I initiated a correspondence with Jim Zemlin, the Executive Director of The Linux Foundation, some months back regarding a blog entry he posted on The Linux Foundation site. I was interested in discussing the future of Linux, the suitability of Linux for netbook platforms, and my observations that Linux really pioneered the concept of an app store with Debian distro repos but dropped the ball on realizing the promise of that breakthrough. I’m sure I’m not the only guy who has made this observation, so I’m not taking credit for a lot of the themes in Jim’s recent Wired interview, but I found several things about his interview interesting, and felt the lightbulb go on over my head.
The most interesting part of the Wired interview from my perspective was Jim talking about the Intel Atom processor App Store initiative, AppUp. Obviously Intel has realized that the Atom line in netbooks and tablets is going to be critical to its success going forward. This is after several years of Intel displaying a pretty public contempt for the low-margin Atom line of processors. Let’s face it: Intel would much rather sell CPUs and foundation technology like motherboard reference designs for $1,000+ notebooks built on traditional x86/x64 CPU cores than components designed for sub $500 market consumer electronic technologies. The high end is where Intel’s biggest margins are. The low end turns Intel into a commodities manufacturer.
But the entire article got me thinking about the mobile consumer computing device revolution, who the key players are, how the tech blogging industry has been completely overtaken with articles about these technologies and companies, and how little press Intel has been getting during that time. I called some of my friends at Intel and asked if I could be put in touch with someone who could explain more about the Intel AppUp application market that Jim Zemlin revealed in his Wired interview.
Intel recently announced an aggressive new initiative to revitalize its presence in the mobile digital consumer device market built around their Atom core mobile processors. Intel has established an application marketplace, branded the AppUp store, to support marketplace frameworks bridging the gap between mobile app developers and consumer end users.
My interview with Intel’s Peter Biddle
I interviewed Peter Biddle, Head of App-Up Products and Services at Intel Corporation, and we discussed the future of Intel mobile digital consumer devices, app markets, and vendor relationships; we also talked about whether Intel can stay relevant in this very competitive segment of the marketplace despite having let other vendors get a significant head start. Below are highlights and excerpts from the interview, which took place on Sept. 8, 2010. If you would like to listen to the entire interview, you can download the podcast. (Disclosure: I am a former Intel employee.)
Does AppUp signal a change of direction from Intel?
My first question focused on the fact that I’m currently using an iPad and that Intel spent a lot of time in the last several years de-emphasizing the importance of the Atom line of processors. I wanted to know if the AppUp store signaled a change of direction from Intel in supporting mobile oriented Atom devices. Peter was somewhat elusive about past marketing from Intel on the Atom, and he made it clear that Intel still considers its core processor technologies viable and important to the future of the company. He stated that the PC had been “declared dead” many times before and that the PC platform remains dynamic, flexible, and at the right price point with the widest range of available software. In his words, PCs are “clearly the world’s best software socket… effectually run[ning] the widest array of software.”
But he went on to say that Intel has acknowledged observing a change in software consumption. Device-centric apps, console apps, and phone apps suddenly gained incredible momentum. The traditional PC software market suddenly became Office and “web oriented” consumption. The AppUp store is Intel’s response to try to help the “small developer” who has “no infrastructure in place today to help you reach 10s or 100s of millions of end users” — for what he called “general purpose” PCs. It helps to understand, from Intel’s perspective, a general purpose PC could be the latest Intel Core processor or an Atom; they share a compatible instruction set.
The goal of the AppUp store according to Peter is “connecting creators to consumers via curators.” He spoke about developing specialized app store markets, and he gave examples (app stores that cater to working moms, to back-to-school apps, to hip 20-somethings, to productivity apps); he said AppUp was designed to provide a framework to help realize this. The theme seemed to be that Intel doesn’t want to control the AppUp store in the same way that Apple controls its marketplace. I got the sense of an old world market or bazaar. Someone owns and operates the framework, but not the actual individual shops that sell from this location. Peter touched on the concept that Intel realizes that in order to remain viable, it will need to develop a strategy to bring familiar mobile-apps to the traditional PC market, and AppUp seems to be its initiative to help achieve that goal.
Is the AppUp store the right kind of business for Intel?
I asked Peter if an app market is the right kind of business for Intel. My experience with Intel is that the company usually doesn’t do well when it strays from manufacturing, its core competency; this is something that Intel freely acknowledges within its company culture. When I think about how the company’s forays into other business segments have traditionally had poor results, I immediately think of the Intel Play line of consumer electronic toys.
Peter acknowledged this, but he stressed that Intel has an open roadmap for achieving success with the AppUp market. According to Peter, the equation Intel wants to solve is enabling the developer community to build excellent client apps that are highly marketable/saleable and that are not difficult to test and deploy. To this end, Intel wants to work with its hardware customers, OEMs, the telcos, and infrastructure partners (for example, Adobe, whose AIR platform will be a key part of the AppUp experience). Intel considers this ability to partner with best in industry partners a core asset, and thinks that this same perspective applies with the AppUp store.
