The "Connected Store" and the future of retailing

Intel and a number of partners, including Adidas, Procter & Gamble, Best Buy, Kraft Foods and the MIT Media Lab, are showcasing a two-story, 2,400 square foot "Connected Store" at this week's National Retail Federation Convention in New York. Intel is using large flat-panel, touch-sensitive displays to showcase how its processors can be used to improve the retail "experience".

The Adidas demonstration uses three displays to showcase Adidas' entire shoe line. Shoes that aren't in stock can be delivered to the store for pickup or directly to the customer's home within 48 hours. The Best Buy demo, built in conjunction with MIT's Media Lab, enables customers to get a demonstration of any product on any vertical surface in the store.

These demonstrations go far beyond the digital signage systems found in many retail stores, in that consumers can interact with them instead of simply passively watching them. These systems could help drive a trend to downsize retail locations. For decades in the U.S., there's been a proliferation of "big box" stores, led by Walmart, Target, Costco and Best Buy. The problem is that real estate is leased based on square feet, and bigger stores means higher real estate costs, as well as higher costs for utilities and more employees needed to staff the physical space.

The "big box" cost problem is very clear at Borders and Barnes & Noble. They've built bigger and bigger bookstores, even as print book sales have declined. Many consumers still like to look at physical books in bookstores, but they then go home and purchase them online as eBooks. Bookstores can recreate the "books on bookshelves" experience using much less floor space with big-screen touch-sensitive displays. Consumers can open "books", page through them, and make a purchase decision right there. eBooks can be sent to their portable devices for immediate download, and the sale is made by the retailer, not by Amazon or another competitor.

These systems will also enable consumer electronics retailers to shrink their stores. They can keep the most popular products in stock in the stores for immediate delivery, yet let consumers interact with and order the same assortment of products that they already carry on their websites. One might argue that consumers can do that already from home on their personal computers and tablets, but closing the sale in the store means that those customers won't compare prices with other online merchants and buy at the lowest price.

In short, the era of the "big box" store isn't over, but "little box" stores enhanced with interactive digital technologies will take their place in many markets and segments.

Intel's Sandy Bridge...and a little Mac (and Final Cut Pro) speculation

At Intel's IDF Conference, the company provided more details about its new Sandy Bridge architecture. According to Anandtech, the first Sandy Bridge processors will be shipped for performance-level (gaming and media creation) PCs in early 2011, and will migrate to entry-level PCs in 2012. One major change in Sandy Bridge is that its GPU core should provide performance easily double that of Intel's existing integrated graphics. It won't keep NVIDIA or AMD up at night, but it should be good enough to lessen the demand for add-on graphics cards in entry-level PCs.

Sandy Bridge has an integrated MPEG2, VC1 and H.264/AVC decoder that Intel claims will use only half the processor power for HD playback as existing processors. It also has a AVC encoder/transcoder that, in a demonstration, was able to transcode a three minute 1080P 30Mbps video into a 640 x 360 iPhone video in 14 seconds, at a rate of 400 fps. This gets into the performance range of high-end, GPU-accelerated encoders. Sandy Bridge will also have an enhanced Turbo Boost feature that will allow the clock speed of individual cores to be boosted beyond the normal thermal design power (the maximum safe power dissipation of the chip) for brief periods of time.

Okay. so I said something about the Mac in the title, right? According to Anandtech, Core i3, i5 and i7 processors with Sandy Bridge architectures will ship in Q1 2011. Every MacBook Pro, iMac and Mac Pro ships today with at least a Core i3. Next, let's add Light Peak, Intel's new 10Gbps optical competitor to USB 3.0. There's no support for Light Peak in the first Intel chipset announced for Sandy Bridge, but that doesn't mean that it can't be added.

Add it all together, and it sounds like the Apple notebook and iMac product lines will be fully refreshed some time next year with a combination of Sandy Bridge and Light Peak. There goes the I/O limitations of Apple's notebooks and desktops. (Apple won't have to reengineer the Mac Pro right away--they can simply offer a PCI Light Peak card.)

Let's throw one more thing into the mix. There's been a lot of rumor and speculation surrounding the next release of Apple's Final Cut Studio, with a battle between bloggers--some saying that a new version will be released no later that NAB in April 2011, and others saying that it won't happen until 2012. The "2011" school says that the new version of Final Cut Studio will have Adobe Mercury Engine-like performance, but the "2012" school says that's not possible without major architectural changes. If Apple is writing the next version of Final Cut Studio to take full advantage of the features in Sandy Bridge, it most definitely is possible in the 2011 timeframe.

So here's my semi-informed speculation: In Q1 (perhaps late Q1), we'll see the first updated MacBook Pros with Sandy Bridge and Light Peak announced. New iMacs will follow. Then, in the April timeframe, just before NAB, Apple will announce the new Final Cut Studio that takes full advantage of the new computers ("great time to upgrade!"). Deliveries of everything will occur by the end of Q2.

New AppleTV uses A4 processor; what impact will ARM's A15 have?

According to the latest EETimes magazine, the new AppleTV uses Apple's own A4 processor in place of three Intel chips in the previous design: A Pentium M-class CPU, a memory controller with embedded graphics, and an I/O chip. Those three chips cost Apple $60 to $65 and took up 975 square mm of space. By comparison, the A4, manufactured by Samsung, costs Apple $15 to $20 and takes up 196 square mm of space.

The switch to the A4 explains why Apple won't port the new AppleTV user interface or features to the previous model. It also puts Intel at an increasing disadvantage in Apple's future product plans, since only the Mac family now uses Intel processors.

