feature article
Subscribe Now

Counting the Cost of 2008

It’s been a funny old year, has 2008. Traditionally this is the time to pull your chair close to the fire, grab a glass of mulled wine and tell ghost stories. We won’t do that this year: no yarn we could spin would chill your bones half as effectively as the news about sub-prime mortgages, Ponzi schemes/scams, financial meltdowns, factory closures and Reductions in Force.

(Interestingly, a quick Google search show that the use of the term, Reduction in Force, seems to be mainly in press releases from North America. I suppose it is easier on the corporate Human Resources’ tongues than to use words like “lay-off” or “fire.”)

Years ago, I learned that accountancy and financial people see the world in a very different way from the rest of us, certainly different from the way that most engineers see it. Now there are all sorts of reasons for this: one is the constant pressure on companies, particularly in the US, to ensure that the quarterly earnings figures look good. Another is that there are areas where arithmetic does not explain simply what is happening. Yet another is the continual battle between regulators and the regulated: a government devises a set of rules for taxes – the taxed find a way around them before the ink is dry. Accounting regulatory bodies devise rules to ensure that a company’s figures are reported in a way that is as honest as possible – the auditors and the audited are immediately on the hunt for ways of presenting them in the best possible light.

It still seems incredible that a group of people can borrow large sums of money to buy a company and then, as soon as the deal is done, the loan becomes the company’s problem. A healthy company, paying reasonable returns to the investors, suddenly finds that it is saddled with debts for which it has to find hundreds of millions of dollars a quarter to service. Immediately the company looks a lot less healthy. Investment slows, risk taking is reduced to nothing, and bits and pieces of the company are closed down or sold off to reduce the debt. (Oh, and people lose their jobs.) Two major semi-conductor companies are working in this position.

How companies account for sales is another area of concern. The next few sentences are about Cadence, but this is not meant to mean that they are any worse in their behaviour than any other big company. When Cadence makes a sale it is, broadly, many tens of thousands, or even tens of millions, of dollars for a licence to use their software for, say, three years. This money is paid over the entire life of the deal, not up front. Yet, until recently, Cadence showed the majority of this income as a sale in the accounting period that the licence was signed (less a few percent for “maintenance”). The problem for someone like Cadence is that this is like holding a tiger by the tail. If you let go, the tiger will eat you, or at least hammer your share price. So only when you have some calamity like losing a huge chunk of your senior management, including the CEO, can you do something about changing the position – after all, by stating more accurately when the money actually arrives in the company’s coffers you effectively write a chunk out of this year’s income. Cadence is moving towards making some of these changes. The way that future sales will be reported will change (“We have changed the business model”), and this has been applied retroactively to some sales in the first part of the year. This means that the quarterly earnings reports for the first six months of the year are being restated, changing a loss of $7.6 million into a loss of $42.9 million. There will be an RiF of 12% of their staff, mainly by “flattening the hierarchy,”, as well as significant internal reorganisations

Cadence is by no means alone in this way of showing revenue, nor is it the worst. A computer company, now well dead, used to show a sale the moment a machine left the loading bay in the factory in the United States. These were mainframe machines, and if the sale was to somewhere like Italy, it might be many months before the machine was shipped, assembled, integrated and passed through acceptance testing and then many more months before the invoice was paid.

So, what does this have to do with engineering, apart from the phrase “financial engineering” – which is almost as much an insult to an accountant as “creative accountancy”?

Firstly, it means that perfectly good companies, making products people want and employing good engineers, can change from looking financially healthy to financially ailing, purely through the way in which financial people behave and choose to treat numbers.

It also means that, as the recession bites more deeply, many financial chickens will be coming home to roost. A small business may be totally sound one day, then the next day their customers, the bigger companies, exercise their financial muscle and arbitrarily extend the time they take to pay invoices, to protect themselves in a time of financial crisis. Their bank will respond by cutting back on the line of credit and increasing their interest payments. Suddenly the small business is in deep trouble.