“We can support essentially an infinite number of app-stores.” Intel wants to build the market and allow different storekeepers to focus on their core market demographics. If Intel can be successful running its own AppStore on a similar model to Apple’s marketplace, they will; but if not, Intel wants to provide the infrastructure that allows other vendors to compete successfully in this space. If this provides the application content to sell Intel mobile device hardware, Intel wins either way.
Peter says, “But if five store vendors come along and use Intel infrastructure to completely kick our ass at running stores to service consumers, and they do it, you know, in partnership with Intel using our infrastructure, we on-board the apps, we work with the developer community, we make sure stuff works, we make sure no malware is there, we give the billing and monetization and marketing tools and they run away with it, that is great!”
According to Peter, building on the foundation of a web service gives this flexibility to adapt to these challenges. It adds direct-to-consumer value, but this same framework can be leveraged by partners and other vendors.
How will Intel build excitement about the Atom processor and AppUp apps?
When I asked Peter how Intel plans to build excitement about the Atom processor and AppUp applications, he responded, “The Atom is a screamer. We don’t need tricks to enhance the perspective of the user experience” in the way that Apple uses perceptual tricks to make iOS appear smooth and responsive.
There are fewer reasons to compromise when developing for an Atom processor, and Intel plans on investing heavily to continue to enhance this performance edge. “We’re Intel; we really know how to do that part really well.”
In addition, at the annual Intel Developer’s Forum, the Intel Atom Developer Program was renamed the Intel AppUp developer program. Peter said that although this “may sound like a nomenclature change, but [it] strategically shows where we are headed with AppUp as a developer platform that spans the entire range of Intel offerings.”
What about the app store’s maturity?
My next question was about the app store’s maturity. I visited the beta, and it was slow on a 1.6ghz Atom running Windows 7, and there was an obvious lack of “big-ticket” applications. No Facebook app, no Twitter apps (official or otherwise), and no DropBox, TweetDeck, or Evernote. Peter assured me that many of these major players were in negotiations with Intel and the AppUp team, and that at least one major application had already landed on the store. The performance issues are well known, and a new version of the AppUp store has been released. Intel is committed to continuous improvement of the performance and stability of the store.
Peter was also excited to reveal that at the Intel Developer Forum a partnership between Best Buy and Intel around “Blue Label” consumer PCs, starting with netbooks would be announced shortly. “You’ll see AppUp pre-loaded on devices sold at Best Buy.” Peter also stated that he does not think Best Buy will be the only retailer or vendor bundling AppUp onto PCs.
What are your thoughts about Intel’s position on the Atom?
Next I discussed my observation that IBM once dismissed the PC/XT as a toy that was not a threat to its long term dominance of the business computing market. This was a mistake that left IBM trying to redefine its core business for the next two decades. I drew an analogy between this misstep on IBM’s part to Intel’s position on the Atom.
Peter responded, “Legions at Intel are passionate about the Atom, including Paul Otellini, for whom the Atom represents the basis for where he wants to take Intel. Atom is a foundational pillar for Intel.” Peter stressed that the Atom plays to Intel’s traditional strengths, and that the company has the expertise, the facilities, and the R&D budget to make Atom extraordinarily small in the future. “The beauty is they are general purpose computing devices — all of them — everything you’ve ever learned [about Intel architecture devices], works on all Atoms.” He called this, a “no compromise kind of thing.” Atom “gets tiny, runs cool, doesn’t need a lot of juice” and grows “more powerful than today’s consumer class netbook PCs, and is a general purpose PC that would run Windows 7 if that is still around.”
Is Intel still relevant?
I asked if Intel is still relevant in a “post PC creation, tablet digital content consumption” society. Peter feels that with its current technology lead, Intel has a 2 - 5 year “free ride” of remaining relevant in shaping technology patterns if they only did the same things they are doing today. But with their ability to engineer ever superior core processors, by continuing to make massive investments in keeping that true, “we’ll be small, we’ll run cool, we’ll be battery efficient and we’ll be incredibly powerful.”
1. The platform must be “sexy” or define the “digital tribe” that the device community associates with. Is it fun, easy, self-sustaining? Are there people on forums who can help you with issues at 2 AM?
2. Size of the total addressable market for a single given library.
3. Money. But Peter says it isn’t just money, it is recognition. Developers, and creators in general, aren’t always in it just for the money — they want the “creative outlet, for that moment, even if it is only 13 people, when they can delight some fans.”
4. and 5. Validation and distribution. “Creators don’t want to think about those two things.” They want to create.
