Design Article
Opinion: EDA is not well -- where is it heading?
Paul McLellan
11/6/2008 9:16 AM EST
There used to be a healthy EDA world. A few years ago, there were: many EDA journalists; Gartner covered EDA; the large Wall Street firms all had analysts covering EDA; many venture capitalists invested in EDA startups; and every quarter, a startup would go public or be acquired for an attractive valuation. EDA was democratizing semiconductor design, pushing it out from its heartland within the large semiconductor companies themselves into system companies, fabless startups and design houses. Life was good. The industry was profitable.
The world's $3 trillion dollar electronics industry is largely dependent on the world's $400 billion dollar semiconductor industry which, in turn, is dependent on EDA, a tiny five billion dollar industry. EDA technology is a requirement for this value chain. But the EDA industry, as currently structured, might not continue to be the place that continues to provide it.
There is a tragedy of the commons with respect to money flowing from semiconductor companies into EDA. Overall, semiconductor companies require EDA to be well-funded. But it is in the interest of each semiconductor company to negotiate the lowest possible cost for its own EDA purchases, hoping and assuming that other semiconductor companies will take up the slack and pump enough money into EDA to keep the innovation engine running.
That doesn't seem to be happening. With hiring freezes and large layoffs, the big EDA companies are going to be underpowered on innovation. Further, with official and not so official policies against acquisition of startups, venture capitalists have largely exited the industry, looking for more attractive rates of return elsewhere, and so the flow of small companies is running dry.
Part of the reason that EDA is sick because it is sailing into a strong headwind. Chip design starts are falling and semiconductor R&D budgets are under pressure.
EDA companies like to believe that product companies want to design chips to achieve differentiation. While it is true that product companies want to achieve differentiation, designing a chip is the last resort. Nothing is more costly or more risky. Differentiation is increasingly in software and design.
Look at the original iPhone. All the differentiation is in Apple's software, industrial design and branding. Nobody cared about the silicon, mainly an old baseband chip from Infineon that had already been superseded. Ten years ago, Apple would have had to build some chips to get out a product like that. To do that, it would have had to buy lots of design tools. Nokia used to design lots of chips, now it lets ST and Broadcom do all the heavy lifting and ST already has all the tools it needs.
Innovation in EDA occurred mainly in startups. This is just a statistical fact. If 30 design-for-X projects are kicked off, a few in big companies and most in startups, it is numerically most likely that the one that wins and achieves market traction comes from a startup, not the internal projects. As the largest company at the time, Cadence's deliberate killing off of the startup ecosystem to focus on internal development didn't just hurt Cadence when its internal development failed to produce. It hurt the entire industry since it drove away the VCs who would otherwise have funded more startups, which in turn meant that the innovation hasn't been done at all.
So what is likely to change? One scenario that is clearly not going to happen is that the world's electronic industry stands around and waits for EDA to get its house in order. One company alone, TSMC, makes so much of the world's leading-edge silicon that it alone suffers hugely if tapeouts of chips slow up due to lack of new technology in EDA and the cost of one of its fabs is more than the market cap of the entire EDA industry.
For EDA to get healthy, there are several possible scenarios:
- EDA can be structured like it used to be, with a few big healthy companies and a thriving ecosystem of small startups that serve as technology incubators. For this to happen, prices will need to increase which could happen by changes in business models, or by the natural increase in pricing power that will accrue to Synopsys if Cadence doesn't recover. Or,
- EDA can become part of a larger industry, say, mechanical design or enterprise software. Or,
- design can return to being done internally in the semiconductor companies as it was in the early 1980s and still largely is within Intel and IBM, perhaps jump-started by some acquisitions. TSMC acquiring Synopsys anyone?
- Finally, perhaps a consortium of semiconductor companies can start a new EDA company to serve their needs, much as SDA (now Cadence) was originally funded. Full circle.
About the Author:
Paul McLellan is a turnaround CEO, marketing expert in EDA and semiconductor industries. Most recently, he was CEO of Envis.
|
|
|
|
|
Alan Smith at Altium
11/6/2008 8:00 PM EST
Interestingly, we at Altium share Paul's sentiments and agree with most of his observations. Why, you may ask, should another so-called EDA company say this? Because Altium does not think of itself as "another" EDA company, or even an "EDA" company at all. We think of ourselves as Altium, provider of next-generation electronics design solutions.
Those whose business models have diverted them from what Paul describes as innovation in the development of the tools required by engineers and designers to exploit the power in semiconductor & FPGA devices, may well be in strife. In Altium's case, building a solution that helps electronics organizations exploit the increasing power of programmable devices, wireless devices, processing power and the internet is exactly where we have been focused, and our results are telling. We reported sales growth of 22% in the year to 30th June 2008, and sales growth of 28% in the most recent quarter.
Earlier this year we made the commitment to release new versions of our software to the market every six months. The current release hit the market in May, and the next will appear before year-end. Interestingly, our company has always focused 100% on developing solutions expressly to help electronics engineers bring successful designs to market, not creating solutions as a hybrid or adjunct to some other business activity.
