Ajay Bhatt: Intel’s Rock-Star Inventor
The creator of USB hasn’t stopped inventing. Now he wants to build an all-day computer
Steven Cherry: Hi, this is Steven Cherry for IEEE Spectrum’s “Techwise Conversations.”
One day, about 25 years ago, an engineer at Intel was trying to install a multimedia card into his computer and came up with the idea for a universal port for peripherals. The engineer was Ajay Bhatt, and the port became USB.
Today, there are by one count 6 billion USB devices in the world, and not just printers and keyboards and mice. There’s a USB butt cooler for your chair, USB heated gloves for your hands, and a USB disco ball for your inner John Travolta. Ajay Bhatt wasn’t kidding when he envisioned a universal serial bus port.
Ajay Bhatt hasn’t stood still, and neither has USB. There’s a USB 2.0 and 3.0, and Bhatt has worked on the Accelerated Graphics Port and PCI Express. He’s worked on Intel’s desktop power-management architecture and is now helping to develop a computer that will work all day. He’s my guest today.
Ajay, welcome to the podcast.
Ajay Bhatt: It’s nice to be with you, Steven.
Steven Cherry: You ended up with collaborators from six other companies: Compaq, DEC, IBM, Microsoft, NEC, and Nortel. How did you end up with so many cats to herd?
Ajay Bhatt: Well, the thing is, even though we started with, you know, sort of a silicon company, we started with a certain view of the computer. We wanted to be inclusive. We wanted to make this port universal. We wanted to comprehend requirements from system vendors and peripheral vendors and software developers. So by gathering a group of diverse folks, it made USB even more successful, because we got, you know, input that we wouldn’t have from Intel’s side.
Steven Cherry: It really is universal, and I was sort of joking about the butt warmer, although it does exist. But there’s different pin settings for different applications—printers, smart cards, audio/video. There’s one for health care. Does that get used?
Ajay Bhatt: Yes. Actually, there are some companies that make glucometers, and there’s some equipment where USB is used, particularly to upload and download information. Just the way you upload and download information from a smartphone, for example.
Steven Cherry: I love the convenience of USB, but I have to say it drives me crazy that it has an up and a down. You’d think that I’d get it right at least half of the time, but I don’t think I do.
Ajay Bhatt: That’s an interesting point. If I were to do this all over again, that’s one thing I would fix. When we started, though, I’d like to remind you that some of the ports were such that the degree of freedom was about 360 degrees, and blind mating a cable to a port was very difficult. So when we started, by taking the connector we had, we made a big jump, an advancement. However, the limitations we had were with respect to the cost. So we found the cheapest connector that we could afford at that point, and that’s what we’ve ended up living with. But as we go forward, I think we may get an opportunity to fix that problem as well, at some point in the future.
Steven Cherry: I’m glad you mentioned cost, because I wanted to ask you: FireWire, IEEE 1394, already existed in 1995, and it was much faster. Did USB win entirely on price, or were there other things as well?
Ajay Bhatt: Well, a few other things as well, but when we started, we reached out to everybody in the computer industry with the view that we had. We even made an attempt at approaching people to revise their specs to meet our requirements. So there was a real effort made in earnest to sort of bring people together. But one of the big things that we also focused on, besides cost, was an architecture that was open, that was widely available to developers, royalty-free, and at no cost. So we developed this technology and made it available.
Now, such terms were not available for all the other interfaces that were out there. They had a different business model. Our view was to really promote an open standard that would enable new users of the computers, and attract new users, because computers would become much easier to use.
Steven Cherry: Apple turned out to be the leading user of FireWire, and it seems like they keep doing this again and again. I’m wondering what you think of their Lightning, which I guess is similar to USB 2.0, with some proprietary extensions.
Ajay Bhatt: I haven’t looked at Lightning in much more detail, but I think the problem that they’re trying to solve for all Apple ecosystems is to make—I guess one of the most advanced features of that connector is that it’s flippable, right? So the problem that you were talking about with USB, which only goes one way, they’ve sort of solved the problem. I think the other problem that they’ve fixed is the size. The connector that they’ve chosen for Lightning is a very, very small size. And when you have very portable devices, size does matter.
