Take a look at today's Barrons.
barrons.com
Requires free registration. For those of you who don't want to
register here are some excerpts:
Visa and MasterCard, for example, can't issue credit cards with expiration dates
beyond 1999.
Q: They can't?
A: No. In one case a bank sent out a bunch of cards with ``00'' in the date field
and they all bombed. It was a big embarrassment. The banks' card readers and the
software systems weren't ready. That's just one example - and there are many.
Drivers' licenses. We've had defense contracts screwed up - one contractor got a
97-year delinquency notice on a contract that's to be completed in the year 2000.
Q: Back up. Why is it that computers choke on something so simple?
A: Essentially, for efficiency and expense reasons, software programmers have for
decades used two-digit year formats to represent date fields. So 1997 is written as
``97,'' and the software simply assumes that the first two digits are ``19.'' Come
Jan. 1, 2000, however, it's predicted that all hell will break loose as computers see
double zeros in the date field and applications crash or malfunction. In fact, with
computer technology permeating our lives, the doomsayers are in overdrive. But
while some of the imagined repercussions may be a bit melodramatic, the Y2K
problem is very real. And as we get closer to 2000, it's only going to get worse.
The question is: How much will it cost to fix? And what impact will the Y2K bug
have on the systems that, inevitably, won't be corrected in time? Gartner Group
claims that 90% of all computer applications will be affected and that it will cost
most major corporations an average of $50 million-$100 million to solve it.
Total fix estimates are all over the place. International Data Corp. says it could cost
the U.S. alone $100 billion. Other estimates go as high as $1.5 trillion globally. But
whatever the number, the cost will be huge. Then, too, there are some who fear
that this problem has been ignored for too long. META Group, the research house,
for one, is saying that it's impossible to get all of the systems corrected in time
because so little time is left to do the work. Longtime computer software efficiency
guru Capers Jones, chairman of Software Productivity Research, believes it's too
late to scrap and replace major legacy applications - a process that generally takes
two years or more. The experts agree that such changes must be made by Jan. 1,
1999, to allow enough time for testing and debugging. In the March issue of
Datamation magazine, Jones calls Y2K ``one of the most expensive problems in
human history.'' His estimate of the tab in the U.S.: $276 billion.
Q: Come on. People didn't just wake up
yesterday and discover the millennium,
and this bug, around the corner.
A: Nobody in the technology world was not
aware of this. Some people have known
about this for 20-30 years. When the first
code was written, they knew it was a
problem. But because of the enormous cost
and complexity of fixing it, most
organizations ignored it. I know what that
procrastination was like, because I was
involved in it. Almost 10 years ago, when I
was working in the industry, I had oversight
responsibility for a major information system
project that was very similar. We had a
contract number code that affected every
one of our company's critical software systems - just like this year 2000 issue. The
code field was no longer large enough because the company was about to
dramatically expand, via a merger. And the solution, in that case, was the same.
We had to go in and identify all the software systems affected and find all the
instances where we used the contract codes. Then we had to expand the fields to
correct them and test the corrected systems extensively before bringing them up
live. It was a massive, exceedingly unglamorous, labor-intensive effort to comb
through every line of code - and fraught with risk. As we were going through that
process, we discussed the Y2K problem (``maybe we should do it now''), since
the programmers were already scanning all the code, line by line. But ultimately, it
came down to time and cost constraints, so we said: ``Well, it is kind of early to do
it. Maybe these systems will go away by then. Maybe we will find a magic bullet.''
Q: Okay. We'll grant that some software has to be changed. But why should
adding two little numbers to dates be such a big, expensive deal?
A: On the surface, it sounds easy. You go in, you correct the code. But that
assumes you can find the date fields - which are not identified in any systematic
way.
Q: Isn't that precisely what the Y2K tools have been designed to do?
