Limiting Excess Profitability: A New Intel Motherboard Theory
Issue 12
Welcome to Bill Gurley Notes. Together, we’re taking a step back in time to the early days of the internet and into the writings of Bill Gurley, the famed 6’9 venture capitalist and author of his aptly titled blog Above the Crowd. We started with his maiden issue and are working forward in chronological order. To follow along with the un-notated versions, check out my friend KG’s comprehensive collection of Gurley’s writings here.
Limiting Excess Profitability: A New Intel Motherboard Theory
5/1/1995
“There are only two ways of getting on in this world: by one's own industry, or by the weakness of others.”
—Jean de La Bruyère
A New Motherboard Theory
We had an interesting discussion this past week with Sam Moore, an important industry expert and participant. (Of course the real name has been changed to protect the innocent.) Sam has a new theory regarding Intel's motherboard participation that could explain (1) Intel's reasons for being extremely price competitive on the motherboards, and (2) why Intel will never try to control more than 15-20% of the worldwide PC motherboard. Pay attention, this gets good.
Third newsletter touching on this topic! Clear he thinks this is going to have a huge impact on the industry.
Sam's theory works as follows. First, what is Intel's primary goal? To sell more processors, right? Now, what causes demand to rise? We think there are two major lever points for increased demand: increased usefulness and lower price. As you know, Intel has already begun to pull the first lever. Efforts such as the ProShare video conferencing system and NSP (Native Signal Processing) are clearly geared toward improving the overall usefulness of the personal computer. Now let's think a little more about this price issue.
Proshare: “Better business by being there.” Intel’s video conferencing product, developed in partnership with AT&T.
Native Signal Processing: included in Intel’s 1996 Pentium products, an innovation bringing sound processing to the CPU rather than offloading it to a soundboard as had been done previously. For better multimedia PCs.
The ultimate buyers of personal computers buy complete systems and not microprocessors. Therefore the price that affects overall PC demand, and by implication demand for Intel processors, is the price of a complete PC system. In other words, when system prices fall, Intel benefits because more processors are sold as a result.
This could theoretically allow them to charge more for individual components as long as they are consolidating the needs of other system components (eg., eliminating the dedicated sound board in the case of NSP).
Now let's think about ways that Intel could lower the overall price of a PC system. The most obvious example is to lower the price of the microprocessor. Since the processor has more margin than any other component in the system and is one of the most expensive components in the system, this seems reasonable. However, since Intel controls approximately 85% of the x86 processor market, this type of strategy would be very costly to Intel. For instance, if Intel wanted to lower the price of an overall system by $50, it could lower the price of an average processor by $50. If we assume the current market to be somewhere near 60 million units, this would cost Intel 2.5 billion dollars in marginal profit! (60 million units x 85% market share x $50 discount). Of course, this decline would be offset by the increase in overall demand spurred by the price decline.
As $2.5 billion dollars is a pricey figure, let us see if a more reasonable alternative exists. Intel's recent push into the motherboard business may be the answer. Once again, according to Sam, it is Intel's plan to permanently control about 15% of the overall motherboard market. If we assume that a PC motherboard is for the most part a commodity, Intel's pricing on motherboards will have to be met across-the-board by the competition.
Now, with this newfound tool, Intel can reduce the price of its motherboards by $50 and as a result reduce the price of all systems worldwide by $50 as well. This action only costs Intel $450 million in marginal profit (60 million units x 15% market share x $50 discount), yet accomplishes the same goal as lowering the price of the processor itself.
Essentially forcing other players to lower the cost of their motherboard and motherboard components in order to be competitive with Intel. Clever.
Is Sam's Theory Plausible?
We must say that we found Sam's theory very intriguing (which is why we published it). Is it reasonable? Well, Intel has suggested that its prices are in line with those of Solectron, which has a profitable contract manufacturing business. If this is true, Sam's "loss leader" theory would indeed be hard to support. (If any industry executive cares to validate Intel's motherboard pricing, we would be more than willing to listen!)
Additionally, this motherboard pricing lever would be an extremely limited tool. Returns on assets and equity in the motherboard manufacturing business have been quite low historically. If Intel lowered prices too much, it would simply put everyone else out of business, soak up all the business, and eliminate the usefulness of this idea.
Right, then they’re stuck with the low margins.
What Intel may be able to accomplish is to ensure that PC motherboard design and manufacturing is a value-neutral business. We define a value-neutral business as any business where no one player earns a return above its cost of capital. In other words, no one earns extraneous profits. Eliminating non-Intel extraneous profits from PC system prices is clearly beneficial to Intel.
