Infotech: Inside New Millennium Computing

He knows the future of computing since he is the man who is programming it. BT presents a test-drive of tomorrow's business paradigm today with Craig Barrett, CEO, Intel Corporation.

He's been inside the future of computing--and it works. But then, it's sheer power, and not strategic prescience, that gives 59-year-old Craig Radford Barrett the capability of charting the destiny of computers and their usage. As the CEO of the $25 billion Intel, the world's largest chip-maker and half of the Wintel duo that controls computing, Barrett leads a mega-team of 64,000 people who are creating our tomorrow. Crucially, that is no longer limited to swifter, smaller, and cheaper chips that deliver exponentially larger quanta of processing power to the computer on your lap. It is, instead, expanding into a world of connected computing, where a billion computers, networked with one another, are opening up unprecedented opportunities in business, entertainment, and, indeed, all forms of intra- and inter-organisational interaction. This is the vision which, with Intel at the forefront, is being pursued by the global infotech industry. The best view, not surprisingly, comes from the leading edge of the wave, where Barrett is uniquely positioned. Converting, for a few hours, his unique perspective into shareware, the man who succeeded the legendary Andy Grove explained last month in New Delhi the dynamics of computing tomorrow. BT's S. Chandrashekar and Anshu Tandon re-create the dazzling multimedia presentation, and offer tantalising glimpses into Intel's business strategy for the Next Millennium.

It's a pleasure to be back in India. I hope to demonstrate some of the things we will be doing in the coming years in the areas of business computing, especially over the Net. Our vision of the future of computing is that we are rapidly moving towards an array of connected computers. There are an estimated 150 million computers connected over the Net today, and that should increase to 300 million in the next 3 years. Over the next 6-8 years, we will have over 1 billion connected computers. The real issue is the impact of these connected computers on business--on the way we do business, and on the ways we become competitive in the business environment. The question is, in what way will these connected computers affect the competitiveness of local economies, or--if you are an international company--economies around the world?

Specifically, this phenomenon gives you instant access to information for taking decisions. There have been several discussions on whether computers aid productivity. Actually, the issue is not whether computers raise the productivity of businesses today, but the fact that you can't begin to play the game unless you have a solid computer infrastructure behind your company. It allows you to compete in your own as well as other markets with equal felicity. Big companies can behave like small companies, and vice-versa. You can provide personalised service to any customer. And, like technologies that have confronted us in the past, you cannot but accept this new technology, and use it to move forward. If you try to deny its existence or ignore it, it will just leave you in an utterly uncompetitive position.

Business over the Net comprises 2 types of commerce. The first is business-to-business commerce, which replaces the classic way of doing business--using telephone calls, purchase orders, faxes, mail messages, and so on--with real-time interfacing over the Net. This is expected to be the dominant form of business transaction as we move forward. The second is, perhaps, glamorous, and so, gets greater publicity in the press: home-to-business, or individual consumer-to-business transactions--where the individual end-user can buy flowers, books, real estate, and automobiles over the Net. The interesting fact about this forecast--which suggests that, by 2002, $500 billion of business will be transacted on the Net--is that each time this forecast is updated, the figure tends to increase by 30 per cent.

Some of the companies associated with our industry--Cisco, Compaq, Dell, and, of course, Intel--are showing how rapidly you can become involved with business over the Net. Dell and Compaq, for instance, are doing $4-5 million worth of business each day, with end-users ordering computers over the Net. At Intel, we do business with customers who deal with end-users. We started a pilot project in July this year, and hope to do $2.50 billion of business in 1998--and substantially more next year. Add these figures to those of IBM, which does business worth approximately $5 billion over the Net--and you can see why the estimate of $400-500 billion isn't very conservative, and will, probably, be exceeded phenomenally.

These businesses are indicative of what many other companies are doing on the Net: personalising our interaction with customers--to provide them with the information they need. They can personalise their databases the same way that you can personalise your own interface to the World Wide Web. With strong encryption technology being used to ensure the security and privacy of such transactions, the capabilities are available 24 hours a day, 7 days a week.