A: Yes, software tools can be helpful, especially in identifying date fields. But there
are no magic automated software solutions that can be used to correct all those
problems, as Fred Schuff explained very well in the March edition of Enterprise
Systems Journal. Schuff, who's been designing and developing software for more
than 20 years, points out, among other things, that when you go into old software
and crack codes, you can do all kinds of insidious things. A lot of the logic in
software indirectly uses date calculations. A lot of programmers were quite
creative, whimsical or even mischievous in the ways they wrote code. And they're
long since gone, generally. So it is not simple. It's inevitable that, experts say,
maybe 10% of fixes will be bad, and there'll be some degradation of application
throughput and data-center efficiency in fixed systems. Also, there is a certain
percentage of the code - some of the software gurus, like Jones, say 10% or 20%
- that can never be fixed by any automated methodology. Using ``factory''
solutions, a certain percentage of the code will be identified falsely as needing
correction or be missed. So in mission-critical systems, you can't afford not to have
eyeballs going through your programs line by line. There's also the liability issue,
which is just starting to come into focus now that a proposed tobacco settlement
has been hammered out. The lawyers are getting all excited about this potential
problem, talking about it being better than asbestos. There was a conference in
London recently where some members of the plaintiff's bar estimated there could
be a trillion dollars in lawsuits in the U.S. alone. So, the legal industry is gearing up.
Q: Didn't we see a piece in The Wall Street Journal recently about a software
industry pioneer coming out of retirement with a way to solve Y2K problems
rather slickly at the binary machine-code level, instead of messing around
with billions of lines of application code?
A: I saw that, too. There actually are a number of ``bridging solutions'' that use
certain logic to avoid the issue without making fixes. That particular one is untested
and unproven. Besides, it's highly risky to meddle with binary code. In the
corporate world, it's not something people would want to rely on, necessarily.
Q: Still, isn't this a problem only in ``legacy systems'' that use old
mainframe-based application software that probably should have been
upgraded long ago? In fact, isn't Y2K spurring their replacement by new
client/server systems?
A: Yes, it's primarily a mainframe problem. However, it's not just mainframes. It's
more widespread than people realize. As IBM's chief scientist on year 2000 has
said, ``The more you know about the issue, the worse it looks.'' They've found that
it effects even embedded systems - the microprocessors now in almost everything
from traffic lights to telephone systems to tractors that use software with data fields.
Most PCs manufactured before this year aren't Y2K-compliant.
It also affects a lot of client/server software. For example, not its database
software, but Oracle's applications software had been going out, until very
recently, not year-2000-compliant. Oracle is telling people they must upgrade.
They aren't going to offer fixes for these ``older'' client/server applications.
Q: A great sales tool.
A: It is. It may be one of the reasons Oracle's application software sales were up
70% last quarter and everything else was weak. That seemed to save their bacon.
PeopleSoft, meanwhile, has said that 30% of their business is from people who are
upgrading systems to get away from the year 2000 problem. So the earnings at
client/server software companies - PeopleSoft, SAP, Baan and Oracle - are good
and their stocks are out of this world. You have Baan at 202 times trailing earnings
and PeopleSoft at 135. But how long can companies continue to buy their systems
and still have enough time to install and test them before it's too late? So, while the
client/server software segment is currently benefiting from an accelerated buying
cycle due to the looming Y2K problem, the closer we get to 2000, the less likely it
is that those system upgrades can be finished in time. Oracle's ads actually ask,
``But is there enough time to adequately design, convert and test before January
2000?'' Their reply is: ``The answer is yes, if you start now.'' But that implies that if
you don't start now, you are in big trouble. So, as we go into 1998, there's going to
be a question, which these companies' triple-digit P/E ratios don't reflect. Likewise,
the Dells and the Compaqs of the world - some of the leaders in server hardware
- have been doing better than the regular PC makers in a highly competitive market.
Possibly because they, too, are being pulled along by people replacing mainframes.
So they, too, may be at risk.
Q: Which is why you're calling the Y2K problem a time bomb?
A: It goes beyond the client/server market. While all the analysts have focused on
the potential beneficiaries of the Y2K solution, I'm worried about all the losers. I
believe the computer hardware, peripherals and software industries may be at risk
of a significant slowdown due to purchase deferrals, as companies address the
massive Y2K problem over the next two to three years. And if the computer
hardware market slows, it's inevitable that the semiconductor and
semiconductor-equipment industries will follow.