And to Paul's example of Apple's iPhone, yes - innovation will be in the device intelligence and the software in the centre of the design in the future. That's Apple's great innovation - not the sleek hand-held unit (although that's part of it) but the embedded interface and connectivity into a wider ecosystem of value. Being connected, being updated in the field, not being reliant on hardware that is easily copied (even Apple's great designs have multiple clones within weeks) is where the design innovation will come from.
Continuing to do tomorrow what was done yesterday is of course one definition of madness. So the focus needs to be on creating the tools required to do electronics design in new ways, focusing on where the value is created, and supporting engineers in getting that innovation onto boards, into manufacturing, and out to market.
(Note the change of sequence in the phrasing here - device intelligence, board, manufacture; not board, which device, and then oh - it's not working, let's start the process again.)
It requires a change of approach. And engineers need tools to help them explore new concepts, experiment with different devices, and rapidly move to reconfigurable prototypes and then be able to take them through the design process out to manufacture - preferably all within a single design solution.
But this won't be done by simply accelerating what went before.
Whatever the EDA sector is must change. We disagree with Paul in that designers won't wait - they can't afford to wait (and their CEOs won't let them, any way). Provide the tools they require for the next generation of products, let go of 30 years of habit, and embrace that change. Let's focus less on measuring an artifice called the EDA sector, and let's focus on creating great solutions for tomorrow's electronic products.
Sign in to Reply
NoelH
11/6/2008 10:19 PM EST
I think your analysis is good however there has also been excess in the EDA industry. Good salaries, expensive marketing and a good deal of, shall we say, over-confidence in the market. This is born out in the numbers too; Cadence spends 52% of its reveneue on sales and marketing, compare that with ARM who spend 18% of revenue on S&M.
I accept that this is a crude measure, they are different companies with some different business dynamics. However, they sell into the same industry (semiconductor design and manufacture) and the contrast is quite stark!
My view, we will see a smaller but leaner, meaner, more technology lead EDA industry.
Sign in to Reply
wieslaw
11/7/2008 4:43 AM EST
Just one remark Paul, I do not think prices should increase for EDA tools. Probably they are ok.
But disastrous discount policies should just be banned. How can EDA make money if some leading company practices 80% or 90% off, or even worse, buy one at high discount and get 5 free of charge?,
As we, selling EDA tools and trying to keep the prices as high as possible (aka small discounts) say you do not get 80% off on a car!
Sign in to Reply
SysTom
11/14/2008 12:39 PM EST
Wrong on all counts:
1) EDA budgets are down so your solution is the EDA co.s should raise their prices. No, you are wrong sir. The solution is for open-source EDA software to return sanity to EDA tool pricing.
2) EDA should be consolidated into a larger industry. Why?
3) "Not invented here" mentality should return - yeah that's the answer.
4) The consortium idea is not bad - except it would be dominated by The Big Three and would only exacerbate the problems they have created.
I think you must have a particular axe to grind coming from the EDA industry.
Sign in to Reply
Hillol
12/7/2008 11:55 PM EST
Engineering Market is $3 Trillion Dollar.
EDA is focused on IC SoC design. The growth
is limited. In Mobile computing environment
SoC and number of new design is not growing.
EDA market is not growing because number of
new design are limited. It is not going to
change. If EDA can merger with total engineering solutions, it will create a new market for EDA. We can call it Total System
Engineering TSE. We see that FPGA is doing well because it is low cost system solution.
The uP and DSP can solve most system related issues.
Hillol Sarkar
hillolsarkar@hotmail.com
Sign in to Reply
ekne
2/8/2009 12:03 AM EST
EDA tool vendor survival strategy ??
Will the EDA tool industry put the life back into SFLM ?
Looking at the type of companies around the business concept of selling a tool for a long term fee but the usage is a few weeks a year is a dying concept, and surly the tool vendors sees that...
So why not revive the concept of pay-per-use ? And at the same time also consider hardware/software bundles.... Well, some of you in accounting will turn around a couple of times, since now you can't have a predicted income to book, well welcome to the 2:nd millenia.... The technology is there and the "accounting" can be utilizing the many different technologies such as "prepaid" accounts....
Smaller companies will twist and tweak for any number that is in the 6 figures, for a tool license and a short usage..
Just let this fly around your head a bit......
You get a "pre-paid" account with the possibility of recharging it. a la phone cards, yes you heard it for the first time here, and I will let the idea fly around
You get the license when you need it and then no more charge, instead you move on to another phase and another license.
You also need some more compute power during 1 month... well bundle it with a license.
So what is preventing this, well some legals and where the data resides when it comes to hardware...
Well you can deliver hardware onsite, just like you deliver a photocopier or rent-a-tv/sofa so it's not really that hard to do it.... and licenses are still part of the software space so managing the licenses a la SFLM or local license servers where you go to an offsite to download a license file has been around for many years so all bit's and pieces are there...
So why isn't the EDA tool vendors looking at this, to get some market from smaller design houses ? Even medium and big size would be interested ..
So my fellow EDA Experts, why isn't this an already working concept in this industry ?
Bengt-Erik Embretsen
Sign in to Reply