It is very simple for customers to enter Intel through the Net, to find out about product-availability, place orders, and track the status of those orders. So, it involves much more than just ordering products; it is a form of constant communication. Today, only about 10 per cent of the time is actually spent on ordering, or making a change in an order or product. Much more time is spent on checking availability, pricing, and the delivery status. Customers can effectively manage their inventory pipelines on a real-time basis by accessing our data, and can input that information into their product pipeline to ensure that the product is delivered on time for their production. It is a simple, yet powerful, tool that increases productivity and competitiveness in the marketplace--a classic example how businesses are going to be in the future.

Businesses can interface electronically with their inventory supply systems and their customers, and can use instantaneous information over the Net, or a private network, to communicate with customers, warehouses and inventory systems, suppliers, and even brokers who may have access to suppliers beyond your knowledge or the realm of people you do business with. Investment in infrastructure is critical for this kind of business. Your customers and suppliers must have this capability. They have to be networked with one another, they must all have instantaneous access to information. This kind of baseline capability is mandatory for building e-business competitiveness.

In India, the computer business and computer technology has been growing very rapidly, but on a relatively small base. But the total investment level is still lower than in any other country, and that is a tough proposition facing a country that wants to make strides in software technology, that wants to become more of a world player in infotech. If you want consumers and small businesses to be involved with the Net, you have to make the reliability and affordability of Net access appropriate. There is a strong coalition between the cost of Net access and Net penetration. The higher the former, the lower the latter, and vice-versa. But, for India's overall competitiveness, Net access is going to be a key issue. Highly reliable, low-cost access to the Net is the key area that the country needs to work on.

To introduce parallel communication channels to the consumer, there are several things that need to be done to augment the standard communication and e-mail that takes place. Since you have narrow bandwidth to reach the customer, for adequate communication, you have to deploy compression-decompression technology to overcome those narrow pipes. One of the interesting features of the Net is that you have access to all the information in the world. But this could also be a negative factor. How do you sieve through all the information in a reasonable time-frame to get only the information you want?

Well, you can do that in the form of background work on your computer while you are doing your daily work in the foreground. To conduct secure transactions over the Net, various aspects of security must be involved. You need encryption and de-encryption technologies parallel to your communication. Make sure you are not passing viruses back and forth. So, while virus scans, encryption-de-encryption, compression-decompression, and information searches go on in the background, the standard activities go on in the foreground. This is the concept we call Constant Computing.

We are constantly making our computers do more and more things as we get more and more processing power. A simple model of Constant Computing is something like an iceberg. The visible things are running your office applications, communicating or e-mailing, but, behind the scenes, you are doing a number of things such as searching out information, management and maintenance of your computers--a number of hidden activities which are eating up some of your computing power. We want to do as many things as we can with the computing power we have, and automate as many processes as possible to increase that productivity, and use as much technology as we have to be as productive as possible.

People often ask me about the next killer application for computers, which is going to use all the processing power available each year. My answer is that it is not so much a killer application as a killer environment with high processing capabilities that we need as we move forward in this form of e-commerce. Constant computing, which increases individual productivity, requires a lot of processing power.

Similarly, today's array of networked computers and transactions over networks such as the Net will create an explosion of servers. Consider even simple transactions using a network to communicate with suppliers and customers. There is a database server for managing inventory. If you use e-mail, there is a mail-server. If this is done on the Net, there has to be a Web-server. If you communicate with other companies through their firewalls, there is going to be a proxy or a firewall-server. And, every time you have an additional transaction, there is usually another server involved. Sometimes, these functions are combined onto one server but, usually, they are on independent servers--for simple transactions.

If you want to order something, there's a whole array of servers involved--just to carry out a simple financial transaction between a supplier and a customer. For instance, banks will have firewall-servers to protect the integrity of individual corporate clients, integration-servers for commerce transactions, authentication-servers, billing-, and proxy-servers. In an end-to-end transaction on the Net, there may be 10 servers involved. That is one reason why, worldwide, the fastest-growing segment in computing today is not in the client space, but in the server space. The requirement here will be for high-performance scaleable servers. Importantly, we are now able to bring some of the economics of the pcs space into the high-performance scaleable server space. And, as servers become more important, their cost-effectiveness becomes more important too.