Another story in the same issue of EETimes has a great deal of potential importance to Apple's future desktop computer plans. ARM, the RISC processor developer that licenses its designs to a variety of semiconductor companies, disclosed a small amount of information about its new 2.5 GHz Cortex A15 processor. This processor is multicore-capable and uses much less power than the current generation of server chips from Intel and AMD. Apple could build a notebook- and desktop-oriented version of the A4 using the Cortex A15 as its foundation, and displace Intel from its product line entirely. The A15 is scheduled to begin shipping in 2012.

ARM to support Google TV, but will Intel let them?

According to Engadget, the president of ARM, Tudor Brown, said that ARM support for Google TV is "in the works, and now that they support Flash and Google Chrome, they plan to "optimize" for Google TV on Android. If it happens, it'll be a big boost for the Google TV platform, which to date only supports Intel's Atom platform. Atom is expensive and power-hungry relative to ARM's processors, and the availability of ARM processors will make Google TV much more practical and cost-effective to implement.

There's a question, however, as to whether Intel has veto power over other processor vendors joining the Google TV initiative. The way that the Google TV project has been run makes it look like Intel is actually calling the shots rather than Google. Consider that Google TV is being released on a down-rev version of Android without any support for WebM. If Google was running the show, you'd think that it would want to showcase WebM and use the most advanced version of Android, but that's not what happening.

ARM may be waiting until Google open sources the Google TV platform, which Google said would happen next summer. By that time, however, Intel could change some portion of the platform to make it work better on Intel's architecture, or not work at all on any architecture other than Intel's. So, just as Adobe kept saying that support for Flash on the iPhone was "coming real soon" until it was obvious that it wasn't, I'll believe that ARM is participating in Google TV when I see representatives from ARM, Intel and Google sitting side-by-side at a Google TV press conference.

Google TV: Right hand, meet left hand

When Google announced Google TV at last week's I/O Conference, one of the jaw-dropping omissions was that there was no mention whatsoever of support for WebM or the VP8 codec, which Google had announced with a great deal of industry support the day before. YouTube, the world's largest online video site, owned by Google, announced that it would make all its videos available in WebM going forward. At the Google TV announcement, however, YouTube's participation was limited to a strange "we pick the content for you" service that seemed to have no connection to Google TV whatsoever. Yesterday, an Intel executive said that the company might support WebM in the future, "if VP8 establishes itself in the Smart TV space," whatever that means. And, as if that isn't enough, Google TV will be based on Android 2.1, even though the company launched Android 2.2 to great fanfare minutes before the Google TV announcement.

When companies get big, groups stop talking to each other and bureaucracy takes hold. It's clear that the WebM and Google TV teams didn't talk to each other, or if they did, there was no adult supervision applied to insure that they worked together. It also looks like the Google TV and Android teams didn't talk very much, either. If WebM is such an important initiative, why isn't it in Google TV? From Intel's perspective, it may never be in Google TV. Why on Earth is Google planning to release Google TV with an old version of Android? Why did YouTube look like a "bit player" in the Google TV announcement?

Looking from the outside, this suggests that there are some serious management problems at Google. The WebM product group and YouTube are working closely together, but Google TV is completely out of sync with both WebM and Android, and Google TV is barely talking to YouTube. They're all in the web video business, but they're working at cross purposes. Intel won't adopt WebM unless "VP8 establishes itself in the Smart TV space"? "Smart TV" is what the whole Google TV initiative is about. How can VP8 establish itself if it's not included?

It takes successful startups to build more startups

There's a number of elements that are generally accepted as essential for a city or region to have a successful startup culture:

• Good colleges and universities, preferably with strong engineering programs, in order to provide an ongoing supply of qualified young talent
• A good quality of life that encourages graduates to stay in the area rather than relocate after graduation
• Availability of venture capital

There's another element that's less-discussed, but possibly the most important of all: Local, successful startups that get big enough to spin off other startups. This is one of the reasons that Silicon Valley has been able to maintain its lead for several generations now. Shockley Semiconductor begat Fairchild Semiconductor, which begat Intel, National Semi, Signetics and others. Those companies spun off other semiconductor companies, and the chain continues today. Hewlett-Packard employees founded Apple, Tandem and many other companies, and those companies have spun off countless other startups. Yahoo and Google have each spun off many startups. eBay acquired PayPal, and now the ex-PayPal team has built a new generation of startups. Twitter (which spun off from Google) has already sprouted Square, and many other companies will bud off over time. Facebook will have the same effect.

It takes a big success to first recruit and then spin off the founders of future startups. In Chicago, where I'm located, Groupon looks like it has the potential to spin off more startups. It's too early to tell whether that will be the start of a sustainable startup culture in Chicago, but it's a big move in the right direction.

Intel: We'll leapfrog USB 3.0 with Light Peak

Intel just demonstrated its new Light Peak optical interconnection technology running on a laptop, at a technology showcase in Brussels. For the demonstration, Intel integrated Light Peak fiber into a conventional USB cable (although the connectors are obviously very different.) The initial bandwidth of Light Peak is 10Gbps, more than three times as fast as USB 3.0.

The Brussels demonstration showed Light Peak sending two separate HD video streams to a television monitor. Intel claims that Light Peak will eventually be able to replace every PC/peripheral interface, including HDMI, USB and DisplayPort. The company also claims that the first Light Peak hardware will ship to manufacturers by the end of the year.

This explains Intel's hesitancy to include USB 3.0 in its chip sets. If it believes that it can start shipping Light Peak by the end of the year, introducing support for USB 3.0 now would simply "muddy the waters", especially since it's highly likely that Light Peak hardware is going to be more expensive than USB 3.0. The critical questions are: Will Intel make its end-of-the-year deadline, how much will it cost for computer and peripheral manufacturers to implement Light Peak, and when will Microsoft and Apple implement support for Light Peak in their operating systems?