"This year, 1999, is our year of piloting." Exactly! Pilot-testing this year, rollout next, and you don't get clobbered -- either in the marketplace or in the market.

That's what Michael Armstrong, CEO of

AT&T

(T) - Get Report

, which I am long, said this morning in the analysts' conference call on the AT&T-

Microsoft

(MSFT) - Get Report

deal.

Now the country's largest cable company, AT&T not only has to manage those cable franchises -- hardly its biggest problem, or its biggest priority -- but much more importantly, it must move quickly to prove it knows what it's doing in building, packaging and selling its telephone services over those cable lines into America's homes.

The $50 billion-plus

TCI

deal last year, now finalized, has brought skepticism and frustration among former TCI cable customers, who were already unhappy with TCI's slow, spotty and often unreliable fast access services, especially in high-profile parts of the San Francisco Bay Area. TCI promised a lot, delivered a little, then in effect dumped the problems into Telephone's lap.

Now, with another $50 billion-plus set to be invested in

MediaOne

(UMG)

cable systems as a result of this week's deal-making round, plus another $2 billion for the

Lenfest

cable systems in the Philadelphia area, plus the side deal to sell local phone service over

Comcast's

(CMCSA) - Get Report

cable systems, AT&T is in so deep it simply has to make this one-wire/one-provider notion work.

If, as promised, AT&T can sell us a wide range of services, all digital, over a single drop into our homes -- plus maybe Digital One Rate cellular service as well, and let's not forget long-distance! -- monthly per-subscriber bills can shoot through the roof. Which is exactly what AT&T had better deliver, to begin to amortize these purchases in your lifetime and mine.

Big problems lie ahead. I'm an Armstrong fan, and as both a holder and admirer of AT&T in general, I believe those problems can be solved. But with the cable industry's standards and technologies for delivering Net access, and voice, over cable wiring, still so messy and in some ways unproven, AT&T has to deliver on the operational level with a speed and quality of execution that is simply breathtaking to consider.

Anything short of getting it really right will lead to a pillorying on Wall Street for Armstrong and T ... and the company knows it. So "the year of piloting" idea is a brilliant strategy, and a great public stance. This is the year, AT&T says, when it will try things out, set up small initiatives to work out the kinks and get things right before rolling out big and expensive mistakes.

Tough, tough job. And a kind of test of corporate character, as well as operational and technical savvy, for Telephone.

Clearly the market thinks this can be done: AT&T was up 5 yesterday, and almost 4 so far today, to a little over 60 in late morning.

Media king? We'd better hope so.

What Might Have Been at National Semi

Shed a tear, maybe, and say goodbye to what was once a source of real innovation in the PC business, as

National Semiconductor

(NSM)

yesterday announced it's finally bailing on the PC microprocessor market. Its expensive purchase of Cyrix in mid-1997 was its final lunge at success in PC chips. Cyrix has done fairly well at the bottom of the personal computer CPU market, and low-end boxmakers

Microworkz

,

eMachines

and

Packard Bell

will have to scramble to find replacements for the

Cyrix

chips in their current bottom-of-the-bottom-of-the-line machines.

But they won't have to look far: What finally squeezed the life out of the Cyrix line was

Intel's

(INTC) - Get Report

aggressive marketing of its

Celeron

chips, plus

AMD's

(AMD) - Get Report

success with its

K6

series.

Those of us with long memories and longer histories in this business recall National Semi fondly for its early failures -- I'm not being sarcastic -- in the PC CPU business. It is not an accident of history that Intel won that war -- Intel brought focus, intensity, speed and the willingness to gamble to bear upon what at the time did not look remotely like today's huge, rich personal computer market. But if history were written on the basis of technical innovation, National Semi's 32032 chip would have been one of the big, big winners. When Intel was fooling around with such dead-end products as the 80286, National was jumping way ahead with more far-seeing, more innovative products.

Then again, history is written by the winners, and National Semi was never able to persuade an ever-cautious, me-too PC-clone industry that technological superiority was a step toward market success. National became, along with AMD, a classic heartbreaker for technology investors.

So National bails on PC CPUs, shuts down Cyrix, and puts the name, intellectual property and its South Portland, Maine, high-tech chip fab up for sale. Sooner or later you've gotta face the music and stop losing money, and from an investor's standpoint, this is a very positive step.

But there's still a tear or two for those burdened by memories of what might have been.

Jim Seymour is president of Seymour Group, an information-strategies consulting firm working with corporate clients in the U.S., Europe and Asia, and a longtime columnist for PC Magazine. Under no circumstances does the information in this column represent a recommendation to buy or sell stocks. At time of publication, Seymour was long AT&T, although positions can change at any time. Seymour does not write about companies that are consulting clients of Seymour Group, or have been in recent years. While Seymour cannot provide investment advice or recommendations, he invites your feedback at

jseymour@thestreet.com.