As businesses become more involved in computing, and as people like you and I travel, we carry the computer, the cords, the plugs, and the accessories that we need in order to get networked when we travel. But wouldn't it be nice if you could just take your computer and have it act as a cellular communication device--no cords to plug into networks--just like a wireless, any time, any place? Of course, it isn't just connectivity, you also want bandwidth. If you are in an office-domain, and you want wireless connectivity to a Local Area Network (LAN), you want that bandwidth to be consistent with the bandwidth of the LAN--up in the megabits rather than the 56 kilobytes at which telephone calls are made.

One of the exciting things about mobile technology is that it is consistently approaching desktop capability in absolute performance. If we can get a two- or three-fold improvement in mobile computing capability, many corporations around the world will ask why, instead of buying your employees both a desktop system and a laptop, we can't buy just the laptop, provided it has the performance and graphics capability of the desktop systems? It is much cheaper to give employees one computer, let them use it at home, on the road, or in their office, especially if the computer functions in a wireless fashion.

The other issue is the ease of connectivity and the ability to interface with your computer--not only with local area networks, but also with other peripheral devices--in a wireless fashion. One of the various wireless technologies at work is named BlueTooth. It is a product of a consortium between companies involved in cellular communication, computers, and consumer electronics. The objective is to allow scaleable access and communication between different appliances, including computers.

The capability that has been developed is for short-range, wireless connectivity, scaleable from 9.60 to 56 kb when connecting from a specific remote location. But, if you put yourself in a hotel or an airport where you have XDSL or isdn technology, you can get up into the 256 kb region. And if you were in a corporate environment with a LAN, the speed would go up into megabytes. This is a simple, short-range 10-metre or 20-metre type of connectivity, with low power requirement, and scaleability. This is the principal aspect of making mobile computers easier to use, covering simplicity of plug-and-play, wireless connectivity, and full functionality: from voice input to downloading an image to video-conferencing.

Part of the ease-of-use comes from making computers work like standard consumer electronic equipment--like a TV set. In a TV set, you turn it on, and it goes on. You turn it off, and it goes off. If you want a different channel, you simply change the channel. Computers should be exactly analogous to that. You should be able to turn them on; you should be able to change applications the way you change channels. And when you're ready to turn it off, you should be able to do so; you shouldn't have to go back to the Start button, and be asked whether you really want to turn it off, and have to say yes, and be told to wait until it is safe to turn your computer off. This is not user-friendly.

The computer industry is, in fact, working on ease-of-use capability. And among the first things that will be developed are the abilities to have your computer always connected, to communicate with other people, to be incredibly easy-to-use, and to plug your peripherals into. As we go forward, we will get rid of the need for serial ports and parallel ports. We can have one simple universal serial port. And the computer can be much simpler. We don't have to carry all the excess baggage, cost, and complexity associated with legacy capabilities. We can have computers made instantly available by putting them to sleep--into a very low power configuration--rather than turning them off when they're not needed. There is going to be a big movement in the industry in the next several years towards ease-of-use and simplicity so that anyone in the world who can play with a consumer electronics product can play with the computer too.

What comes next in terms of processor capability and computing capability? A typical roadmap from a company like Intel shows that technology does follow Moore's Law. Gordon Moore, Intel's founder, had forecast in 1966 that processing power would double every 18 months. Moore's Law has been valid for the past 32 years, and will probably be valid for another 15 or 20 years. By the second half of 1999, Intel should have processors in the 700 Mhz range, and, by the end of 2000, in the gigahertz range at the high-end.

We have a whole range of processors in different segments: from entry-level desktops to workstations and servers. There are products even at the bottom, using a different architecture, named StrongArm, for hand-held appliances and set-top boxes, where legacy and compatibility with the conventional instruction set is not necessary, but where you do need a powerful, low-power-consumption processor. In the next few generations of Intel's microprocessors, upto 2002, the 64-bit family will have 4 processors, the 32-bit family will have a couple of new architectures, and newer micro-architectures that appear.

We can continue to bring the same increase in processing power into the marketplace every year as we have in the last 30 years. The 64-bit family will be an industry-standard for workstations and servers while the 32-bit family will be very satisfactory and very competitive for the desktop and lower-performance servers. But, while this roadmap is very interesting, it doesn't tell the whole story. A look at the platforms will reveal the capabilities that will be available. A study of the various aspects of the basic computing platform reveals that, over the next few years, we will get a three-fold increase in processing power, a six-fold increase in graphics performance, and a dramatic improvement in memory bandwidth, giving processors access to bigger banks of memory.

