Despite my misgivings about the merger, HPC has managed to meet expectations financially in its early days after the merger. Achieving this in the current recession speaks well of the new executives as a well planned consolidation. Credit is due for competent execution of a complex plan.
How about the long term? That outlook is not so rosy. Some problems HPC faces over the next few years apply to more than HPC, yet HPC faces the broadest range of problems.
First, let's list the positives for HPC:
Now let's look at the challenges:
HP's profits have been driven for some time by the printer division. That profitable business now faces an attack on multiple fronts.
First, Dell will be sourcing printers from a competitor. With Dell's volume and low cost sales model, that could cost HPC 25% of its printer business over time. Additionally, the printer business is maturing and growth will slow there too.
Second, the profit from printers is mostly in the supplies business. Just like razors, they almost give the printers away so they can sell branded supplies at a high profit. The problem is that high supply prices draw competition which has already driven supply costs down and will continue to erode margins.
A third factor is that printer technology has reached a level that more than satisfies even picky users. Ink jet printers claim up to 1200 x 2400 dpi print resolution, and most laser printers manage 1200 dpi or higher. Most people or businesses don't need that level of detail.
Practical technology limits, maturing sales and competition will reduce volume of new printers, and supplier competition will cut into supply margins. HPC's cash cow is going on a diet.
Uptake of personal computers, in the home and business, has already taken the easy two-thirds of households in the US. The remaining third of the market will be slower uptake and mostly lower price systems. The new $200 & $300 systems from Walmart will take at least 50% of that third.
Since the excitement of the first 1 GHz processors from AMD and Intel in 2000, speeds have ramped rapidly to 2 GHz and over. From here on in, diminishing returns in performance have set in, slowing replacement sales. Even my older K6 processors still have useful lives in less compute limited applications. Just how much speed do you need to browse the web or write a letter?
With processors that are 'fast enough' for most applications, the replacement cycle in business and home systems has already lengthened. Businesses that used to upgrade every two to three years are now on a four year cycle and only in the last year have 486 based systems finally disappeared from home use, if not from the home.
With extremely tight margins in PC hardware, volume and efficiency are critical to profitability. For the past few years, first place has belonged to Dell and its direct sales model. With approximately 40% of the market, well above HPC and IBM, Dell will keep the prices down to where profits in the PC business will be slim for everyone else.
HPC's only solution is to match the Dell model for their low price systems. These days that includes smaller servers too. As Dell's sales continue to rise, eventually IBM and HPC will have to decide what they want to do with the PC business because it will be only a single digit profit source, below what those companies want to achieve.
Given the current economic situation, making any profit in personal computers is is a real challenge. It's not going to get easier.
HP and Compaq both have had powerful and unique high end systems, with as many as 64 processors in a single system. For Compaq it was the Alpha chip, bought from Digital Equipment. For HP, it was the PA-RISC line of chips, internally developed, that powers their Superdome line of large multiprocessors. HP's RISC chip was no slouch either and remains competitive with two more versions, the 8800 and 8900, promised.
Both of these systems had success in the market, yet neither really exploited these chips in the workstation market where the biggest volumes were. Alpha was second sourced to Samsung, but PA-RISC was always an HP product. Alpha was the leader in floating point (FP) performance for a long time, and only recently has been matched by Intel. Alpha has one more version nearing release.
When I first heard that Compaq was selling Alpha to Intel, I was stunned. The speedy Alpha chip was one of the few things that gave Compaq a unique advantage over other vendors. Compaq's multiprocessor systems, up to 32 Alpha chips, was a perfect mid to high end server for business, scientific or mixed workloads.
Even more, the Alpha systems ran under a mature and powerful operating system, VMS, which was solid and had better cluster capability than other vendors. For me, the Alpha chips, servers and workstations were Compaq's crown jewels, both as products and intellectual property.
In addition to selling these jewels, they only got a pittance of $200 million for the whole batch - the chip design, patents essential to Alpha's performance and a fab line. In return, they got only the rights any other Intel purchaser could have.
