Scott McNealy on: Sun's secret weapon and its biggest mistake (not you, Apple)
Interview Hard to imagine, in all the time we've been here the most we've gotten out of Scott is smalltalk about haircuts and parking. The following should help to put that right.
On Monday we had an hour at Sun's HQ in Menlo Park, where the CEO and his executive team have relocated, Sun says, to be closer to their customer center. They haven't moved there for the view, that's for sure. On one side there's a salt flat wilderness that looks like the Russian steppes in winter, with the Dumbarton Bridge disappearing into the fog, while on the other is Silicon Valley's dirty little secret: deprived East Palo Alto, the murder capital of the peninsular. So you're only here if either - like me, you were born and brought up with such a view, or else if you got serious work to do.
Barely a word in that hour deviated from McNealy's insistence that Sun's future is as a systems company, and that Big Systems are precious and are difficult things to do. The one-ness permeated every answer. Now that would be pretty, derr! obvious in saner times only now we've got a bunch of Wall Street smart alecs - asset-strippers, basically - who simply see the world sliced into horizontal segments they can bleed to each individual core. [As we lamented here] This leaves many, if not all computer companies who exist outside the Dell model having to justify their very existence, and high spending R&D companies like Sun have to justify it more than most.
So for you, dear reader, that means hearing a ton of stuff about cars and webtone switches you've heard before. We've focused on the rest, which we hope has some value: observations on Sun's integrator strategy and company history that are bang up to date.
I get the impression you're quite a Romantic in one sense and you like the purity and symmetry of things at Sun, like being able to say you have one SPARC platform that scales -
The customers like that. Go ask them - they like the scalability. I'm not romantic about it - I'm romantic about making money. That's where I get romantic. I'm a capitalist.
But so was Colonel Sanders. You're not in the fast food business and it's hard to think of KFC taking a pride in something like that, which is pretty rare
By the way, we are in the business to make money. The advantage of being a technology company is that we have a much longer view of the discontinuities occurring in the market - so we focus more on the right thing as opposed to today's thing. SPARC is the best 64-bit processor on the market today and customer's have rewarded us with the lion's share of the 64-bit market. We care about doing the right thing because that's the best way to make money and keep customers over the long haul.
Sun is a long-term play.
Linux and Open Source
So was it hard to be convinced you needed a Linux business - is that something you did with a heavy heart?
Not really, I like the open source movement. The amount of Open Source participation and donation and collaborations we do is great and we don't go around bragging about it. We do our charitable giving anonymously.
I think people who take a good hard look will see that. We've been a long time player in community based development from the NFS days and the Berkeley BSD days with Bill Joy. We've been there with open interfaces and a whole bunch of open implementations. It's not like a new deal for us. We're leveraging MySQL, and Ximian desktop, we're pretty excited about that.
It wasn't such a big deal - people are looking for news. It's a piston ring, it's a disc brake; it's a component and it's free: and here are a lot of people out there making it better.
There's a lot of free components. But there aren't a lot of free systems out there.
The trouble is, and have you heard of companies employing their own "Directors of Linux Kernel Engineering"? It reminds me of a DEC story. Ken Olsen used to give a turkey every Christmas to every employee. And when the company got to 70,000 employees someone came to him and said, why don't we have a turkey farm? Why don't we vertically integrate it all?
Well, a bank having a Director of Linux Kernel Engineering is like - I put that in the Turkey Farm folder. It's like me getting into the ATM branch banking business. What's up with that?
So we hear you've set aggressive targets for success in the Linux business, but we don't know what they are. How will you judge success - volume, market share or something more intangible?
It's kinda like a car you get in and you don't know if it's a long haul or a straight six or a v8 - except when you step on the gas, that might go faster.
Well at recent Sun briefings you guys were saying that a third or half of the people buying LX50s were putting Solaris on Intel on them...
Well, you know. I had a very interesting meeting with some Swiss company a year and half ago and they decided not to deploy StarOffice because it's free. They said: it scares us, there's no contract, no agreement.
So we put a price tag on it and all of a sudden people are paying for it. You've seen more activity since we charged for it than you have the whole time it was free. It's a better product. And all of a sudden StarOffice is funding the R&D.
Its own R&D?
