Blog
20
May
Brazil as VC hotspot?
Chris Morrison over at VentureBeat has just posted an article about Brazil being a new venture capital hotspot. I can’t say I’m surprised. Last year we did a project for a major F500 firm about key software growth markets, and Brazil was one of the hottest.
There are some amazing stats about the Brazilian software market:
- $13 billion market
- 15% year over year growth
- Tech spending expected to double by 2010
- 6000 new PhDs in Computer Science per year
- $400 million of investment in 2007
Obviously something is going on in the Brazilian market and, to add fuel to the fire, Brazil has just announced that they would use their sovereign wealth fund to make venture investments. Not that the tech business really needs any incentive, as it’s been calculated that losses due to lack of IT investments in agribusiness, textiles, health, electrical appliances and auto making cost the country around $21 billion/year…
One of the interesting things about Brazil has been that the government has been very open about encouraging the use of Open Source throughout Brazil’s computing infrastructure. They also have a target of having $2 billion in tech exports by 2009, something which may come from Brazil’s leadership in several green tech areas, notably fuel biomass production.
Given all this, and their recent discovery of huge new oil fields, Brazil may yet become the first global Latin American powerhouse since the 19th century.
18
May
Failure: Open Source & OLPC
I never thought much of the One Laptop per Child effort. My belief was that spending so much effort and capital on technology could be re-direct to better uses, such as food and books (notably The Book Thing in Baltimore, MD which has shipped thousands of books to emerging countries…). The other part of my skepticism of the OLPC effort was that anyone could actually do this effectively. This was further reinforced in Spring 2007 when someone showed me a OLPC computer. To say I was less than impressed is an understatement.
Now, Ivan Krsti?, the ex-CTO of OLPC, has written a long rant about what happened to the OLPC project. Not surprisingly, perhaps, it has largely been a victim of Nicolas Negroponte’s ego and hubris, which is useful at times, but, in this case, has turned OLPC into a sham. Ivan makes several very good points about OLPC and Open Source in general, particularly in regards to real-world usability, but his core theme is that OLPC has turned into a mainstream laptop manufacturer with zero support backend (and running Windows…).
However, there are two other memes in Krsti?’s post that the Open Source community should pay attention to. The first is that changing computing paradigms is not always the best thing to do. OLPC tried to change too many things at once, everything from general computing paradigm (time based instead of files and folders) to the UI (which emphasized circles of activity and communication), resulting in a hugely complicated software design fraught with issues:
- "Choosing to reinvent the desktop UI paradigm means we are spending our extremely overconstrained resources fighting graphical interfaces, not developing better tools for learning. …"
There is a real lesson here that if the goal is to do something radical, better focus on doing just one radical thing (e.g. one laptop per child – literally, even if they are used laptops running Ubuntu & Gcompris), rather than 20. Comes back to that focus thing I discussed in a previous post….
The second, and perhaps most damning, meme that Krsti? brings forward is the "elephant in the room" that no Open Source person really wants to discuss, that Open Source desktops suck:
- "About eight months ago, when I caught myself fighting yet another battle with suspend/resume on my Linux-running laptop, I got so furious that I went to the nearest Apple store and bought a MacBook. … As one of the people who actually can hack my kernel to suit, I find that I don’t miss the ability in the least. There, I said it. Hang me for treason."
I can empathize. Over the years, I have gotten flamed for running Windows on my laptop, particularly while doing conference presentations. I’ve also gotten flamed for saying that Linux on the desktop was not ready for prime-time. Now, like Krsti?, I’ve been a Linux user for about 15 years, but I won’t run it on my laptop as it’s almost impossible to get everything working correctly at the same time (try hot swapping a DVD drive during suspend…) and I don’t have time to fiddle with every single package to get things running just right. I would note that the problem seems to have gotten worse in the last 5 years as hardware capabilities have grown.
To be fair, Krsti? does point out that this is not the Linux community’s fault, but a result of hardware vendors not releasing specs. Regardless, the net effect is that Linux desktops are partially broken, particularly on laptops. And a chunk of the Open Source community is just not interested in solving this as it does not fit into their philosophy of OSS to work with proprietary anything. That’s great if you live in a vacuum, but, in the real world, these problems have to be solved regardless. And, right now, the Linux community is largely failing to solve them. Perhaps a non-profit like the Linux Foundation could step in and fix this mess…
13
May
Tesla & why startups fail
Well, I posted something on Sunday about Tesla and, lone behold, they are in the news again with more internal turmoil.It seems that Martin Eberhard, the visionary [ousted] founder of Tesla was promised car #2, which has not yet been delivered despite the fact that car #3 has…
Whether this is due to internal production incompetence or a desire to further punish Eberhard is largely immaterial, what is clear is that Tesla is not capable of delivering a car as promised and they are terrible at communicating issues. Carolyn Eberhard (Martin’s wife, I think, but I don’t know either of them) put it best in a comment "… not one person at Tesla felt the need to pick up the phone and tell Martin about the "defects" and "delays." He had to learn about it through the Tesla blog. This is not great customer service." No, not great customer service, and, worse than that, perhaps a sign of impending failure and certainly a sign of very, very poor planning. Not good.