Compaq became just another Intel reseller and lost it's unique identity and technology edge. This move almost certainly sealed Intel's hegemony and Compaq's doom.
Enter now HP and the acquisition. Excuse me, that's The Acquisition. The good news is that they pulled it off. The bad news is that HP is also discontinuing a unique product edge, the PA-RISC chip. Rather than invest in something to differentiate and outperform the competition, they joined Intel in designing something completely different.
In fact it's so different that even today, eight years after the project started with a massive engineering effort, we are just seeing the first usable results with the second version of the Itanium. Not only does it have an ugly name, but it can be used as a room heater in cold weather. Single chips dissipate 130 watts for a performance barely better than an Alpha chip.
The next version, Madison, will have a six MB level 2 cache, and contain approximately 500 million transistors, most of them in the cache. This is brute force silicon engineering and production, not an efficient design. Only with Intel's large base of expensive FABs could they afford to build such a large chip.
HP originally planned to end the PA-RISC series at the 8700, but as Itanium slipped and slipped, HP's customers and possibly common sense, caused HP to continue design work on the line. This now includes the 8800 dual core design nearing delivery, and the 8900. At least for now, that's the end of the PA-RISC line.
HP at least has the advantage of helping design the Itanium and planning to make conversion easier from PA-RISC. But they will also face the fact that they are just another Intel reseller, albeit with an inside track and probable pricing advantage. Whether that's enough to maintain an edge remains to be seen. I'm skeptical about the prospects, but it will be a while before the results are in.
The fast growth, high margin days for personal computers, and to a lesser extent, the computer industry, are fading fast. The industry is maturing, and with maturity comes slower growth. Few if any of the IT businesses are prepared to live in that environment. Most of them simply look on it as due to the recession, a temporary situation.
It isn't. Yes, there will be growth spurts as new parts of the world reach the economic level that enables rapid uptake of computers, but that has already started. With prices for basic systems without screens already down to as low as $200, the margins will be minuscule. A price increase in shipping could significantly affect sales.
Earning reasonable margins on computers will require that the vendor add real value, a concept foreign to many companies today. Advertising that extols tiny differences between systems for large price increases will lose their effectiveness when the latest 2.1 GHz chip is only 5% higher clock speed, and maybe 2% better performance. Who needs it?
In order to survive, never mind compete and make a profit, IT companies must add value to the base hardware, most of which is reaching commodity status this year. Some will take advantage of this and identify their company as an education specialist, or a home user special software package, or maybe the best dual processor software.
Even many white box builders will face extinction as competition drives down prices and profits so that only builders in low wage countries will make a profit on the base box. But since the systems are so powerful already, the issue of the exact components in a box will only be of interest to specialists and writers struggling to make a detail into a difference.
HPC has passed the turmoil of acquisition. While there are details to finish the job, the big step is done, and HPC gets a good grade. From here on, it gets tougher.
HPC will face small margins on the PC business, yet they need that to be a full line supplier. As PA-RISC fades into the sunset, other competitors will match their Itanium based performance and challenge their market share and margins.
HPC faces slowing volume in printers, increasing competition in printers and supplies, limits in new technology that can make a visible difference. From here on it's tuning and tweaking, not major advances.
Is HPC management up to the challenge? Their performance in the turbulent acquisition phase is an encouraging data point. What they will face in the future will be more difficult than deciding which product lines will die and which survive, and laying off good employees who happen to be in the wrong place at the wrong time.
Ultimately, it will be HPC's creative responses to the challenges I have outlined that will determine their mode of survival. The best answers must be found if HPC is to do more than just survive. On that issue, current rumors to the change of the "HP Way" to a more centrally organized company is not encouraging.
The original HP was known for its open access and experimentation known as "The HP Way" and "Management by Walking Around." I believe those were essential elements of HP's success, and a key benefit that retained good people. If they become just another centrally managed business with one research center, they lose an essential part of what created HP's success.
Without that flexibility and creativity, HPC's future looks dim.
[30]