Q: Y2K will cost so much to fix that companies will slam the brakes on other
technology spending?
A: Exactly. Managements today concentrate on the bottom line as never before -
probably because so many have been quietly raping their shareholders with
outrageous options awards to themselves (no one seems to care as long as the
stock market keeps rising). The game is to inflate short-term profits, which inflates
stock prices and makes the CEOs instant multimillionaires. Meanwhile, it's
estimated that 60% of the cost of Y2K fixes will fall in 1997 and 1998. So is it
likely, for example, that Coca-Cola's chairman, Roberto Goizueta, will authorize a
mammoth increase in his company's information technology budget to deal with
Y2K, thus lowering its earnings and cratering the stock? I doubt it, considering that
his 1995 options award will bring him $36 million if the stock rises just 5% a year,
and $91 million if it rises 10%.
Will our great budget-balancing President pony up the additional $30 billion (per
the Gartner Group) estimated cost to fix federal systems? We already know the
answer. The Office of Management and Budget is lowballing its Y2K estimate
($2.8 billion for fiscal year 1998) and government IT professionals are screaming
like stuck pigs. The chief of the OMB's Inflation Policy and Technology Branch,
Bruce McConnell, told Government Computer News: ``There will be no new
money to fund year 2000 fixes.''
The GAO has estimated that fewer than 25% of state and local government
computer systems will be ready by 2000.
So, over the next two to three years, a panic will sweep government and industry.
It will be too late to scrap the old legacy systems; they'll have to be patched. And
with shortages of programming labor, the fixes will be costly.
Q: And that money will come out of other tech budgets?
A: The answer is yes in all the articles I've been reading. Bruce Hall, a VP of
marketing for the Y2K project at Trigent Software, told Federal Computer Week
that inadequate central funding will lead to a reallocation of money from a range of
IT projects. ``There will be a giant sucking sound . . . that's going to pull from all
corners of IT.'' Barry Ingram, VP and chief technology officer for Electronic Data
Systems' government-services group, told that same publication that operations
and maintenance budgets have fallen in recent years and money for less important
systems has already been pulled. As a result, Y2K costs will ``pull dollars out of
critical systems.''
Q: And the same goes in the private sector?
A: Well, Chase Manhattan (with 200 million lines of code) expects to spend
$200-$250 million over three years from funds in its IT budget. Union Pacific
claims that a number of its new client/server projects have been postponed while it
deals with Y2K. When the Society of Information Management surveyed IT
professionals on Y2K funding, almost 75% believed the money would mostly come
from their regular IT budgets. The report warned, ``Something else will be
sacrificed, and there will be consequences,'' and it estimated that the costs would
absorb nearly 25% of IT budgets. J.P. Morgan, meanwhile, has forecast that
Y2K will sap 50% of technology budgets from now until 2000.
Q: But the money will still be going somewhere in the tech sector.
A: The dislocations could be enormous - and all that spending won't be buying
productivity enhancements. According to Commerce Department figures, business
spending on computers has jumped by almost 45% over the past three years. The
reasons: a corporate upgrade cycle that began in mid-1992, a rebounding economy
and acceptance of graphics-based Windows. But last year, we saw PC growth
slow at both the corporate and consumer levels. Personal computer sales growth,
according to IDC, was just 11% worldwide in the fourth quarter - the lowest level
since the 1989-1991 recession. The reason, I think, is that corporations are
saturated with technology after years of above-trend computer spending - many are
even loudly complaining about their inability to absorb it all. And now they're facing
their biggest one-time technology cost in history. What's most amazing is that Wall
Street is completely clueless about this threat. Of course, most analysts aren't even
aware of the current slowdown in corporate PC spending that corporate resellers
like CompuCom, Vanstar, Intelligent Electronics and MicroAge have recently
warned about.
Q: The distributors aren't talking saturation.