Steven Cherry: What about Thunderbolt, which I guess is also sort of an Apple port technology—and I guess it’s also a superset of one of yours, that is to say, PCI Express.
Ajay Bhatt: Yeah, so I’m on one of the—two of the primary patents on Thunderbolt. I actually was the original guy who worked on Thunderbolt for a while before I moved over to a different job assignment at Intel. But Thunderbolt basically is addressing a need of supporting two protocols: display port or very high-resolution displays. And PCI Express, in certain of these cases, you want to desegregate IO and take it outside the box. And those are the two needs that Thunderbolt meets. So it is targeted toward a specific segment of the market, and I think a specific set of applications.
Steven Cherry: Is it the next big thing, or will there be a USB 4.0?
Ajay Bhatt: Well, I think both these, Thunderbolt and USB, will keep evolving. I know USB IF, or Implementers Forum, has announced their intentions for extensions, so I expect USB to be extended. And so will Thunderbolt be extended as well.
Steven Cherry: And the main point of extending USB would be the large devices? Especially being able to charge them, right? Like monitors?
Ajay Bhatt: Well, a couple things: One is, there’s a recent extension to USB—it’s called USB Power Delivery, or USB PD. And USB PD is an extension that allows a power source and a computing device to negotiate power delivery mechanisms, or the voltages and the current. So that’s the spec that enhances the way we charge devices, or the devices give charge to external peripherals.
So USB PD is one of the extensions. The other things that are happening are related to the data rate. As you can see, as we go to devices that store more and more data—for example, I have SLR camera. When I shoot RAW, each picture is about 25 to 30 megabytes. Now, to transfer those pictures from the camera to the computer takes a long time using USB 2.0. Using computer extensions, we would have the speed in a few seconds, rather than a few minutes.
Steven Cherry: Maybe you could just tell us about the all-day computer: Why has Intel put its best engineer on the case there?
Ajay Bhatt: Well, one of the major pain points that we all have using portable computing, and for me it is tablets and laptops, is the power efficiency. You know, even though in the past people have claimed that it has six-hour battery life, in normal use you’re lucky if you get half that, right? So that was clearly the pain point for people: They want to use the computer all day long; they are sort of tethered; they have to bring the power brick with them. And I delivered the advances made in silicon technology, as well as the design techniques involved in architecture, so that we can actually fix this problem.
So in the last four years or so, I’ve spent a lot of time looking at all aspects of PC architecture and sort of audited each part of the architecture and systematically gone in and tried to fix the issues in the system that would result in draining the power unnecessarily. So we’ve really done a big overhaul in PC architecture to enable PCs to now run all day on a single battery.
Steven Cherry: In some ways it’s really just a weight problem, right, not a power problem? I mean, if we were just willing to carry computers, you know, that were state-of-the-art 10 years ago, we would have all-day computers, wouldn’t we?
Ajay Bhatt: Well, that’s not the right way of solving the problem. You know, we didn’t want computers to be luggable. We still want computers to be very sleek, lightweight, attractive, yet have a good dynamic range when it comes to performance. So adjust the power consumption of the computer based on the tasks that you’re doing. Of course there are certain tasks that are very, very demanding, and we want to provide that level of processing power to support the most demanding application. At the same time, when you’re doing some simple browsing or word-processing kind of tasks, then we want it to be much more power efficient. So we want to provide power efficiency while maintaining reasonable size and weight and temperature of the device.
Steven Cherry: Up to now, producing light, sleek, attractive computers was a sort of a differentiator for a manufacturer, and I find it interesting that Intel is now trying to solve this problem for the industry as a whole. And you’ve said that the point of USB is that the companies should not be competing at the level of infrastructure. But I guess I’m wondering, how do you know what’s infrastructure and what should be a manufacturer’s value-added?
Ajay Bhatt: That’s a very good question. Usually when the buses interconnect, where two sides of the interconnect are used by two or more parties, that becomes an infrastructure issue, right? Because when you want to communicate with other devices, you’d better agree with the rest of the engineering community on the specs. But then once you define the rules of communication, how efficiently you communicate, or if you can create a much more efficient implementation or architecture, then that becomes the differentiation.