Yes, it's not a sinkhole any more. Now people know it's funding its R&D they go well, I'm willing to buy into it because they're not going to shut it down
So we shut down Solaris on x86 at about the same time - and people went nuts on us. Same thing. Now you can download it.
But how will you know when you've done well, or not in the Linux business?
That's like asking Ford how well you've done with four cylinder or six cylinder machines. They don't care about that. They care abut revenue and earnings per share. It's like asking me about the camshaft.
We sell systems. Like Legoland.
I'm not so sure it is.
We sell a system - and if you draw a picture of the datacenter there'll be a switch, a firewall, a loadbalancer, a whole load of edge servers that might be Netra or Wintel or 'Lintel' boxes, over here you might have some app servers, over here's the SAN, over here's the database servers… when you put it all together and metal wrap it it's a big webtone switch.
I was just with a large new partner, a very large Australian telephone company and they don't want dialtone switches. You've seen the org.chart: for voice there are three procurement people, but for data there are ten thousand.
Microsoft and the Law
Are you going to be bothering the Microsoft compliance committee? You haven't commented on the judgement.
The day we announced our private lawsuit, I said I'm not interested in talking about it. We will let an independent judge with a non-political attorney general and a non-political private company pressing the case decide, as opposed to it coming from a politically appointed attorney general . There are no governments involved here - just a judge looking at the laws
So do we have a non-political Attorney General?
I, uh - I know nothing.
It was a reward not a remedy.
When's the last time you spoke to Gates, when did you last have a conversation?
I haven't talked to Gates recently. I talked to Ballmer - we played golf. I was one round behind him - but I didn't hit him.
Were you surprised by the comments on the management changes?
It was noisy at the time but would it have been better to do it all at once - who knows? It all kinda happened naturally and was all part of a plan.
I have an org chart with all the positions penciled in, a year to a year and half ago and I presented it to the board. I said I hope to have it done by July 1.
Was the goal to have more people reporting to you?
Well at Sun we've got down from 19 per cent managers to something like 12 per cent managers, and we've done it. Just to flatten the organization, reduce the overhead; so we've taken a couple of layers out.
Because you weren't getting information fast enough?
Everybody has the information, in fact, I've got more information at my table now than I did before. It's kinda nice to have Yen not reporting to Schumaker not reporting to Ed, to have him right at the table. I have a much better feel for things.
We've all been at Sun more than ten years so we know each other, and we can all finish each other's sentences. And I don't think as you go around talking to people you feel like you're talking to people from different planets.
The Biggest Mistake
What's the best decision you've never made? One months or years later you're really glad you didn't do.
In other words, what was the biggest mistake we avoided?
Well clearly there was a huge, huge war back in the early 90s and we were in Hawaii. It was one of our SunRise clubs for our sales people And it was an 11:1 vote, my staff against me about whether or not we should ship a Wintel computer. And I said no.
And it was practically a mutiny. And I can remember being in this conference room - and I remember walking out to get my composure so I could go back in, thinking I don't get this. I don't understand.
And along the same lines we had some interesting conversations with a fairly large PC company with potentially merging. And I remember saying to the board I'm happy to execute on my merger, but I'll immediately sell all of my stock.
You can't be a PC reseller and a computer creator company in the same company. I think IBM's proven that; HP has proven that; DEC has proven that. The Asians have proven that. And the Europeans have proven that.
Every company that went onto Wintel ultimately hollowed themselves out: it was a self-imposed lobotomy. They're no longer R&D companies and it was massively painful. They were shareholder disasters.
But IBM and HP are still system companies
Well HP's R&D budget is elsewhere and in fact two weeks ago IBM took $1 billion out of product and into services.
So the biggest benefit is what, the focus? Or the freedom of not being a Windows licensee?
Everybody knows we're committed to SPARC and Solaris. It's a lack of ambiguity around Sun -everybody knows SPARC, Solaris and Sun ONE is the priority it's where we're going to invest, and have the best product line. If you look at our x86 strategy it helps Solaris and Sun ONE; our LX50 helps Sun ONE. While reselling the Intel box or the AMD box is no harder for us than anyone else.
Sun was interested in Apple more than once - would you classify that as a mistake avoided?
We could have survived merging with Apple, I don't think we could have survived merging with, or adopting, the Wintel architecture.
Men In Suits
You've invested heavily in China. Can you take money out of the PRC?
