Sun Announces
New Software Strategy And Pricing
Sun used its NetWork Conference
in San Francisco last week to announce both software and hardware
products. There is so
much software and strategy news that with a few exceptions we’ll
leave the hardware to the UNIX hardware aficionados and devote this
article to trying to make sense out of Sun’s foray into breaking
new ground in software.
Sun started off with the usual
McNealy bravado – we’re doing great.
We’re incredibly innovative.
We have lots of cash ($5.7 billion).
They waited until the end of the event to announce that Sun
was laying off 1,000 of its employees – not usually a sign of how
well things are going.
These announcements are so
important to the future of Sun that everyone took a turn at having
something to say. In
fact, most Sun executives took more than one turn – using both the
podium for keynotes and multiple interviews in the press to try to
get their points across. We
listened – carefully – and we think we understand what Sun
THINKS it’s doing. Unfortunately,
we’re not sure we agree that what they’re expecting will
actually happen. Perhaps
this is an example of that economic principle the Law of Unintended
Consequences.
The Strategy
Sun has been faced, for longer
than it wants to think about, with the hard fact that many customers
prefer to buy their hardware from other sources, often taking
advantage of the lower prices available in the PC market, especially
in combination with Linux, which is much more appealing to customers
than Sun expected.
So Sun has come up with a
multipart strategy:
- Offer Linux boxes but don’t
really care if anyone buys them (Sun certainly isn’t pushing
them; just offering them in case someone has to have them.)
If you want -x86 architecture, Sun thinks you’d be much
better off with Solaris for –x86.
Sun appears to have failed to notice that if you didn’t
want it before there really isn’t any particular reason why
you’d change your mind now.
There was some noise about the SCO law suit making it
clear that you wouldn’t want Linux, but that belies the fact
that customers haven’t made that decision.
- No Linux on servers; that
space belongs to Solaris. I
have trouble thinking of that as a strategy, but it’s
certainly Sun’s. That
means, by the way, if the customer wants Linux servers they are
automatically going to buy from IBM or HP or Dell.
But they were probably going to do that anyway.
- Put all the infrastructure
software together in sets of suites (there are six) and sell
them as integrated sets, upgraded once a quarter, on an annual
per employee license fee. Sun
is hoping that the combination of a bargain price ($100 per
employee per year for the basic stack) will be very appealing.
We have some questions:
- We
understand that this could be interesting for a company that
has relatively few employees but uses its servers to interact with a very large
number of web-based customers (since there is no charge for customer usage).
Sun may gain some real traction with organizations in this particular demographic
– if these customers are willing to give up their current applications.
Our experience is that customers are very reluctant to do that.
Sun’s peculiar pricing would make it tough for
customers to choose its software unless everyone is actually going to
move over to that software
since you pay for every employee – but if the number of
employees was very small (say a few hundred) and the number of customers very large
(hundreds of thousands or millions), paying for unused employee software to get
free customer software might be seen as a very reasonable investment.
We’re assuming Sun’s rules would permit that.
- We wonder if Sun has done any realistic research as to what
percentage of its customers might actually buy this deal.
Jonathan Schwartz, for example, in an interview during
NetWork, commented that Sun’s top 65
customers have enough employees to provide $1 billion in revenue to Sun at $100 a
head for the Java Enterprise software stack.
My guess would be that Sun would be really lucky to get 10% of them onto the Sun stack – but then, we haven’t asked Sun’s customers.
yes". We hope Sun has.
- We suspect that the offer of Java Office desktop software for
$50 per employee
(with the Java Enterprise Stack) or $100 per employee (on
its own) simply ignores the reality of how corporations are organized.
In North America today, Microsoft Office
Suite has a 95% penetration rate in the corporate market.
If we assume that some
percentage of the employees in every company would need to
keep their Microsoft
software (and be supported with software purchases and
administrative support), but nevertheless be included in the employee head count for Sun
licensing, the rationale falls apart pretty fast.
Sun would have been better off pricing their desktop
offering
on a different basis – perhaps on a volume discount, with
sharp discounts for high volumes. That
would have allowed organizations with call centers or other
groups that are candidates for the Sun office desktop to use it
without the need to double pay for unchanging Microsoft desktops.
Outside of North America,
things might be a bit different, especially in places where few
customers are heavily penetrated with office suite software and
this is a new decision rather than a replacement decision.
- Sun is offering its
Development Suite separately.
Again, Java Enterprise buyers get a “bargain,” only
$5 per employee more. Otherwise,
it’s $1895 per seat. We
talked to a few development software providers who told us that
even in the very largest companies it’s unusual to sell more
than a 100 or so development environment licenses.