It’s very similar to building a highway by agreeing on the size and the lane width and what have you, but then building a car or set of cars that run on that highway is akin to building a differentiated computer device, right? So may that be a wireless or wired connection, there’s certain aspects of architecture where you need to be open with respect to architecture and specification, such that anybody can build interoperable devices, because ultimately these devices will be used by users, and you don’t want them to be frustrated. Because if each company does proprietary design, then these devices don’t interoperate. So for the sake of interoperability, you must invest in common infrastructure.
Steven Cherry: So what’s the interoperability issue when it comes to the all-day computer?
Ajay Bhatt: Well, all-day computers are, remember, in the computer we have components from a variety of vendors, right? So if we all agree to a common set of power privileges in this case, then we know when the devices need to wake up and when they go to sleep, otherwise a misbehaved device could actually keep the rest of the computer on and drain the battery. So even when it comes to all-day computing, each devices, along with the software, have to be able to communicate with each other with messages that communicate the state of the machine with various substations.
I think that’s where some of the work we’re doing is invaluable, because you have to agree to entering and exiting certain power states at the right time without actually being visible to the end user. So if the end user clicks on some application, appropriate behavior should be demonstrated. However, underneath, the subsystems that are not being used have to be powered down for however long they’re not needed. And then they should wake up transparently to the user, so there’s a tremendous amount of work that needs to be done inside the computer to make it much more efficient.
Steven Cherry: I wanted to ask you a question about Intel in general, and I thought I’d compare it to some other organizations. Bell Labs used to be a place where people could do university-style research without having to teach, and today Microsoft Research is like that. But it seems like a lot less makes its way out of the lab and into products—the Kinect, maybe, being the notable recent exception—than does at Intel. Does Intel have some secret sauce when it comes to generating immediately usable R&D?
Ajay Bhatt: Well, so, we have a rather large investment in Intel Labs. Now, Intel Labs does various things. One, we have close relationships with academia. So we have Intel Science and Technology Centers at universities around the world, including various cities in the U.S. So Intel researchers, along with university researchers, collaborate on certain things.
We also, at Intel, have researchers who focus on fundamental research, may that be in process technologies, architectures, or design technologies. And they take a long-term view of technologies that will be needed in the future. And then we also have a view where we bring together people from different disciplines to come together and develop something called Rapid Prototypes. And these prototypes, or these technologies, are based on fundamental research that we may have done, and then based on that, we create new technologies that can be deployed in our future products. So we work with universities, do fundamental research as well as applied research, I would say, and a lot of these ideas make it into the product.
Steven Cherry: Well, Ajay, Intel did a somewhat silly TV commercial that featured you as a technology rock star, and Conan O’Brien did a version of that that managed to be even sillier, although it was hilarious. And you were in it, and it seemed like you really enjoyed it. I wonder, though, if you find it more humbling than anything else to know that you’ve improved the computing efficiency of basically half the planet every day.
Ajay Bhatt: Well, you know, I feel extremely privileged to get this opportunity to leave my fingerprints on the computing industry, right? Clearly the vision that we had was, you know, we were at the right place at the right time, and we were able to assemble the right group of people to make USB happen.
And with respect to Intel’s commercial, well they were looking to highlight thousands of engineers that we have, and they just chose me as an example because people could relate to USB, and hence they asked that my name be included in that commercial. And that’s how I ended up on Conan, and clearly Conan really—it was sort of out of my comfort zone to be on Conan. But I went along with it, and it was a lot of fun.
Steven Cherry: Yeah, it does look like a lot of fun, but there’s nothing silly, to be sure, about the achievement of USB, and PCI Express, so thanks for those things, and thank you for joining us today.
Ajay Bhatt: Thank you so much for having me.
Steven Cherry: We’ve been speaking with Intel’s Ajay Bhatt about the past, present, and future of USB.
For IEEE Spectrum’s “Techwise Conversations,” I’m Steven Cherry.
Photo: European Inventor Award
This interview was recorded Tuesday, 25 June 2013.
Segment producer: Barbara Finkelstein; audio engineer: Francesco Ferorelli
Read more “Techwise Conversations,” find us in iTunes, or follow us on Twitter.
NOTE: Transcripts are created for the convenience of our readers and listeners and may not perfectly match their associated interviews and narratives. The authoritative record of IEEE Spectrum’s audio programming is the audio version.