YouTube Facebook LinkedIn Google+ Twitter Xingrss  





CONVERSATION · DAVID L. TENNENHOUSE

 

Bio·IT World Executive Editor-IT John Dodge recently interviewed Tennenhouse about Proactive Computing and the company's foray into bioscience technology. 
Sept. 9, 2002 | As the world's biggest chipmaker, Intel Corp. is looking to the technology-thirsty biosciences as an ideal environment to debut future semiconductor technology — chips that promise to talk to one another, behave autonomously, detect "wet" and "dry" states, and even do single molecule detection. The company has 7,000 people in 75 labs around the world and spends $5.2 billion annually on R&D. Intel's chief architect of the future is 45-year-old David L. Tennenhouse, who serves as director of research for the company's corporate technology group. He joined Intel three years ago after serving as director and chief scientist for the Information Technology Office at the Defense Advanced Research Projects Agency (DARPA). There he formed his vision of Proactive Computing, which Intel has adopted. Several Proactive Computing technologies will be explored in depth at the Intel Development Forum in San Jose, Calif., the week of Sept. 9.


Q: What is Proactive Computing? 
A: People are interposed between computers and data, and such activity demands too much of their time. Proactive Computing is getting data from the physical world to the virtual world without human intervention. It's computers working to anticipate people's needs and transacting on our behalf. Right now, the computer sits around waiting for humans to ask questions, and that's the way it's been for 40 years. That's interactive computing.

We want to get the computer connected with the physical world, because shuttling data between these worlds is such a waste of human time. Our long-term agenda is proactive computers anticipating people's needs. If we can unlock data from the physical world, we get rid of the [human] bottleneck. Then you've enabled a lot of interesting applications. There's nothing wrong with the people. They're doing the best they can.


Q: How does Proactive Computing relate to the biosciences?
A: There are a lot of implications for the biosciences. Labs are chock-full of embedded micro-controllers shuttling data into computers. People are shuttling the data around. Can you imagine anything stupider? You want it collected electronically so [scientists] can mine it.


Q: What are the cultural, legal, and social challenges of Proactive Computing? It sounds a bit scary.
A: Privacy and security are some of the software challenges that need to be dealt with. If you're able to walk into an intelligent environment, it needs to be at your discretion how much personal data is exchanged. If you walk into your lab, you're going to want the system to identify you and bring up your personal lab notes. You're not going to want your notes to pop up if someone else walks in the room.


Q: Describe some Proactive Computing technologies.
A: A big piece of this puzzle is embedded controllers. There are 80 to 100 in your car. What's sad is that they are not networking. It does something very basic and then drops it on the floor.

A quantum step forward is ad hoc sensor networks, where nodes are getting data and relaying it on behalf of other nodes. You can query this data as it comes off the network and feed it into XML databases or an electronic lab network.


Q: Can you give examples of sensory networks?
A: You could drop thousands of nodes (each includes a microprocessor, memory, sensors, antenna and battery, and is nicknamed "mote") from an airplane into a big forest fire. They find each other, form a network, track the movement of heat, and relay the hot spots back to a fire captain.

You could also sprinkle them throughout an office high-rise to pinpoint and analyze earthquake damage after the fact. A more sophisticated sensory network comes out of a project we call Labscape. Everything a biologist does is recorded automatically and put on a network so he or she doesn't have to worry about it. Wherever the scientist goes, his or her data appears on the nearest screen.

Q: Wireless communications and radio technology built onto a chip are a big part of sensory networks. Please describe what Intel is doing there.

A: We're using MEMS [microelectromechanical systems] devices and doing more of analog RF [radio frequency] in classical CMOS [complementary metal oxide semiconductor]. After all, an antenna is a sensor. There's always a tradeoff of what goes on the chip and what goes into the processor. Some of that last step of this has to do with marketing. (Intel published a white paper in April entitled "CMOS Radio.")


Q: What MEMS projects at Intel relate to the biosciences?
A: We have a MEMS activity called Precision Biology. It's really chemistry and physics in service of biology and biologists. We're looking at single molecule detection. We're fairly new to the space, but we think we can do it and we have approaches that will involve MEMS fluidics. Fluidics (Intel calls this piece of the technology "Wetware") holds the sample in place. Then the sensor's computational ability does the detection.


Q: How far away is something in terms of a prototype chip and commercialization?
A: We're two to three years from a working lab prototype. We have a couple of groups working on it that started five to six years ago. Turning it into a widespread business is much more difficult.


Q: Many companies work on similar-sounding technologies known as lab-on-a-chip. What does Intel bring to this party?
A: Intel internally has good nanodetection ability considering we're [building chips] at 90 nanometers (see August Bio·IT World, page 1). And we have instruments [for chip-making] that are 10x better than [the chips we produce]. We know others researchers are working on the problem.


Q: You mentioned creating a market for a molecule detection chip is difficult. How are the bioscience markets different from Intel's traditionally horizontal marketplace?
A: The biotech space frustrates me a bit, and we [Intel] are even investors there. It's extremely vertically integrated and focused on two or three long-shot drugs. Say someone comes up with a great technology. If there is no payoff, the technology disappears. If there is a payoff, a big pharma will acquire the company and the technology.

It's kind of like us selling our Pentiums to just one customer. [Bioscience] technology does not get the widest possible usage. It's a structural problem and the startups can't be blamed. There's really no great mechanism to extract value [from these technologies]. We bring a different market perspective, applying technology to the broadest potential market. * 








For reprints and/or copyright permission, please contact  Jay Mulhern, (781) 972-1359, jmulhern@healthtech.com.