Yet was there originally a real crisis outside a few banks lending injudiciously, or did we talk ourselves into this mess? In the last major recession, Franklin Roosevelt said, “We have nothing to fear but fear itself.” Were the current financial dominoes knocked over by a genuine major weakness in the economy, or by a minor weakness exacerbated by fear?

The future is hard to predict, and my crystal ball needs a new communications interface, but two things are certain about 2009: it is going to be difficult for many companies and there will also be some tremendous opportunities. If you are working for a company heavily burdened with private-equity-induced debt, then it is clearly going to be difficult as consumer demand continues to fall but the debt interest remains high. If you are working for a privately held company, then there is a fighting chance that you will do better.

But there are areas where there will have to be opportunities. The world needs to wean itself from its addiction to oil, so solar power is going to continue to be important. The output from photo-voltaic cells will need to be controlled and integrated with the local power supply. Electric-powered vehicles, either hybrid or pure, will need very sophisticated control electronics to get the best from their batteries and fuel cells, both discharge and charge. Elsewhere in automotive the legislative pressure for safer cars already means that new models will be required to have stability control, with lane change warning compulsory for larger vehicles. Governments are mandating improved efficiency for internal combustion engines, from improved engine management throughout the drive train to tyre pressure monitoring. Improved information to the driver can increase fuel efficiencies by minimising congestion.

Across the embedded spectrum, there will be incredible pressures to reduce power consumption. This is already well underway in portable and hand-held devices, but the demand for lower power consumption will spread into consumer and industrial. Electric motors are terribly inefficient. While the semiconductor companies have for several years been producing control chips and software that can make significant improvements, it will take strong financial pressures, and possibly legislative action, to implement these.

I didn’t do any specific interviews for this piece, but over the last few months, I have been asking people how they see the future. And no one was gloomy. Since the embedded community employs bright people, I don’t think this was whistling happy tunes, but a genuine feeling that, although the future is going to be difficult, many of the problems that do need solving will require the help of embedded engineers. From gossip, it appears that development tools providers are seeing an up-turn in tool spend as designers, under pressure to work faster, see working smarter and using tools to increase their efficiency as the answer.

Whatever the results for the economy as a whole, may I, personally and on behalf of Techfocus Media, wish you all a prosperous 2009, and hope that this time next year we looking at a turned-around economy and a booming market.

Leave a Reply

And That Makes 1

A Look Back at 2008

A look outside confirms it. The sky ranges from dull white to steely dark gray. An occasional glimpse of blue is quickly corrected to bring it into conformance with the surrounding severity. The sound of the traffic is enhanced by the swish of tire on wet road. And it’s cold. By California standards, anyway… Yup, it’s winter. No, you don’t see this in the travel guides, but as the old year gives way to the new, Silicon Valley receives – hopefully – its annual drink of water from the skies, and would-be outdoor adventurers, confined to close quarters, are given to musing on the waning year. As shall we this week.

All in all, it’s been a chaotic, tumultuous year, characterized by both dread and hope. And not even the staid world of semiconductors has been immune to what might be seen as excesses confined to the more high-flying daredevil businesses like hoardingbanking, mismanagementinvestment, and fraudspeculation. But the year started on a much more pedestrian level as the first 45-nm designs worked their way towards production – and made it into the world, bringing to fruition yet another round of miniaturization.

We are in another of those end-of-days times, when the physical limits of standard run-of-the-mill chip technology are threatening to bring to a close the world as we know it, leaving behind the general populace and taking the chosen few, those of the exotic boutique, to the next exalted heights. And yet, even as the select-in-waiting polish the bows on their pretty packages, the discount-store processors continue to find ways to extend the life of their shoulda-been-dead-years-ago technologies. 45 nm became 40 nm, and visibility is growing even beyond 32 nm to 22 nm.