A: No. I sit in on pretty much all the corporate resellers' conference calls with
analysts and they've all been blaming a number of issues for the weakness in their
businesses: a ``pause'' in corporate buying. Intel's confusion of processors and the
enormous cost of upgrades. But whatever it is, they're seeing a slowdown, which
they didn't expect. This should be a strong period - with the economy good and
product cycles favorable. You have Windows NT 4.0 shipping. There was
supposed to be a big upgrade cycle to 32-bit computing. Intel has shipped all of
their MMX chips, their graphics-based stuff, the Pentium IIs, in the last several
months. Yet we're seeing quarter after quarter slower than the prior one at virtually
all technology companies - with the possible exception of Dell. Hewlett-
Packard's order rates are at the lowest levels we've seen in 10 years. There are
problems at Intel, Seagate, virtually every PC component vendor - from graphics
to motherboards to microprocessors. There's been a collapse in DRAM prices
again. There's a fantastic PC price war going on. Compaq is going to crush Dell,
and Dell is going to crush Gateway and HP is going to crush Dell and Compaq
back. We are seeing a sub-$1,000 PC market explode here. So it's really ugly out
there in what should be the best of times. And no one seems to want to ascribe this
slowdown in corporate PC spending to deferrals because of the costs of
addressing Y2K - because that would mean it isn't a short-term problem, but a
multi-year one.
Q: But companies, not just
consultants, are now issuing
eye-popping estimates of what the
fixes will cost.
A: Well, Merrill Lynch says their
problem is $200 million. EDS says it
will cost them $144 million.
Prudential says $100 million.
Hughes says $125 million. We have
companies that are actually now
pre-announcing that their earnings
will be impacted because their IT
budgets have been completely
overwhelmed by Y2K. Southern
New England Telecom, for one.
The numbers are significant - and
investors have so far viewed all that
spending as an opportunity. There
hasn't been any focus on the other
side - that people are spending
money they may not have in their
budgets. Or that there will be tradeoffs. That's the big risk: a real long-term decline
in technology product growth rates that are already weakening. In that case, what
happens to the valuation levels of the Intels, the Microsofts, the Dells, the
Gateways, the Compaqs? We've seen what happens to tech companies during
economic slowdowns - they get slaughtered.
Q: Clearly, though, some companies will be raking in all that dough.
A: Sure, but the real winners won't be the software or tool vendors. The year 2000
correction is a labor issue. You need COBOL programmers to go through code,
make the corrections, do the testing.
Q: People, ironically, with skills that had been thought obsolete.
A: Right. You're talking 30%-50% increases in their pay scales over the last year. I
have seen $25,000-$30,000 signing bonuses. Retention bonuses. There are people
abandoning government jobs because, of course, it can't offer these great
incentives, like the corporate world can. These are the winners. Also the companies
that happen to have a large number of these people. But their stocks also reflect
that, unfortunately.
The problem is that there is a date-certain end point to their business. Take, for
example, Acceler8, one of the highflyers at the end of last year. They haven't
broken $1 million in quarterly revenues yet. Their last quarter was $716,182. But it
still has a $150 million market cap. Its earnings in that quarter were $387,129.
Now, there aren't that many quarters left between now and the year 2000. After
that, the business starts to tail off. So how do you rationalize a $150 million market
cap? You know, Viasoft has a great tool set. But with an $943 million market cap,
I don't know. I expect these stocks to continue to go wild - especially at year-end,
when people focus on the calendar. But they aren't valuation stories, that's for sure.
Q: There's no hope of a silver bullet?
A: No. That was the hype on Matridigm, but once they were actually testing, they
found out that it didn't quite work as advertised. Or, at least, it hasn't, to date. Like
I said, if you appreciate the complexity of the problem, it doesn't seem possible that
there could be a single solution.
Q: How about the big computer service providers? Isn't Y2K a boon for
them?