I did a presentation some time ago about why startups fail. It had three slides:
- Do whatever it takes – Most startups don’t
- Focus on your customer – Lack of focus will kill you quickly
- Remember the exit – Your investors care about this above all else…
It strikes me that Tesla is at least breaking rules one and two, and probably three as well. Car companies live in die by customer experience, because, in the end, all the brand and lifestyle attributes can’t make up for the fact that the car won’t start, rattles or the dealership is rude and unethical. Mercedes found this out the hard way.
This is exactly the kind of thing that will kill a car company faster than anything else. There is no excuse for the lack of communication and failing to follow through on promises, particularly to the company founder (whatever the internal politics). Like I said in my previous post about Tesla, they are off to a bad start, and Fisker looks likely to eat their lunch, at least in the larger four door market. That is, if Fisker follows the rules.
11
May
Tesla vs Fisker – The battle for the electric car…
I was at Tesla in March, hosted by Kurt Kelty, their head of battery technology. As I’m sure most people who are following Tesla know, it’s basically a stretched Lotus Elise with a different body. This was largely made possible by Lotus’ use of aluminum extrusions bonded together (rather than welded or mechanically attached). Two basic things struck me during my visit to Tesla.
The first was the complete lack of any sports cars in the parking lot. Most of the cars were either family cars, outdoor sport enthusiast cars or hybrids, which is very strange for a company building a sports car. Apparently, not many car nuts work there. This would be OK at some level, but part of the reason GM and Ford lost their market is that they stopped building cars on gut instinct and started building cars purely based on financial metrics. Fundamentally, a combination of the two is needed. Most cars are not appliances, they are often part and parcel of people’s identity and that requires a certain amount of ‘soul’. Companies like BMW and Mazda have really taken this to heart and it has translated into market success. GM, with the aggressive guidance of Bob Lutz, is getting back to building cars with ‘soul’, like the new Pontiac G8 and the Cadillac CTS-V. Even though most consumers won’t necessarily buy these cars, they represent brand aspirations and image that matter to buyers of less performance oriented models.
The second was that Tesla designed and built a USA-certified car for less than $200 million. Just to but this into perspective, it has traditionally cost car companies $2-3 billion to do this, and Tesla, with no experience, has done it for $200 million. Sure, part of this is due to the shear brilliance of Lotus’ extrude bonded chassis, but that’s not the only story. Most of the investment in fact went into power electronics, specifically the torque vector control motor electronics, the motor and charging system. Still, $200 million for total development costs (including setting up the company and marketing) is peanuts in the car business.
What does all this have to do with Fisker? Well, for Tesla to bring on a true revolution in the automotive business, it must move beyond selling 10,000 exotic sports cars. Their are plenty of companies out there that have been successful in the low volume, high price market, most of them in Italy and the UK (Zonda, Aston Martin, Marcos, etc). The real breakthrough will be to build 100,000 mass market cars. To do this, Tesla needs to bring both it’s electronics engineering expertise and appealing design. For the latter, it had hired Henrik Fisker, one of the most well known current automotive designers, and generally regarded as the designer of some of the most beautiful and desirable cars currently on the road. This relationship has since ended in a lawsuit. And it’s not surprising.
While working with Tesla, Fisker was working on his own mass market electric car, the Karma. The Karma has been seen testing a prototype recently and is due to be available in the next 12 months, likely before Tesla has their mass market car ready. So what does this mean for Tesla? Well, Fisker is capable of building a car with ‘soul’, he has done it before. Fisker has demonstrated a working prototype of the Karma and has a track record of delivering roughly in stated timeframes. Remember the parking lot I mentioned earlier? If Fisker comes out with a mass-market car before Tesla, Tesla might be relegated to a low-volume, high-dollar niche manufacturer, without the drive of car nuts like Bob Lutz. At that point, GM or someone else will probably buy Tesla, although GM might not need to since they are about to launch a rival of their own, the Volt.
And remember those exotic motor controls Tesla spent millions engineering? Turns out a bunch of people on the internet had already done it. Fundamentally, the real lesson with Tesla is not that it’s possible to build an electric car, but that the barrier to entry to the car market has been lowered to $200 million, which is almost VC money. And that’s a revolution.
09
May
Skype, the GPL and the courts
As many of you may have heard, Skype today dropped an appeal of their GPL violations case, accepting a lower court decision that they had violated the GPL (the details of the case are available here).
What I don’t understand is that, after the Linksys/Broadcom fiasco, how could anyone think that it was OK to ship a Linux-based device without making the source available. I fully recognize that sub-contractors may put companies in a difficult position by doing things that are not entirely in compliance with licenses, but still, you would think companies would pay more attention.
Of course, the reality is that everything is lax until something happens. We have done several large engagements in the last few years where enormous amounts of time was spent search the entire codebase for open source so that it could be documented and appropriately managed. Most of the time, it was because of a shocking lack of control over incoming code and a failure to properly tag sources of code. Not evil, mind you, just sloppy.
