The impact of technology convergence is at its earliest stages, just visible if you know where to look. This is a much broader convergence than the move of voice, data and video into an all digital mode. This convergence will affect all industries and the public because they all use computers.
This convergence is one of processor and system architectures into simpler forms, yet connecting those forms in more complex arrangements. These complex arrangements are networks, but in this context, the word 'Networks' means more than the lines that connect phones and computers. Networks now exist inside of processors (IBM's Power4), between processors in an SMP system, and between the motherboard and the peripherals. These internal networks are in rapid development, presaging rapid change in the high performance system area.
Recently, for economic rather than technical reasons, the powerful Alpha architecture, originating at Digital Equipment, was sold to Intel. The sole remaining future for Alpha is Samsung, if they are willing to go it alone. The MIPS processors, once competitive in the general purpose arena, are now mostly for embedded use. Competition in general purpose computers is pushing aside those that cannot keep up.
The x86 architecture is now down to four vendors - two major, Intel and AMD, two minor, VIA and Transmeta. The low power variants must fight for space in the embedded arena, where they should do well because of the broad software support legacy of the x86.
The embedded arena is in the early stages of consolidation. Volume and financial backing will count for more than technical excellence, much to the disgust of technical people everywhere. That's the reality of the New World, where the ubiquity of the digital processor and its broad support infrastructure makes economics and timeliness the key to success.
Portable phones, PDA and Tablets are converging too, but this process will be slower because of two critical players in this space. First, governments must allocate scarce frequencies, which takes years to become available in semiconductor technology that can support a mass market.
The second player, telecom companies, have to build expensive infrastructure, decide on which protocols and devices to support, and only then can they sell products which bring revenue to repay the large front end investment. To quote a statement about designing processors:
"You put the gun to your head and pull the trigger. Four years later you find out if you have blown your brains out."
Telecom companies may have to wait much longer, with the additional risk that new technology will obsolete what they have done before earning a dime. No wonder the telecom sector has become extremely conservative. They are recovering from the third degree burns of the .COM bust, and probably will be for another year or two.
We can't ignore computer memory development either. For a long period from the mid 1970s to mid 1990s, semiconductor memory density doubled every 18 months, per Moore's law (which was actually an observation at the time). But memory access and bandwidth imitated the inchworm's progress, barely moving 50% faster in ten years.
After ten years of much faster processors and slow memory, the demand for faster memory finally got the memory manufacturers off the dime. Since 1995, typical memory access times have dropped from 60 nanoseconds (ns) to to 10 ns in 2000, and 6 ns today. At the same time, bus bandwidth rose from 32 bits (486), to 64 bits (P5) to 128 bits with Nvidia's NForce chipsets for AMD Athlons and new chipsets for Intel's latest processors.
The combination of advances from 1995 to 2002 have created an explosion of computer power that has impacted PC growth. No longer do people and companies desperately need to upgrade PCs to run more demanding software. We have reached the point where many have enough power to do what they need in a reasonable time, and little incentive to spend money for an upgrade.
It is then not a surprise that PC sales in the US have dropped off, impacted by both the hit to the economy and the 'fast enough' phenomenon. One result has been prices dropping to unheard of levels, such as the mail order Lindows system from Walmart's web site for $199. Add a monitor, or use an old one, and you are on the web for cheap.
Another result, already well along, is consolidation of the PC makers. When HP bought Compaq, it was just the latest step in this process, not the last. Economics, driven by volume and timeliness, strikes again.
There is good news, a bright light at the end of the tunnel. While things look on the dark side for telecom, the computer industry and the overall economy, it simply reflects that we are having to absorb multiple waves of change at the same time. After this painful transition, the results will be more dramatic and beneficial than you can imagine.
Following here and in Part IIIB are some of the good things to come.
One of the reasons for the coming change is the development of the computer utility. Unlike the electric utility system where central generation and distribution is the norm, the computer utility will be widely distributed and competitive rather than monopolistic. IBM has already announced its interest in this capability.