A: The traditional service providers, like EDS and IBM, don't have a huge
backlog of legacy contracts. Even so, their costs might be greater than what they'll
ever earn on this, because - on the legacy contracts they do have - they'll have to
fix code for free. EDS is hoping to generate enough new Y2K business to offset
that. But they may have some liability issues. So understand what you're getting
into. Now, EDS is relatively cheap because it has gotten hit in earnings. Potentially,
if they get their ship turned around, there is some upside. Companies like
Computer Horizons or Keane don't have that kind of liability because they don't
have the legacy contracts. And they've been generating a lot of new business in
Y2K. Keane has over 250 backlogged Y2K projects, representing future revenues
of $300 million. They've been around for a long time. The problem is that the P/E is
61. At Computer Horizons, the P/E is 80-something. Viasoft has a tool, and an
astronomical P/E. They will all do well, if you're a momentum player; I am not. The
earnings will keep coming through the year 2000, it looks like.
If you're a value guy, there are few pure plays. There's a company called Intersolv
that has a real business. It's a configuration software provider with a product called
Factory 2000 that's very good. And it isn't strictly a mainframe software company
(most of which were failed companies that all of a sudden have found riches).
Intersolv is a leader in the NT world. Its earnings have been rocky, so it sells at
only one times sales, a market cap of $180 million. Its Y2K tools and services
were released in April. It has just started signing contracts. So you have a company
that will likely be in existence after the year 2000 and isn't real overpriced.
Another is Sterling Software. It's an old-time mainframe company, but the
attraction here is it has a lot of cash. They also have year 2000 capabilities, which
they've tied in with James Martin's widely noted Y2K tools. It's not a moonshot
pure-play tools business. But if you have to own a Y2K stock, and you want to
participate in a chicken's way, you can do it in Sterling, where the valuation isn't too
high.
Q: What are you shorting here?
A: It's a very dangerous market to short, even for professionals. We have reached
a new level of lunacy, where even pre-announcements of companies missing
numbers by fantastic amounts don't seem to matter. In fact, they seem to energize
the bulls. Gateway 2000 made a horrendous pre-announcement. The estimates
were 45 cents and they said they're going to come in around 36 cents
(split-adjusted). Meaning almost no earnings growth versus last year, down from a
50% growth rate before. And today, it's one point off its all-time high - in the
middle of a price war and slowing PC sales. It's dumbfounding. So I've been
shorting only in very limited amounts. I've been buying more put options, LEAPs or
some of the longer ones that give me time and limit my exposure. But that's not for
everyone, either, because you could lose all your investment. My suggestion to
individuals is that the place to be, if you're a value investor and believe we will get
back to that kind of market again, is in cash. Historically, a 6% return isn't bad. It
isn't what you've gotten out of Dell in the last year or so. But it also isn't what
you've gotten from being in PBHG. The technology market has been pretty
dangerous, bifurcated. The big names keep going higher, and the little names just sit
there.
Q: Okay, caveats acknowledged. What are you betting against?
A: The Gateways and Dells, the PC makers. They're tremendously overvalued.
Another commodity play on the short side is Micron Technology. The DRAM
business is in trouble. Micron is facing a transition to 64 Mbits. Electronic Buyers
News in late June noted that DRAM bit growth has slowed to 50% from prior
70% levels - which is well below many analysts' forecasts. And 16 Mbit DRAM
prices have fallen $1 in the last month to almost record lows. There is a tremendous
oversupply and their earnings are going to decline. So the valuation, which is pretty
high, won't be sustained. In Gateway's case, they're slugging it out with Dell, IBM
and Hewlett-Packard. That price war isn't going to get any better. And Dell, I'd bet
against just on valuation - at over two times sales. PC companies don't usually sell
beyond one times sales. And the stock has rocketed from 22 to 125 in the past
year. Granted, Dell hasn't shown fundamental problems. They have reported
consistently strong growth. But the Y2K issue is going to accelerate the price war.
So even if Dell remains a successful PC company, its valuation can't be sustained.
It's unheard of to have a $20-something-billion market cap on a PC company with
a limited product line and limited distribution.
Q:Thanks, Fred.
TABLE OF CONTENTS | WEEKDAY EXTRA | COMPANY INDEX | ARCHIVES | QUOTES & CHARTS
MARKET DAY | MARKET LAB | CORPORATE LINK CENTER | MARKETPLACE | HELP
Copyright c 1997 - Barron's Online, All Rights Reserved
Two World Trade Center - 18th Floor
New York, NY 10048
We welcome your comments, questions and feedback. |
