Gauging the return on R&D: Do tech giants spend too much?
Tech giants are spending heavily on research and development, but it's difficult to follow the bouncing return on investment ball.The research and development return topic was brought up by my ZDNet UK colleague Rupert Goodwins earlier this week.
Tech giants are spending heavily on research and development, but it's difficult to follow the bouncing return on investment ball.
The research and development return topic was brought up by my ZDNet UK colleague Rupert Goodwins earlier this week. Rupert went on a tour of Microsoft Research in Cambridge and questioned whether R&D was the best use of the software giant's cash. The post also spurred a rebuttal. And in February Microsoft CEO Steve Ballmer defended Microsoft's R&D spending and outlined the company's big bets.
Ballmer said the time to invest is during a downturn. "We certainly invest in research and the incubation of new technologies, and new ideas, and then unfortunately every company has got a corporate overhead and G&A associated with it that we invest in," said Ballmer.
Rupert isn't so sure. He writes:
At a Microsoft Research's open day at Cambridge recently, this line was promoted heavily in the keynote speech: fundamental research helps generate brand new technologies that give companies competitive advantages. In some cases, that's unarguable. If Intel shut down its research, it would die overnight. But what would happen if Microsoft stopped doing research? Based on the researchers' demonstrations, the answer is 'not very much'.
The ideas on show in Cambridge, while good and interesting, did not address Microsoft's core problems, nor even any of their minor ones. There was research into ecological systems, into displaying networks of influence, into low-power network hardware, into capturing people's lives as a timeline. Try matching those with any known Microsoft strategy — or any conceivable one — in a way that makes compelling sense.
Rupert was looking for something tangible. His big idea: Microsoft should scrimp on R&D and maybe even kill the research unit. While that may be a fine short-term move it would be a mistake. You can argue that Microsoft should refine its R&D approach to become more efficient. Perhaps Microsoft should focus more on commercialization. But dropping R&D completely would be a miscue.
Why? Every company on the planet is trying to reposition itself on this curve (graphic courtesy of Leaderpoint) below. Notice that branch at the middle of that graphic, which I scanned from course materials so apologies for the appearance. That branch---market development---is future business. Those future businesses are what R&D is all about.
You can certainly question Microsoft's approach. According to IBM research presented at its investor analyst meeting Microsoft has spent $36 billion on R&D over the last five years. IBM has spent $30 billion. Here's the chart:
Is there a return on research amid those billions of dollars or is it just about patent bragging rights?
Intel was left off IBM's R&D spending chart, but according to SEC filings the chip giant spent roughly $27 billion on R&D over the last five years. Much of that R&D has translated into Intel's manufacturing prowess, but what's the real return?
Rupert mentions that Intel has delivered value from its R&D efforts, but that point can be debated.
Following Intel's analyst meeting this week, JP Morgan analyst Christopher Danely made a few interesting points. First he pooh poohed Intel's talk of growth initiatives and remained decidedly skeptical.
The company also spoke on its growth initiatives, largely centered around leveraging its Atom processor into the Netbook, handheld, consumer, and embedded markets. Intel stated its advantages are in process technology, scale, and platforms. The company also indicated it could command a premium in certain markets due to the desire for Intel architecture. We would note Intel’s revenue growth has only averaged 4-5% over the past 10, 7, and 5 years despite a multitude of growth initiatives and has been below the average growth of both US and world wide GDP.
Then Danely broke out this chart (click to enlarge):
In a nutshell, Intel's revenue growth lags GDP on average. Danely also notes:
We believe Intel’s disappointing revenue and earnings growth has been due to the disappointing returns on growth initiatives such as Itanium, Xscale, communications ICs, and Wimax over the past ten years.
Some of those aforementioned items used to be lumped into R&D spending.
Following the bouncing revenue ball is also difficult for other companies. IBM has really stepped up the commercialization of its research unit. Some of these advancements could be game changers. Others may just be a conversation starter that turns into a global services deal. Simply put, research is a fuzzy item. IBM has turned to more of a commercialization machine. Note the focus areas here:
IBM certainly seems to be better at commercializing its R&D---at least recently, but there are some duds there to be sure. R&D is often about failures too.
Simply put, it's difficult to draw a straight line from R&D to revenue. Innovation is sprinkled between multiple business units and its effect is often indirect. The IEEE Spectrum has a handy calculator that shows the top R&D spenders as a percentage of sales. The top of the list doesn't feature a lot of stock market all-stars. AMD, Micron Technology, Nortel, Lucent and Sun Microsystems were some of the top spenders as a percentage of sales. All of those stock charts---assuming they still exist---are disasters.
Tracking returns on Microsoft's research is also fuzzy. A programming language, F#, has gone commercial after it was cooked up by Microsoft Research. Vista includes numerous Microsoft Research items. Surface may turn into something. Azure will definitely turn into something. You can find Microsoft Research's technology transfers on its site.
Meanwhile, Ballmer indicated that a lot of the R&D team is working on servers, a very profitable business for Microsoft. He said:
On the investment side, we just chose to take Windows and mobility, client operating systems all up, so that we could get a real comparison with Apple.We think we're the big R&D spender in this area, but you should remember or R&D group is also doing shared technology for the server.I didn't try to tear that apart, so some of those 8.2 thousand people in our case are actually working on server operating system technologies.
How much has Microsoft Research produced in terms of revenue? Who knows. But it's clearly not something that should be jettisoned. The R&D call should be for more efficiency not wholesale cutting.