The enabling of this continued progression was conducted amidst the din of the latest critical discipline for ensuring that chips could be made long after they should no longer be makable: design for manufacturing (DFM). Early in 2008, DFM announcements were frequent, even as it was questioned whether much of it was really just a boondoggle; after all, many of the techniques weren’t really being tested through solid experimental testing – doing that many test runs on processes just out of the oven is just too damned expensive. DFM excited the investment community and was largely the realm of small startups, after which, as is the pattern in EDA, the big guys started cherry-picking DFM players, integrating DFM into their own flows.

It’s one thing to joust as a small company against numerous other companies your own size, pitting point tool against point tool, knowing that your customer will need to integrate it into the overall flow, which is presumably dominated by a Big Guy. But once the Big Guys bring DFM into their own flows, it gets harder to sell against. If your point tool does the same thing as what’s in their flow, even if incrementally better, you’ve got some hard slogging to do. At some point, you’ve got to have a really strong story to get any attention. So it seems that, as we come to the end of 2008, the DFM noise has abated rather dramatically, DFM having been duly tamed, civilized, and made suitable for patrician company.

The gradual fading of the three letters DFM has been accompanied by a rise in the prominence of three new letters: AMS. Analog has been accorded what feels like an unprecedented level of respect as the world has come back to the realization that we are, in fact, analog beings, and if we’re going to bring everything together onto a single piece of silicon, we must allow the hard-working, understated analog denizens to mix and mingle comfortably with their hipster digital brethren. Not an easy task physically, given that things analog are notoriously sensitive and easily offended by the rather more boisterous and noisy digital types. But the approaching day of reckoning between the two is being abetted by modeling standards for analog in the VHDL and Verilog worlds and by increased tools focus on both the verification (via SPICE) and integration of analog within a single flow.

Power reduction has been another key theme throughout the year. While prior focuses on power have been dedicated to fundamental brute-force approaches to reducing leakage through such means as using low-κ inter-metal dielectrics or back-biasing transistors, recent discussions have been much more nuanced, dealing with the effective allocation of power to where it’s required. It heralds an incipient appreciation for using only as much power as absolutely needed, and only where needed. Excessive displays of speed are no longer lionized; the muscle cars of yesteryear are giving way to the more prosaic, aerodynamic, computer-controlled sedans of the future. You knew it was just a matter of time…

There is a space, however, where performance still rules supreme. While all tool-makers continually enhance their tools to improve speed and capacity – at least once their core functionality (and as long as their cash position) is solid – there is one key watchword that dominated this effort this year: parallel. Three out of the four Majors announced the results of intensive parallelization work. Unlike past efforts, where nominal attempts at threading were done and tossed onto multicore computers, these releases reflected significant algorithmic effort to provide parallel operation at a much finer-grained level than had been done before. And, while it’s one thing to develop new capabilities with parallelization in mind, these typically involved a rethink of old architectures and moldy code: a non-trivial investment in making stuff-that-already-works work faster.

Meanwhile, in the corner where dreamers prepare the next salvation for the coming apocalypse, exotic memory makers continued to position themselves as white-knights-in-waiting, and technologists furthered their investigations of 3-D structures. Memory, in particular, got more than its fair share of attention in this pub, and that will continue because, well, it’s cool stuff, often involving mechanisms far different from those we’re used to exploiting. And no other technology has the leverage of memory; the winner that manages to displace DRAM when it finally peters out will be able to view his professional life as but a memory as he sips mojitos on a private island in the Caribbean.

3-D structures can also play a part in some new memories in ways we have yet to elucidate in this forum (patience, grasshopper…), but, less exotically, piling chips to allow functionality formed from a veritable Dagwood sandwich of inter-stacked heterogeneous technologies is a 3-D approach that is also aligning itself to be at the ready at the time when we give up trying to mix all technologies on a monolithic, planar chunk o’ silicon. Its promise, however, as with that of much of the nouveau memory, is some years from being realized despite the attention it’s garnering now because, well, yeah… it’s cool.

