Alphacide yesterday, Yamhill tomorrow: HP merger architect talks
If the name Shane Robison doesn't mean much to you, then take heart, because it had barely figured on our radar until yesterday.
Robison has been described as "the man behind the curtain" in the SirCam merger by people we trust, and the HP CTO was instrumental in bringing HP and Compaq together - at one stage, he was one of only four people in the loop.
It was by accident then that we found ourselves at lunch with Shane, and HP's head of labs Dick Lampman, with whom we'd earlier been chatting about MRAM and P2P stuff, and CodeCon, astonishingly free from interruptions. That had been good, but this was going to get a lot better. After a morning of bland corporate videos in which HP executives had strained to find a reason for the SirCam Merger other than "hey, we're really big now!" - failing to find a reason to stir the soul, other than satisfying Wall Street's asset strippers - Robison offered a glimpse for the rest of us why HPQ might need to exist.
For a start, he can play the part of a tech villain from central casting: a vaudeville baddy so malevolent that the audience starts hissing as soon as he comes on stage. Kitted out like a Southern plantation dandy, Robison can put out a small house fire with a glare. And having run AT&T's research division, and endured seven years as a VP at Apple - again looking after research - he's clearly damn smart and knows how to survive and prosper in the games of executive politics. You'd want to hire him as an assassin… then hide all the cutlery. So folks, you can see that this is beginning to look very promising…
Robison was CTO at Compaq, so let him tell the story.
"We were evaluating all the options for consolidation for some time. We began talking in May, and I think you'll find this in the SEC filings."
And who was involved?
"Me and Mike from Compaq. There was no need to get anyone else involved at that stage."
But Compaq wasn't talking about "consolidation" to anyone else during these negotiations?
"No. And this was not a negotiation - these were discussions."
And who was involved at HP?
"Carly… and Duane [Zitzner]."
OK, so we know the chronology of the Alphacide fairly well. What went first?
"Internally, it was justifying what made sense. This was the first question I asked Mike when I took over as CTO. [1999 - ed.] How can we justify this?"
"As soon as Alpha stopped having twice the performance advantage, it was 'how can we justify the expense of microprocessor design?'… Don't forget we had three roadmaps, three speed bumps going…. You can't just do it once - you have to do it forever. How could we justify the tools, the compilers?"
So how hard did Compaq then try to license the Alpha, we asked. Samsung and API had been licensees, and SPARC and PA-RISC had some OEMs too.
"We could have licensed it out, but we didn't have the market footprint. We would have been perfectly happy to," said Shane.
Despite Itanic's dismal traction, HP doesn't think the cost of designing chips is worth the expense: and it doesn't think IBM and Sun has met the challenge, just yet.
"You'll see Sun selling a lot more x86 than they are now in five years time." And that goes for IBM, too, he added.
Robison and Lampman re-emphasized the point SRCAM has made throughout the proxy fight: it will invest in chipsets to differentiate itself, but not chips.
"We'll see HP IP in the software and the tools," said Dick. The idea is to use these to eliminate management and services in the enterprise. "We'll still compete on speeds and feeds, but at a different level… we're not in this perpetual horserace about who can make the fastest microprocessor."
"Sun's in a lot of trouble," said Robison.
America, phone home
We'd begun by talking about the phone business. HPQ had made much of wireless, but seems to have missed the wave with smartphones. Shane didn't buy our proposition that as a platform provider, Microsoft has missed its chance, and knows it.
"It's too early to call a winner," he said.
But with devices such as the Sony/Ericsson P800 and the Nokia 7650 taking imaging down to the pocket, wasn't this a challenge? Didn't HP at least know how to stick a camera on a phone, and sell it?
"We've got that in the labs - iPaqs with wireless and a camera, sure," said Shane.
Oh, fscking hell, we thought.
Watching America's tech lead in handhelds disappear is like replaying the automobile industry crash of the 1980s, only in slow motion. Twenty years ago there wasn't a foreign car on the roads here, and US-built cars averaged 13 miles to the gallon. Now the domestic industry designs cars such as SUVs that are so awful that no one else in the world wants to buy them, and the classic aesthetics of American manufacturing's golden age - those beautiful Oldsmobiles - return only in the form of weird pastiches like the PT Cruiser.
So it is that today that American computer brands - Microsoft and Intel - sell the world over, while the East makes components and chipsets. In the phone business, it could well be the reverse: the model is being commoditized, but the only American technology deployed by tomorrow's brand darlings such as Sony and Nokia is in the chipsets (TI's OMAP, with Intel and Moto as contenders) and some software (Sun's Java).
Last week Mike Capellas was asked this very same question, and replied that Microsoft's promise of uniting home and office with a personal devices was the most enticing proposition on the table.
Now across from me were two guys with a $4billion R&D budget. I should have asked for a flying car, and would have settled for a great American smartphone, but in the end got Shane to promise to at least have a look at the P800. I hope he does.
64 bits - take 'em or leave 'em
One final tidbit. We asked the CTO of the world's biggest PC company - and the world's biggest PC server company - how it would react to 64bit x86 - from Intel, or from anyone else.
"Yamhill's going to cause a lot turmoil in the marketplace: a lot of turmoil," said Robison.
Er, if it's ever launched, we silently added, ever being mindful of executive indiscretions.
So will HP use Yamhill in its servers, assuming that such a thing would ever happen?
"We're very committed to IPF," said Robison.
So if say, AMD, had an x86-compatible 64bit processor that met the performance requirements of HP's server division, would HP be a customer?
"We work with AMD now, and we hope to work with them in the future," he replied, his grin now as wide as the Golden Gate. So don't take that as a yes. Or even a maybe.
But it is a glimmer of hope that HP might be more than an asset-stripper's dream. For eight months HP's management has sold this merger on the grounds that consolidation is inevitable, and that IT will eventually become a business as commoditized and as lifeless as cleaning pipes, or selling beds: a business where distribution efficiency and scale count for much more than vision or scientific innovation. Even on these grounds, not all mergers are inevitable or even desirable, as this piece suggests. Now we may be biased, but that will never be completely true: HP has the power to make markets, buck markets, and literally - although such a phrased has been debased by its own disinfected marketing speak - sort of, kind of, invent the future.
Carly and Capellas, while you have to give them their hour in the sun - just gave us the orthodoxy of accelerated capitalism: bigger is better, and consolidate faster and further than your competitors. It's a vision fit for an accountant. We think there's still scope for the latter, however, with a company that's bold enough to get everyone else's worried. If that ever happens, we do hope Robison is behind it.
He asked us what we thought of the merger, and after we'd mumbled about not knowing the integration plans in enough detail, expressed the relief that at least we now knew it wasn't the result of the Sircam virus.
For some reason, this brought a grin and two handshakes. There is hope. ®
Bootnotes: We asked Lamphill if HP had expressed interest in acquiring AT&T's Labs in Cambridge, England - birthplace of VNC and other wonders. The lab has now closed, as reported by John Naughton in last week's Observer. Dick said he'd spoken to Hopper while it was still the Olivetti/Oracle Lab, but no offer had been made. It's no ancient history, he said. But MRAM is now in production…
Sponsored: VersaStack at-a-glance brochure