We first came across the idea of a video kiosk about five years ago, and were called in to talk to a client about how to implement it. When we got to the bottom of the matter what we found was that the client company had a technology for re-engineering the back end of flash memory so that it had parallel write paths into the heart of the memory. This meant that it could write data instead of 12 Mbps onto Flash at 1 Gbps or 150 MBps.
Even today five years on that write speed is impressive, but conventional NAND block based Flash memory is catching up and we have heard commercial products that accept write speeds up to 100 Mbps.
Our position was, license this to the flash memory companies and then take a long holiday, but the client’s position was that there had to be an ideal application and he thought it was video kiosks. Eventually he ran out of money and he would still be pursuing that same kiosk scheme today if he hadn’t.
This is an example of the technology driving the business idea, and this week once again we have Blockbuster offering the same idea 5 years on. If we go back even further IBM and Sears Roebuck tried the same concept in music CDs 20 or more years ago, and it failed for much the same reason.
Let’s think about this. Most people want to play video from a DVD, and even if you solved the DVD writing problem (it can take 30 minutes to write a DVD) pretty soon those same customers would want Blu-ray (how much longer will that take to burn, say 1.5 hours, or a 4 x burner might take the same 30 minutes).
So do kiosk owners want to buy a bank of Blu-ray and DVD burners and put up a web site where they can be pre-ordered, and burned while customers are on their way to the store in the morning. Or do they want to stream the film online or just use Netflix and order one that is already burned and which has been tested?
And there are other issues that will show in the cost of such a service. Instead of each Blockbuster store only having 500 selections available, it will suddenly have 65,000 or some such available. But can you imagine how careful the content owners are going to want to be with a raw unwritten DVD file. This will have to be on a big SATA disk drive, with a duplicate for back up, encrypted and watermarked, with a 20 Mbps online connection or a satellite download delivery to each location, for delivering updates. It will also need to be bricked into the building, to avoid the temptation of a ram-raider dragging the whole device off so that a pirate has 65,000 copies of video that cannot be traced on the internet.
But of course the Blockbuster system isn’t even going this route and instead wants to write initially to an Archos device which comes in two flavors, 30 to 80 GB hard drives and a handful of GB in Flash memory. So we are back to those flash memory copy speeds.
Even with special Flash that allows fast writes, a film in SD might take 15 seconds. But these are not going to have special fast write NAND Flash and it is likely that the write speed for an hour of SD will be closer to 2 to 3 minutes and might be as long as 5 or 10 minutes. Blockbuster says that over time it hopes to get this down to 30 seconds. Now it is entirely possible that a hard drive version of the Archos devices would write considerably faster, but if we assume that each write takes 3 minutes, and that there is a period of searching to be sure that the right file is available etc.. then every transaction will take 5 minutes and a queue will form and a device may only manage 12 transactions an hour, or less than 100 a day. What’s the profit on 100 video rentals and is that more or less than the daily Opex plus capex interest charges on a kiosk.
The DVD route was always a problem, because someone has to fill the device up with blanks each day and that creates new security issues, so the flash and disk write method is best. But exactly how many Archos devices are there out there? Will these have to be rented by the store and given away to customers or will this create another capex cost and a potential Opex drag (as devices get stolen) or is Blockbuster limiting itself to a tiny population of existing machines.
We should stop there because there are only so many things that a company like Blockbuster can say about itself to increase its share value, and some of them are just dumb, and we’re fairly sure this is one of the ones that is dumb. In two or three years someone, possibly not Blockbuster, will have solved the riddle of how to get video to people online, and then not only will the Blockbuster stores be a liability, but the kiosks will become so much junk.
The kiosk prototype was shown and was developed by NCR and to avoid a huge investment Blockbuster says that it will rely on third-party partners. Let me see if we heard right? Someone else will take the cost, and therefore a good piece of the revenue for this, and take customers away from Blockbuster? Is that right?
It is true that Redbox in the US already has DVD rental kiosks in 6800 locations around the US, but it is not paying rental for an entire store but is parked on Wal-Mart or Walgreen real estate. These already have a huge amount of retail foot traffic and rely on impulse purchase. So soon Blockbuster will be copying Netflix to compete with Netflix, copying Redbox to compete with that, and still has no clear online advantage for its own Movielink which it purchased last year from the studios to go wholeheartedly up against online film delivery.