Betting the company
Wily wrote:
(quoting Grove from the "Wired" interview)
"Stagnation. Lack of growth. Being restricted to a market that's further along the S curve than the communications market; a market that in many segments is more of a replacement market than a new penetration market. The risk then is that the scrappiness of the company would go away, the vitality would go away. When there is an opportunity that you can pursue, if you choose not to pursue it you are choosing a different corporate persona. It is entirely possible that the microprocessor business will not have a satisfactory growth rate in the years to come. We could accept that, we could choose to be a 10-percent-a-year company, we could adjust our spending and concentrate on being a very good 10-percent-a-year company. But we would not be the same Intel anymore."
Someone once said that every majorly successful company has "bet the company" at least three times. First when it was founded, then a second time when a major new product was introduced or a major new market entered, and then a third time to ensure catapulting to the top ranks. Boeing, for example, bet the company when it was founded (not interesting, as all companies make at least this first bet), then bet the company again in the 1950s with the Boeing 707 entry into the high-volume passenger market, then did it again ten years later with the very risky 747 (contemporaneous with a decision NOT to pursue work on an SST).
There are sound reasons for this "betting the company" model. And risks. Companies which bet and then fail tend to either vanish or be absorbed or linger as shadows of themselves. The bets may not always be obvious ones. And they may be "smeared" in time in way that make them harder to spot. But the point seems to be a valid one, that the largest and most successful companies tend to make several of these bets.
IBM made its second bet, I would argue, with the 360 family. (Various smaller bets in the decades prior.) Its third bet? Probably the move into PCs and smaller systems, going against the "big iron" culture within the company. While not always successful, and not always profitable, their dominance in the public perception ("the IBM PC") let them escape the fate of the other Big Seven members like Univac and Burroughs and even DEC. So far, I don't think they've made a _fourth_ big bet.
Many such examples abound. Cutting to the chase, what such "bettting the company" events have happened with Intel? I'll make my own guesses/estimates. No guarantee they are the right ones. Others should jump in with corrections or time line changes.
First Bet: as always, the founding of the company. Noyce, Moore, and Company bet that their experiences from Fairchild, and their work on MOS and silicon gate MOS could be leveraged into a new company. The bet was not necessarily on RAMs or processors.
Second Bet: It is very tempting to list the move into microprocessors as the second major bet. But I think the evolution was so gradual, and played on the strengths of Si-gate MOS so obviously, that this was not a "bet the company" event.
What was, then?
Intel pioneered several markets in the early to mid-70s: UV-erasable PROMs, Single Board Computers, Development Systems (for designing with its micros), and various flavors of memory chips. Watch chips, too.
One candidate would be the development of world-class design tools. By the late 70s, Intel had CAD such as no other chip company had. (You all know, I presume, that the x86 chips were laid-out on Applicons and Calmas, while Motorola had the entire 68000 done with "Rubylith and X-Acto knives"? This had, apparently, a major effect on Intel getting the IBM PC contract over the architecturally cleaner 68000.)
Another candidate for a bet which almost _sunk_ the company was the 432 project, in terms of resources spent on it and in terms of the loss of focus on the continuity of the existing product line. (When delays looked unavoidable, a rush job to design a 16-bit version of the 8080-family was mounted. Had this failed, or had IBM not picked the x86, imagine where Intel would have been?)
And there's the switch to CMOS...except everyone in the world was doing it at about the same time, so it was not exactly a "bet" in the sense described here.
I would combine the events swirling around 1977-82 into the Second Bet: Intel's decision to become a leading architect of microprocessors, with uPs not just a way to use fab capacity and not just another product in the product catalog. This mean architectural experimentation (432, iPSC, iWARP, etc.), this meant much more sophisticated design tools, and this meant the need to achieve very high yields on very large die sizes.
On to the Third Bet.
Was getting out of DRAMs the Third Bet? I don't think so, but I am willing to be persuaded that it was. I think getting out of DRAMs, as shocking as it was to some of us at the time, was a consequence of this move above. And Barrett pushed the manufacturing group (the fabs) to obtain the extremely high yields on complex chips WITHOUT having high-volume RAMs as "process drivers." (A lot of us, including me, doubted it could be done.)
So what was, or will be, the Third Bet?
(The obvious issue being whether a move into "the communications space" (!) is such a bet.)
And I cannot resist asking the relevant question right now: what bets has AMD made? (After AMD's first bet, I'd say the Second Bet was to become a second-source for Intel chips. They abandoned the 29000 and basically became just a second-source. This has worked for many companies, so it looks to have been a good bet. What will be _their_ Third Bet?)
I'll defer more discussion to the next post. Post your own ideas, too!
--Tim May |