The government there provides tax incentives for foreign companies to re-invest their profits in China, but they also have regulations allowing repatriation of profits made by Wholly Foreign Owned Companies like Sun. We plan to expand our presence and do a lot of business there.
Now you're recruiting more integrators - something you've left to the channel - how's that going to work without hollowing out their businesses?
We'll definitely partner with them. We'll partner with whoever's competing with Microsoft. Microsoft wants to own all the Service Providers, and directories, they want to be in the subscriber management business, we don't. IBM wants to be in hosting and services, we don't .
But you have to leave something for the partners to do…
Which partners do you mean?
I mean the Cap Geminis.
Oh, there's tons for them to go do. Where the dialtone analogy falls down is that dialtone switches only have a few services: caller ID, voice messages, and that's about it. Think of all the webtone features - let's put a directory into GE, for example, and keep it going in real-time. That's not the business we're in, but that's where an Accenture or an EDS can come in and do the business process re-engineering.
But that's a wonderful partnership. What they won't have to do is make sure that the app server talks to the portal which talks to the storage which talks to the clustering file system which talks to the microprocesser. We will take care of all of that.
By the way, there are still third party pieces we don't have, or that will be customized for each market. We don't have all of the lego blocks. IBM wants to do it all.
Now, when I was about so high, I remember asking what shares where, and part of the answer was that you got rewarded each year if the company was doing well with a dividend. Tech companies don't do this. Why don't you give your shareholders dividends? Why do you do buybacks?
We buy Sun stock because we see it as a good investment. It's that simple. As for paying dividends, we'd only do that if we ran out of ideas and couldn't figure out how to reinvest the money in producing innovative new products and services that increase shareholder value. In other words: it's Not going to happen.
Itanic vs Hammer
I'm curious that Sun is still getting toasted for ignoring Itanic when its impact after all these years has been marginal and Hammer seems to be much more of a direct challenge to your SPARC business. Isn't Opteron more of a threat?
Probably. I agree it's more of a potential threat. Is it a threat that's damaging at this point? No. And by the way, with Solaris running on AMD running the SunONE that's good for us.
Hammer isn't going to fill out all these pieces anyway - there's going to be vertical scaling and horizontal scaling and bandwidth scaling machines and that's our secret weapon.
But for a new chip to blow us out of the water, it's got to be already certified, already tested, already scaled and already branded - and for that you need a 2x price/performance advantage.
The argument is we're already there - it isn't the chip that's expensive it's all the other R&D.
Shane Robison [Compaq, now HP CTO] told me that what killed Alpha was the chipset expense for each iteration
It's not the chipsets - they didn't have the ISVs.
Business Week reckoned that SPARC and Solaris cost $200 million a year. Is that an accurate figure?
We haven't broken out our R&D budget. I can't comment.
So by way of introduction to some more car stuff - what car do you drive?
It depends - we've got a Yuko, a PT Cruiser, and a Chrysler and a Mercedes.
It's kinda retro. It's nice, I've got four boys, you've got two or three of them in the PT cruiser it's what you've got. The Mercedes is for when you don't have any of them, heh.
The car point is this: when I first came to America in the early 80s, I was 16, and every car on the roads was an American-built car. Now half the cars are American. At some point the US car industry just forgot how to build cars Americans wanted.
And a problem for all you systems companies is that people have enough already - there's so much overproduction. There's two things there: lack of demand and oversupply, so how is any systems company going to persuade people to buy your stuff?
Well, I still think people don't want to have a nuclear power plant in their basement, they don't want to have a well in their backyard and they don't want to have a telephone switch in their closet. They want to get the utilities… they probably want out over the network and that's happening.
Think back to when the browser came along. No, back to 1980. Then 100 per cent of the mips you consumed were owned by your company. If you were a working stiff like me, I worked at FMC and they had a big CAD computer and a big mainframe. And then along came the PC, and then 99per cent of the mips were owned directly by your company or your personal computer.
Along came the browser - and now think about how many mips are generated at a networked company: at an eBay, or an AOL, or an SBC or The Register: I would say that when you take out mail, 98 per cent of the mips you consume are their mips. By definition in the browser you're using utility mips. That plays in our favor - not to the PC mainframe model .
I think the market is coming to us. That's why the minicomputer makers are gone, that's why the PC makers are going; the growth is in the modern clients. We like our position.®
Sponsored: RAID: End of an era?