Is an Apple/Intel rivalry or an ARM-Cortex/Intel rivalry in the future?
My next question was if he thinks Intel is likely to find itself back in an adversarial position with Apple, or if the focus from the AMD/Intel rivalry might shift to an ARM-Cortex/Intel rivalry. Peter feels that the small form factor is hot for the foreseeable next 5 to 10 years. He quoted the President of Qualcomm [Steven R. Altman] saying, “it is up to us to get software right, Intel to get the hardware right. Whoever does that, if either of us can do that substantially earlier than the other, we’ll have a distinct advantage.” Obviously the implication is that the AppUp focus is Intel’s initiative to get the software side right substantially earlier than Qualcomm can get the hardware side right.
Peter went on to point out that the ARM market is actually incredibly fragmented. He admitted, “they clearly own that space,” but that although it “may look like a billion devices” (from a core architecture perspective), it really isn’t. “Are you going to do BlackBerry, iOS, or Android? They’re all completely different.” Beyond that, you’ve got proprietary APIs for specific hardware vendors and/or the Telco. The Intel Architecture hardware advantage is a unified instruction set. Code that ran on a 486 is instruction-set compatible with Atom processors, and in Peter’s words, “we think that is really powerful.”
I’ve reviewed the interview several times, and I’ve thought about it quite a bit. Peter makes some strong arguments, and there is no denying that he conveys Intel’s passion and desire to remain competitive and relevant in this market.
I still have my doubts about Intel’s ability to operate a developer-direct-to-consumer retail market bridge, but I think Intel’s honest approach — which amounts to “we know we’re not the best at that, but we want to put the framework in place to let those who are good at it deliver it” — is a generally positive direction. Intel is good at nudging partners in the right direction, and letting them handle the things they do better than Intel. That part of the plan is probably the most reasonable direction Intel could have taken to compete with the other mobile consumer device platforms and markets.
On a personal level, I’m happier when my mobile device apps are available on my desktop PC as well — the more cross platform the better. In particular, I’m a big fan of apps that are effectively device agnostic, like TweetDeck and Evernote. I love these apps that I can use on my iPad, on my Droid 2, on my Windows PC or netbook, and on my Mac. I got the impression from Peter that Intel gets this, and the AppUp focus is designed to encourage and facilitate that type of use from end users.
On the surface, the argument about legacy instruction sets seems to have lost a lot of impact. I think Peter touches on that when he notes that traditional PCs have become “web content consumption and Office devices” early in our interview. I have no personal PC software purchase plans in my future other than those that enhance my mobile device experience. From that perspective, I don’t care if my software is legacy compatible with Intel Architecture, or if I am able to run native legacy Intel Architecture apps on my mobile devices.
But as you dig deeper and focus on the developer’s perspective, Intel’s argument makes more sense. It isn’t so much about the library available to me as an end user, but about the pool of skill, knowledge, and expertise with coding to Intel instruction sets that already exists in the industry, and about those binaries having access to an enormous amount of devices already out there. As I listened to Peter’s answers, I couldn’t help but think that the AppUp store might drive a fork in I/A86 applications. We’ll see those traditional apps that run on a desktop GUI interface — and a growing pool of apps designed for mobile, touch-interface devices. All of those apps will run on any device that has I/A86 (and I/A64) instruction sets, but one fork will be more appropriate to one class of device, while another fork will be more appropriate to another class.
There is no doubt that Intel has the resources and experience to make Atom the highest performing processor available for mobile devices. The question is: Do we need that? Hasn’t the adoption of netbooks, the iPad, and mobile phones as replacement mobile computing devices proven that the speed gains are beyond the development cycle or consumer needs? In the past, when an Intel core processor had a quantum leap forward (e.g., from the 286 to 386, from 386 to 486, from 486 to Pentium), we saw a huge evolution of the apps that were delivered. Since the Pentium 4, that seems to have hit a wall of diminishing returns. The iPhone and iPad illustrate that the perception of a smooth, robust, fluid user experience is often more important than the raw power of the processor that the device runs on. Peter’s argument again seems to be that this isn’t about the consumer’s experience, but about enabling the developer to disregard worrying about these things and focus on actually creating apps.
At one point in the interview, Peter and I got side-tracked in a discussion about Windows 7 and Microsoft’s continued viability and relevance. At one point, Peter said, “If I were a betting man, (and I’m not), but I don’t think I’d be putting money down on Windows being dead.” I’m with Peter on this one — only with Intel as the subject.
Peter concluded the interview by saying, “this game is not over; this game hasn’t actually begun,” and he’s right. It will be interesting to see what plans Intel brings to the game.
Tell us what you think
Is Intel a dinosaur that is hopelessly out of touch with the direction the industry is headed, or is it a sleeping giant just waking up with the potential to redefine our experience with mobile consumer devices? Share your thoughts.