An end to Grove's law?

As part of my reflections on the news of SAP spending $5.8 billion to buy Sybase, I wonder whether we will finally put an end to a piece of self-serving theorizing by Stanford adjunct professor Andy Grove.

Like so much in enterprise software nowadays, the deal is big, expensive, plays to gargantuan egos and does little to increase customer value. (At least this story has one happy ending — the bailout of Sybase shareholders who finally get what the stock was worth back in 1994.)

Apparently I am not the only one questioning the business logic of the deal. The Register headlined their story “SAP buys Sybase - but why?” while the MarketWatch headline said the acquisition “draws mixed reviews.”

One possible explanation is the standard one that I give my strategy undergraduates for most multibillion dollar acquisitions: the desire of CEOs and other execs to run a bigger company. On MarketWatch, a self-identified “SAP Insider” offered a similar explanation:

SAP is simply buying revenue. Their boards technology strategy talking is so inconsistent: they praise their own in-memory DB development, then the Sybase acquisition is mainly about the mobile business while widely avoiding the DB question which is still Sybase's main business. This proves inconsistency all over the place, maybe even panic. I think that SAP as a company is not at peace with itself. They don't trust their own potential anymore. The changes in the board haven't changed anything. Maybe the problem is beyond the board.

Buying revenue is common in mature industries, where firms unable to grow their core business buy other companies so the market cap and revenue graphs show an unbroken string of increases, in an attempt to hide that the core business has simply run out of steam.

However, there is another, more substantive possibility: an end to Grove’s law, which has been offered as a maxim of IT industry strategy for almost 20 years.

Everyone knows Moore’s Law, the 1965 prediction that the number of transistors will double every 18 months (at least until we get down to atom-scale transistors.) But his successor as Intel CEO, Andy Grove, claimed a new pattern of industrial organization in the IT industry, in which horizontal monopolies of components (such as x86 microprocessors) supplanted go-it-alone vertically integrated winners such as IBM and DEC.

In our 2000 paper on PC architecture, Jason Dedrick and I cite Grove’s 1996 book Only the Paranoid Survive (esp. page 44), but other data suggests that Grove was giving speeches to this effect at Harvard b-school during the early 1990s. (As a sidelight to his day job, Grove later became a lecturer at the Stanford Graduate School of Business.)

One of the problems I had with Grove’s claim was that it was so incredibly self-serving: the same claim by (say) Carliss Baldwin would be much more credible than coming from the CEO of a company accused of monopolistic behavior in restraint of trade.

Was the Wintel duoopoly a natural outcome of economies of scale, network effects and other “increasing returns to scale” (as might be argued by Brian Arthur)? Or was it merely two lucky companies that found a window (sigh) to dominate a market niche through a combination of normal and aggressive business practices?

Whatever the explanation, the stylized world described by Grove is just about gone, as IT companies integrate up and down the stack, vertically integrate along the value chain and otherwise increasingly pursue strategies of related diversification. Microsoft, Google, Sony and Nokia are just a few examples of such strategies, as are (to a much more mundane extent) the three enterprise software giants: IBM, Oracle and SAP.

Would it be more efficient to stick with modular component specialization and interoperability, rather than gathering more and more technologies under one roof? I don’t know, but the latter strategy is what firms are doing now, rendering Grove’s law now effectively obsolete.