I can see four levels of participation in a computer utility:
The computer utility is already in place for scientific research and high end companies whose product is based in Intellectual Property (IP). It is currently called Grid systems and is being developed with both public and private funding. IBM announced plans in 2001 to build 50 such centers for corporate use. As fast networks are extended to the final mile via satellite, cable and DSL, home systems will be able to tap into these utilities for needs that exceed local capacity.
The first such application, besides Internet access, is remote backup. Soon mobile individuals and small companies will keep their current work on utilities and work anywhere. The problems of security, viruses, backup and processor capacity will be that of the utility, just as electricity is today.
The big differences from electric utilities will be competition between computer utilities, and local processing capacity at the user's choice. This will make for a very dynamic industry that will require the very best technology and people.
Even before computer utilities become broadly available, technology convergence discussed in Part II will change the IT industry dramatically. Existing Web support centers are the prototype utility, where web server farms are supported by immense communications capacity with the security, backup and infrastructure (power, A/C, floor space) already under separate management from the companies using them.
This is a new version of outsourcing that will mostly supersede the current vision of outsourcing. The current approach is where staff, programming and computers move out, everything except the network and systems required to deliver results to the company employees gets moved from the original company to the outsourcing company.
The driver for the change in outsourcing will be the inability for outsource suppliers locked to multi year contracts at fixed prices to meet the rapid and unpredictable changes that will come down the road. Both business competition and technology convergence will disrupt software plans made earlier.
We will see failures in the current outsourcing model that seriously impact both sides of that table. The planned long term contracts will have to change quickly. Any litigation over the contracts will delay and seriously hamper both parties as competition races on. The time to start changing the contract is now.
Another result of the computer utility model will be corporate emphasis on economics of IT, fast and flexible expansion, and comprehensive management tools to monitor performance and reliability. Where the computer is located, or its type and software will be much less important.
Arguments over operating systems and software languages will happen at the utility, not in customer companies. Since these will be about security, scalability and cost, Microsoft will have a much harder sell at this level given their record to date.
New Industries are already appearing. Genomics, the study of the genome and computational development of new treatments and drugs, exists today. Celera Genomics has just bought a two teraflop cluster of IBM p690 systems for this purpose.
The US health care system is not meeting the needs of the public. Over 40 million people have no medical insurance, medical costs rise at a much faster rate than inflation, nurses are in short supply and doctors are retiring because they cannot afford malpractice insurance.
The good news is that medical science is finding cures for diseases that have been untreatable until now. The bad news is that most people can't get or afford these treatments.
I'm not going to attempt a comprehensive analysis of the health care system problems and their causes. That's not a job, it's a career. Instead I am pointing out a product desired by everyone, good health, is expensive and constrained by many factors.
That's an opportunity waiting for solutions. I can see opportunities for home care systems that can monitor health on a daily or weekly basis and alert the person and/or doctor if problems arise, or diagnose and recommend treatment (Take two aspirin and check back in the morning).
More opportunities exist for diagnostic systems at hospitals which can look at case history, ask questions and recommend tests as the doctor's assistant. A prototype of such a system was developed in the early 1970s by a company named "Health Quest." It was too far ahead of its time and stepped on sensitive toes of doctors.
Now the need is much greater, doctors can better appreciate the help and current approach costs are going too high to be affordable. Now the environment is ready for the Health Quest kind of solution, and technology is capable of delivering it.
Next time I'll wrap up with some really dramatic changes we will see in our lives in the next two to six years. What Part A covered was the more or less obvious consequences of technology development, except that rarely do so many change at the same time.
When we have multiple technologies undergoing massive change at the same time, it is no longer a blip in the economy, a slight readjustment and then all is serene again. Not this time. Fortunately, what I am predicting should be good news for everyone, after the change.
[30]