But while we try to keep our focus here on matters more or less technical, it’s been impossible to ignore the rumblings on the business side of semiconductor design, which, it must be acknowledged, pays the bills. The EDA world is largely bimodal, consisting of a few Big Guys (relatively speaking) and, depending on the era and mood, a smattering of small folks, and no one in between. It’s kind of like watching a fire. A few flames get lit here and there, and if a new small flame grows enough on the fuel it starts with, it may merge with a larger portion of the fire that’s well established. If not, well, it may sputter and die. Occasionally one will grow on its own into a decent-sized fire, but that’s not so common in this world. If you get lit, your exit strategy more often than not is to blend flames with one of the bigger, better-fueled guys. And the mood comes and goes; eventually we run out of matches, and the lighting of new flames goes out of fashion. And then, later on, for no particular reason, folks get some extra cash to buy more matches and start lighting fires again.

This year we’ve had quite a few flames darting about in and amongst the four larger fires that are Cadence, Magma, Mentor, and Synopsys. But the balance of conflagration seems to be shifting rather dramatically, partly on its own, partly driven by the economic meltdown that is rippling far and wide throughout all endeavors. The most obvious victim is Cadence, reeling from management chaos, financial oopses, and a dramatic fall-off in reported business. If that cliff they’ve gone off reflects real business (as opposed to some revenue recognition artifact), it’s absolutely startling and terrifying.

They’re not alone, however. Magma has also had to pull back into its shell a bit; their numbers from this week show a similarly scary trend. They’ve had to refocus efforts on those things most promising and de-emphasize areas perceived as more of a luxury. Analysts are prone to comment here and there that someone needs to buy them; no such overtures have been visible. And one can well imagine that a potential suitor might be cautious, given where stocks have fallen. Cadence’s attempt to buy Mentor fell through; had it happened, Cadence might, in retrospect, be lamenting having paid what today would seem like a ruinous sum.

Mentor itself seems to be sort of holding its own, not hurtling into immediate ruin, but likewise not towards riches either. The only player that has managed to remain profitable, and which has earned the appreciation of numerous analysts lately, is Synopsys. They’ve managed to keep in their lineup pretty much all the technologies needed for a full solution and have made some aggressive competitive moves lately. The only one with a market cap over a billion – in fact, near $2.5B as this is typed – they are valued at more than 3 times the next comer.

As for the smaller guys, well, we’ll see how well they can survive the current panic. Whispers are that some VCs have told their companies that they will pull the plug on anyone not cash-flow-positive by the end of 2009. Funny how, in the financial world, the risk-taking dial seems to have two levels: zero and stupid. There’s no in-between. Oh well, they control the bucks, which makes them the experts. They must know what they’re doing, so we’ll watch with great interest as the products of their wisdom unfold.

Finally, as we draw to the end of 2008, we also mark the approach of the anniversary of the launching of this journal. It’s been exciting to see the readership grow at infinite rates! (Ok, ok, that’s cuz we started from zero… you got me…) Your readership and attention are what matter, and we at TFM are extremely appreciative of your willingness to direct your gaze in our direction. We look forward to bringing you more of what’s exciting in this industry in the coming year and encourage your feedback on how we can keep things relevant and even just a little entertaining.

While 2009 doesn’t look to be a cakewalk, we wish you all the best of what it has to offer and hope to see things looking brighter and stronger when next we review the ups and downs and ins and outs of this business.

Happy New Year.

Leave a Reply

featured blogs
Jun 9, 2024
I've seen several do-it-yourself (DIY) miniature versions of the Las Vegas Sphere, but this is the best contender thus far....

featured chalk talk

GaN FETs: D-Mode Vs E-mode
Sponsored by Mouser Electronics and Nexperia
The use of gallium nitride can offer higher power efficiency, increased power density and can reduce the overall size and weight of many industrial, automotive, and data center applications. In this episode of Chalk Talk, Amelia Dalton and Giuliano Cassataro from Nexperia investigate the benefits of Gan FETs, the difference between D-Mode and E-mode GaN FET technology and how you can utilize GaN FETs in your next design.
Mar 25, 2024