Compaq vs. Intel (Round Two)
All this talk about removing extraneous profits from motherboard design and manufacturing reminds us of the ongoing bout between Intel and Compaq. We want everyone to know that we here at Above the Crowd are dedicated to bringing you a complete round-by-round analysis of this most fascinating event.
Okay, so to get you (and me) up to speed: the first bout of the Intel v. Compaq fight began with Intel’s consumer marketing campaign around its Pentium chips. Compaq’s CEO told an Intel exec in public that “"Intel is not trying to do the best thing for its customers…When you're in a near-monopolistic situation, the temptation is to take advantage of it.”
Intel maintained very high gross margins of 55% through this era, while PC makers were left to battle it out. Compaq had a ~24% margin, the highest in the industry. Head here for more on the original conflict.
The players have just completed round two, and although there were no knockdowns, our scorecard had Intel ahead on points 10-8 (based on a 10-point must system). One of the points was awarded to Intel because of the delay in the K-5, AMD's Pentium competitor. The other point is being awarded for an event that may have been entirely unintentional on Intel's part, but nevertheless quite costly to Compaq.
During Compaq's recent earnings conference call, management mentioned that the output of servers and desktops was limited by the availability of SRAM. As we checked into this issue a bit further, we learned something very interesting. There are currently two flavors of burst SRAM available in the market--flow-through and pipeline. Flow-through has slightly better performance and pipeline is more economical. Compaq originally designed flow-through into its high-performance systems. Meanwhile, Intel decided to use pipeline in the majority of its motherboard designs. The result is that industry capacity began to migrate toward pipeline, restricting the availability of flow-through SRAM. Intel moved the market one direction and Compaq was caught looking the other.
Accidental, but still easy to see how this is painful for Compaq. Intel boasted 85% market share in chips in 1995; Compaq had 10% of the PC market.
Compaq is now supposedly adjusting its designs so that they can accept both flavors of SRAM-- an interesting acknowledgment by Compaq. But keep in mind that this little experience has cost Compaq in two ways. The first is missed sales. Compaq's first quarter revenues were several hundred million dollars below most analysts' estimates. As the PC becomes more of a commodity, the chance that temporarily lost sales become permanently lost sales is increasing. Additionally, Compaq will likely have to respin its cache controller ASIC in order to accommodate pipeline SRAM. This new cache controller and system design will then have to be tested and supported.
Looking Forward to Round Three
On the overall scorecard, we now have Intel ahead 20-18. Round one we called even, because although Intel was quite successful with its "Intel Inside" campaign as well as its consumer Pentium push last Christmas, it did suffer that $475 million setback due to the FDIV bug--an unignorable cost of speeding up processor product cycles.
Caused a problem with the floating-point unit of some Pentium processors.
We expect the fight to heat up in the coming rounds. Intel has the NSP effort and PC server building blocks designed by Sequent in its corner. We have also overheard Compaq's trainer mumbling something about NexGen. Moreover, if push comes to shove, we would not be surprised to see Compaq put on the PowerPC gloves in the later rounds. Ding, ding!
PowerPC deal did not happen, though Compaq did manufacture its own chips beginning in 1998 with its acquisition of Digital Equipment. It returned to Intel beginning in 2001.
Takeaway
A much newsier piece than we’ve seen recently from Gurley—though contains some nice tidbits of competitive theory.
He opens with a theory from an unnamed source about Intel’s participation in the motherboard business—a change he thinks is going to have massive ramifications for the PC industry. The source thinks Intel doesn’t want to control more than a fifth of the motherboard industry, instead using its participation to drive the price of all chipsets down.
The effect of such an action would lower overall PC component prices and result in lower PC prices for consumers and higher volumes of chips sold—all without Intel having to lower its CPU prices, the company’s golden goose. At this time, Intel held 85% market share and its CPUs represented a quarter of the component cost for any given PC.
Gurley ultimately doesn’t think this theory has much explanatory force given the already slim margins in motherboard manufacturing. So why does he write about it? Well, it’s clear from his writing that Gurley appreciates clever theories. Successful businesses—especially dominant behemoths like Intel—don’t always make obvious strategic choices. But if the operators are good at what they do, the choices will be wily and effective nonetheless.
It’s on investors to reverse-engineer such decision-making processes and make their own determinations as to their predicted efficacy. Here, Gurley seems to engage in this exercise simply because he enjoys this kind of thinking and likely thinks exercising this kind of logic is valuable practice for investors—or at the very least, entertaining.
Until next time,
DS