That means $380,000 in Sun’s terms (although there
would be big discounts from a normal ISV). But Sun wants to be paid for EVERY employee.
That’s a thrifty $5,000 for a company with 1,000
employees, but at their cap (120,000 employees – apparently
everything is free after that), it’s $600,000, even if only
those same hundred employees use the software.
- Our biggest fear for Sun is
that they’ve managed to introduce another step into their
hardware sale process. If
trying to sell the customer a software stack at $100 per
employee (or $150, or whatever) is part of the game, every sale
is going to take longer as the customer thinks about whether he
wants to buy Sun software.
In many cases, the guy who makes hardware decisions and
the one who selects software are different people – imagine
that conversation! Sun
can’t afford to lengthen its hardware sales cycle, especially
if, as we suspect, many of its hardware buyers are going to say,
“Oh, thanks, but we already use xyz.”
We’d guess Sun doesn’t know
this, but in most large companies we know, qualifying new software
at the company level takes at least a year(!)
And that’s if someone wants to buy it and supports it
through the testing and decision making process.
In many companies, the software customers buy is a direct
result of consultant recommendations.
As far as we can tell, there are few consulting firms
recommending the Sun software stack just yet.
They won’t recommend it until they are trained to
competently install it, support it, and customize it.
Such ecosystems take time to build.
This Is Not About “Is It
Great SW?”
All of this without even a
comment on how good the software is because, frankly, that’s
irrelevant. All of
these issues would be problems even if the software was somewhere
between good enough and really quite nice.
It might be able to transcend some of these issues (or get by
them more quickly) if it was absolutely terrific, but all of us know
something: most of this
software has been around for quite a while.
It’s not bad, but it certainly isn’t head-and-shoulders
above everything else in the market.
And it’s not clear that even that would be enough to get
hundreds of customers to pitch out what they’re using and take on
the job of learning new products.
Ask companies that sell software every day for a living how
hard that is. If that weren’t true it would not be possible for
all of us to trade war stories about all the times we’ve been in
Big Companies that are still using 25 year old software instead of
something new and shiny. Ask
all the companies that are three upgrades behind.
Change is hard. Convincing
people to change is harder.
And In Conclusion
We love new pricing models and
bold ideas. In this
case we suspect that while some Sun customers will be very pleased
with the new idea, most will take it under consideration and then
not do much about it.
We’ve been waiting for the
Linux Desktop round to roar. Sun
is a natural player to lead this round, with its ownership of
StarOffice and its promise to offer a server-based desktop.
Sun is choosing to sell its Java Office as a software stack,
similar to what could be assembled in the Open Source market.
Its StarOffice is, of course, identical to the free
OpenOffice.org. There
is a version of MadHatter which marries Sun’s SunRay terminals to
a low-end Sun Solaris Server, but I don’t think that’s a Linux
desktop at all. So I
suspect we’ll need to wait for another market entry to prove
whether this market is ready to roll.
Some customers are already
interested. The
Australian telephone company Telestra has been testing Mad Hatter
and is thinking of trading in its 45,000 desktops for a saving of
$750 million off its IT budget. Telestra intends to use both Microsoft and SunDesktops,
which means this will be a good test of interoperability.
We’re still trying to figure out how he’s going to save
that much money, though. Some
Linux folks think they must have forgotten to count the cost of
administrative support to get to that number.
Meanwhile, maybe Sun needs to get
back to basics:
- Maybe
Linux servers are here to stay.
That’s a customer decision, not a vendor decision. If so, Sun should try to take advantage of it.
- Software
pricing models may need some fine-tuning after Sun sees market
reaction to its first round. It’s
possible that while some customers like a single-price model,
others may
prefer to pay only for the employees who will actually use the
software.
- In
his keynote address, Scott McNealy noted that there are now
more than 3 million Java developers (funny I have a note from
last year where I thought he said 7 million – maybe I
got it wrong?) and that the goal is to drive to 10 million in
the next few years. Java
is no longer a new platform. I’d
guess if it hasn’t gone up the ramp yet it isn’t – or
maybe 3 million (with another million or two to go) is the ramp.
Many are moving on to XML.
In quite a few cases that’s because they don’t want to stick with a
development language that’s not open. And as long as Sun controls it, it isn’t.
So maybe Sun could look one more time, since they don’t make money on Java anyway, and think about moving Java
to an Open Standards body – that, I think, would give it (and maybe Sun) a whole
new lease on life.
- Utility
Computing models look like they’re getting traction with the
customers. Sun
has
decided not to offer Utility Computing directly and their sole
offering so far is with ACS. They’ll need to add additional partners and perhaps even
consider a direct offering.
(back to top)