Such sloppiness can be particularly dangerous when it comes to consumer products, as by the time there are hunderds of thousands of devices in the field, it is impossible to do any sort of remediation work.
Open source licenses are not impossible to work with in a commercial environment, but they do require more care and attention. A nifty side effect of this is that they enforce good coding and code management practices, something everyone needs and should be happy to have.
05
May
The coming Linux desktop?
I’ve gotten a number of questions in the last few weeks about Linux desktops and OpenOffice. These have come from enterprise users, the press and venture capitalists. It’s interesting, because this seems to be a meme that is revived every few years. I was really enthused about the possibility of an Open Source desktop rival to Microsoft’s offering several years ago, but there never seemed to be much traction. Clients of mine that seriously looked at Open Source rivals to MS Office did so to negotiate better pricing from MSFT as much as anything else, no one actually migrated to it. Oh, sure there were some high profile attempts (like the city of Munich) and at least one Army command migrated to OpenOffice, but it did not really catch on as I expected.
This largely caused me to examine why there was no widespread adoption and what might eventually drive adoption. What I found is that feature/functionality parity in the office suite is just one part of the equation and probably a small one at that. The reality around migration is that it’s much more complex than the Open Source community assumes and it’s not just a technical problem. The depth of this issue only begins with deploying a new office suite. You then have to train people, provide tech support and tie this new suite into the rest of your systems. If you are an SMB or a consumer, chances are you are using off-the-shelf versions of MS Office and this might not be too painful. But if you are an enterprise (say, Ford Europe with 15,000 desktops) then you are probably not using generic MS Office and these issues take on a whole new dimension. One client of ours has tens of thousands of desktops deployed where MS Word has been scripted to enable customized correspondence with clients, all while controlling the content. What would it take to port that?
And that’s just the thin end of the wedge. When you look at other issues, like having IT staff that’s trained to support both OpenOffice and MS Office, or issues dealing with common authentication, then the problem set becomes logarithmic. Good businesses will do ROI calculations on this and the picture does not look good from that standpoint. If you look at Ford Europe, for example (I’m just picking on them, I don’t have any special knowledge here), it’s easy to see why the incentive to migrate is not that high. Assuming that MS Office is the target for migration, lets run some quick napkin math. Assuming that half of the desktops are engineering centric and can’t be migrated, there are 7,500 potential migrations. Of those, probably half again are using applications only available on Windows, so we have 3,250 desktops available for potential migration. Most enterprises pay around $60 per copy of MS Office and the replacement cycle is roughly 3 years, so that represents around $195,000 over 3 years, or around $65,000 a year. If you just look at the cost of building a tech support team for OpenOffice, that is likely to cost around $400,000 (retraining 100 support people + hiring 5 experts) with around $300,000 of yearly recurring costs. That means that it would take 6 years to recoup the cost of deploying OpenOffice just based on the cost of providing support (and that assumes that you would eliminate some support positions on the MS Office side to offset to cost of the ‘experts’). And if you add end user training, integration costs and other issues, that lengthens the ROI timeline even more.
Yes, I know. It’s depressing to inject this dose of reality in what is otherwise a very, very good thing. However, after seven years of doing Open Source strategy for a very large number of different users (and as an Open Source user for 15 years), I’ve come to realize that there are certain hard realities that are difficult to overcome. These kinds of calculations represent one of the hard realities. And I haven’t even mentioned integration or maintenance.
05
May
Linux only 10% of US datacenter installs?
I was going to mention this earlier, but when Levanta closed, there was a really interesting quote on CNET: "In the U.S., there’s only a 10 percent penetration rate for Linux data centers."
Really? Only 10%? I mentioned this to Tim Golden, VP of Unix at BofA and their lead on Open Source and he didn’t think it was right either. Other people I know in the financial sector have large amounts of Linux in production and if you look at Google, Intel, Amazon or Oracle, all of which have large datacenters, they also deploy large amounts of Linux.
That said, I guess it’s possible that there is only a 10% penetration of Linux in datacenters, but it seems awfully low given what I know . Then again, I do have a relatively Open Source centric view of things and most people I deal with are relatively Open Source savvy. I do know that the dirty little secret of LinuxCare back in the day was that most of their revenue was coming from Windows support, which they needed to do in order to get the Linux portion of support contracts….
I actually know some VCs who invested in Levanta, perhaps I’ll get the inside scoop on this quote someday.
28
Apr
Open Source vs Intellectual Property Misconceptions
The other day, while flying home, I picked up a copy of the Harvard Business Review (HBR). The cover had a teaser about an Open Source strategy article, so I thought they might have something interesting to say.
The article was a (fake) case study* about a CEO worried about Open Source stealing the IP of her company. Great. Here was HBR repeating one of the greatest falsehoods about Open Source, that it destroys intellectual property. It went downhill from there, descending into a ‘all these hackers are stealing my stuff’ manifesto. Read more…