Now, what are we going to use all this processing power for? I know that that is the burning question in your mind. As e-commerce becomes more important around the world, we want to be able to create images in real-time--something you cannot do with the most powerful client or PC you have today. You cannot even do it in real-time off-line. You have to individually render the images, and stitch them back together. In future, with greater processing power, you will be able to approach that capability. And you have to imagine the impact this will have on commerce and shopping and advertising and communicating over the Net. It will lead to a sea-change in the way we shop and communicate today.

At Intel, we have a simple image: a world of 1 billion connected computers. That world of connected computers needs hardware, it needs software, and it needs applications. Companies and countries that choose to accept that image and aggressively pursue it have a successful future. It is not a luxury to invest in this area. It is a necessity if you want to be competitive. And you need to invest not in old technologies, but in newer technologies. It is an investment you have to make if you want to play the game, and want to experience even the possibility of being successful.

INTEL INSIDE STRATEGY

It also makes chips. Within 6 months of taking over from Andy Grove in May, 1998, as its CEO, Craig Barrett has put Intel firmly on course to becoming a networking superpower too. Driven by the vision of a global consumer-base whose computers are interconnected, Intel has reprogrammed its business strategy. Besides capturing the CPU space of PCs, servers, and supercomputers, it wants to provide the other products that will go into converting every stand-alone computer into a connected one. With, of course, (a billion) Intels inside all of them. The roadmap of Intel's alliances, past and future:

February, 1998: The formation of Home Networking Operations by Intel.

June, 1998: The founding of Home Phoneline Networking Alliance, a consortium of 11 companies to develop specifications for home networking products based on phone-line technology.

July, 1998: The launch of the first personalised music channel by Intel, via broadband connections, in association with LAUNCH Media.

October, 1998: The investment of $20 million by Intel in @Home Nederland to access the 1.40 million Dutch homes that will be on broadband cable.

October 1998: The launch of new video technology for the Web by Intel in association with RealNetworks.

October, 1998: The launch of enhanced DTV programmes by Intel, in association with the Public Broadcast Service, for reception on PCs.

Early 1999: The launch of Intel CPUs incorporating Hitachi's All Format Decoder.

Late 1999: The launch of specs on digital connectivity for digital displays and high-performance PCs, with Silicon Image, Dell, IBM, H-P, and Compaq.

Early 2000: The launch of robust, cost-effective, and transparent methods for transmitting, receiving, and exchanging digital entertainment content between electronic appliances.

Clearly, Intel is tuning into, not chipping in, tomorrow

--S. Chandrashekar

INTEL INSIDE INDIA

It also makes money. In the past year, Intel has invested $350 million in 110 start-ups all over the world, picking up equity stakes whose value is now over $1 billion. The company plans to clone the strategy in India. Of course, as elsewhere, the aim is not to watch its investment swell, but to provide venture capital to start-ups in the value-added software and Net-related businesses. Explains Atul Vijaykar, 44, Director (South Asia), Intel: "The company is trying to bring the Silicon Valley culture to this country to encourage entrepreneurs to take their ideas to market." His hotlist:

A 10% equity stake in Rediff Communications.
Six other proposals are under consideration.
Among the 24 short-listed--and yet-to-be-named--software and service companies that will develop products for Intel:
A partnership with the Delhi-based Trisoft Design, which is setting up a shopping mall on the Net, for the use of its Common Data Security Architecture.
A partnership with the cricket site, CricInfo.
A collaboration with the Bangalore-based Silicon Automation Systems to develop infrastructural software for graphics cards.
A partnership with the Chennai-based Polaris Software, which has developed a retail automation software package.
An alliance with the Bangalore-based VR Real Technologies & Vision, which creates 3D software.
A collaboration with the Mumbai-based Maya International (a.k.a. Maya The Magic Shop), which creates content for the entertainment business.

While Intel's target for new investments next year is $400 million, since it hasn't put a cap on the allotments, much of that could be poured into the equity of the cool infotech shops in this country. Obviously, Intel isn't afraid of putting its chips on India.

--Anshu Tandon