Almost everyone accused EDAC of wanting to inflate EDA revenues by adding IP. My position was, and continues to be, that IP is a design tool, because the architectural organization of a product grows with considerations about IP use or re-use. Many IP cores also require specialized EDA tools in order to be integrated into the design and, just as Denali has demonstrated, it is possible to profit from developing and selling both.

Peggy Aycinena said two years ago that a large semiconductor company would eventually purchase an EDA vendor: Cadence has started that transformation, but in the reverse direction than what Peggy predicted. The end result of the prediction was and is correct, just the methodology is, at least for now, a bit off.

I agree that someone could take the position that growing the non-traditional portion of the business is competing with our own customers. But I maintain that being just a producer of tools that allows semiconductor technology to continue to follow Moore's Law is a dead end. As Dr. Bernie Meyerson of IBM stated some years ago, and just restated during the Globalpress Workshop at the end of April this year, atoms do not scale. Innovation will come in the form of new architectures, new computing algorithms and new materials, not in making a substrate 2 atoms thick instead of 5. You got to figure in costs of production, not just technological achievements. The latest process still cannot solve the power consumption problem: you cannot trick physics no matter how small you go.

If we can design a different system implemented with 40nm geometries that processes information better than a traditional one at 20nm, why spend the additional money in production? Just because we can is not a valid answer. The future, as I wrote less than two weeks ago, rests in heterogeneous systems that employ many disciplines to implement solutions for the real world. A world that is analog, mechanical, and biological. A world that can use computation, but the type called numerical analysis, not arithmetic.

The solution has many layers, of which the applications layer is the top one. EDA vendors do not need to develop applications: they need to enable the development of application, through engines that have both hardware and software components. ARM and MIPS have cores that are central processing units that compete with Intel's products. Where is it written that these two entities cannot be a part of an EDA company?

The View From Wall Street

Cadence is a publicly traded company and as such this transaction is important enough to generate an immediate reaction from analysts. Jay Vleeschhouwer, the dean of EDA industry analysts, lost no time in sending out his analysis of the transaction. In spite of the fact that one of my degrees is in Business Administration I have succeeded in the past to misinterpret Jay's remarks a couple of times. I got caught in the semantics that deal with projections of future revenues, and I certainly do not want to make the same mistake again. So I am going to publish Jay's summary of his comments as I received them on Thursday May 13th.

"This morning, Cadence announced an agreement to acquire Denali Software (private) for $270 million in cash, net of Denali's cash on hand. The net price equals over 40% of Cadence's most recent cash balance. Denali had trailing twelve-month revenues of $43 million, implying a 6.3x EV/sales, and its operating margin was said to be more than 30%, unusually high for a company of that size, and above the margins of the four largest publicly held EDA companies (we believe that Synopsys recently paid 3x or more on an EV/sales basis for CoWare, though the latter had little, if any, growth compared with Denali's recent "mid-teens" growth—better than EDA itself— and limited profitability). In comparison with other IP companies, we note that ARM Holdings (ARMH, $11.11, NR), MIPS Technologies (MIPS, $4.66, NR) and Artisan Components (prior to its acquisition by ARM) have, or had, operating margins in the low-to-mid-20% range.

Cadence expects the deal to close during the current quarter; there is no earn-out in this transaction. It will be dilutive by "$0.01-$0.02" a share to FY10 and should be accretive to FY11, for which we're currently estimating non-GAAP earnings of $0.15 a share and $0.47 a share, respectively. Cadence will necessarily have to incur some write-down of acquired deferred revenues.

Cadence had in recent years been opposed to having its own IP portfolio (as distinct from Synopsys' strategy) and preferred instead to partner with IP companies and facilitate customers' use and verification of IP through its tools and services. (One reason was that it didn't want to compete with its own customers; we've had our own doubts about the scalability of the EDA IP model, though Synopsys has now gotten to a certain size—over $100 million—where it's making a better margin than was the case several years ago.)
We view this purchase as a significant change of direction, the idea being to cover a broader part of the design process by owning a leading IP supplier. The company highlighted Denali's specialty in memory modeling and its design IP as a complement to the Cadence Open Integration Platform. There is some question as to the size of the addressable market and whether Cadence will have to engage in additional acquisitions of IP (as Synopsys has done over the years); by our calculation, using EDA Consortium market statistics, Denali's addressable IP categories had total 2009 revenues of just over $100 million, implying good share for Denali, but a limited market nonetheless."

What I understand about Jay's analysis is that although the acquisition will have a positive impact on the bottom line beginning next year, Denali's position in the IP market is small in comparison to the total size of the market. This is a fair statement, that someone will interpret as a negative aspect of the deal. In fact, and I once again return to the EDA360 document, the acquisition is another sign that Cadence intends to be a significant, although not (yet?) dominant, participant in this market. I am sure that Cadence will not stop here. Cadence, and especially John Bruggerman, has recognized that significant growth does not rest in competing with Synopsys, Magma, and Mentor in offering silicon focused EDA tools. In spite of its size, or may be because of it, the traditional EDA industry is mature and does not offer any opportunity for a paradigm shift that would result in significantly increasing its total revenues. The opportunity rests in considering a design in its totality of components and in assembling the tools required to address the challenges found from architectural design to manufacturability. The availability and use of both hardware and software IP is a fundamental aspect to what we call the "front end" design flow.

Jay's complete analysis, not fully reported here, acknowledges that the acquisition is outside the "traditional" EDA industry. Within the industry Cadence becomes more like Synopsys who has over the years grown its IP business to over $100 million. Jay's financial projections for 2011 show Cadence achieving a much better P/E (Price to Earnings) ratio than 2010, and the street's consensus on revenue per share grows from $0.10 in fiscal 2010 to $0.43 in 2011.

The Bottom Line

Those who want to be ready for the coming changes in our industry need to consider that silicon is just a vehicle, not the end to itself. Could we have electronic systems built with other materials and other technologies? This is not an idle question and its answer is: absolutely yes.

Of course EDA must continue to support the design of semiconductors, but we need to enlarge our horizons. We must think of the system, of all the parts of the system, not just the electronic part ,be it digital or analog. Digital was relatively easy, analog is a bit more difficult, but what about software or mechanical and biological interfaces? What about adaptive systems that reconfigure themselves as needed reacting to pre-defined conditions?

When I started in this business, in 1966 it was not called EDA. Now we are beginning to see that the "E" part is a province of the world of Design Automation. We must have the courage and fortitude to change, to explore and exist in other provinces in order to fully understand our world.