The risk of bricking, and the lessons that people need to learn

I read several excellent synopses from SiCortex former staff such as Jeff Darcy, and Matt Reilly. Jeff and Matt gave several excellent arguments as to why SiCortex succeeded, despite getting the plug pulled. Some will guffaw about this, and say that pulling the plug was evidence of failure. They would be wrong. VCs will pull plugs everywhere from pre-term sheet to capital call time. Their rationale for pulling plugs won’t always make sense, to you, or others. Or sometimes to anyone.

Vipin and I once had a term sheet being typed up by a VC for our accelerator play, in 2005. And it was pulled. Since we weren’t in Seattle. That was the only reason. We were willing to move too. Given where accelerators are today and the impact they look like they are going to play in the market … yeah, 4 years later, it still kinda hurts. I can empathize with the SiCortex crew. I don’t know their exact pain, but I understand it. Pulling the plug as the VCs did was not, from my understanding, associated with their business state. They hit their milestones, demonstrated growing revenue streams. The lessons that people will take away from this … I already see some thoughts being voiced that don’t jive well with reality … The fundamental risk in any customer purchase of a technology is, will this technology become a doorstop next week courtesy of the company being bought, changing focus, going out of business. The first incorrect lesson I see various writers espousing is that this will cause customers to be “more conservative” in their choice of vendors, with others explicitly talking about sticking with large vendors. This one is incorrect in pretty much all aspects … end users have to ask themselves if a technology will become a brick if the company has “an event” (illiquidity, purchase, change of focus). With open systems, there is no risk. No, I am not talking about faux-open systems, the marketing phrase “open systems” as applied to proprietary systems that the sales and marketing force want to sell as “open”. Lets ask all the customers of the “conservative vendors” such as, I dunno, Sun, SGI, LNXI, … exactly how they feel about having bought proprietary gear from the “conservative companies”. You know, too big to fail. Right now, many Sun customers are rightfully worried about the future of their systems. They should be, after the close, they won’t likely be long for this world. And worse, if it doesn’t close due to regulators. Think SGI-like implosion. All that proprietary hardware. Turned into a brick overnight. Think about all those Chrysler customers. What if the deal doesn’t go smoothly. What if the creditor protest results in a dissolution rather than an emergence? Lots of nice new bricks, driving on the road. All the owners of these 4-wheel bricks wondering how they are going to get service and support. This … this is the real issue. SiCortex going down means, sadly, there are a number of customers with very cool bricks. They are useful until they break, or you run into a bug that you need their help with. When I was at SGI, we had an Intel based desktop unit. I differ from most of my colleagues in saying this was a good direction for a number of reasons. And we killed it. Just when it started taking off. Our business types pulled the plug. Not for reasons I could really fathom. Later, the event that caused me to give up on SGI, was when Warren Pratt pulled the plug on the x86 servers, effectively leaving the cluster market in 2001. We had developed a product called GenomeCluster atop it. Demoed it at SC2000, at LinuxWorld, at GSAC, … Our performance was great. Sold a few. Even have a patent (well, this is Rackable’s now). And we were forced to throw in the towel. Which meant our customers were “bricked” on the hardware and software side. See the pattern? It actually has nothing whatsoever to do with the size of the company. It has little to do with the business conditions of the company. It has a great deal to do with what the people with the control over the funds want to do … or don’t want to do. So when I hear that SiCortex going down will strengthen the argument to reduce the size of the IT supplier base … I have to ask if they are seriously suggesting customers buy more stuff from Sun, or SGI. I won’t name names. Another very large TLA multinational company has basically decimated its HPC groups. It is, apart from some very large projects, reducing its HPC footprint. This is a business decision. My friends affected didn’t like it. But those with the control over the capital, and how it gets spent are precisely the ones making these hard calls. Now … please understand, we don’t fathom the reasons why they make these calls. They may have information we do not. They may have conditions we cannot fathom. We cannot blame them for making bad calls. We can be annoyed with the calls, but you know, someone has to make hard decisions. I am not sure what decisions were made at this TLA, but made they were. VCs are a different animal. Read thefunded.com for some open and anonymous discussion of the vagaries of working with VCs. Some of this is appalling. Some VC are apparently happy to cut bait very quickly. Then again, think of this market. VC’s have to go to their LPs to get money. In this market, LPs may not have as much money. The LPs may be hurting and might not answer the capital call. So the VCs see their funds not having full capitalization, and they have to make hard decisions about where to spend it. What to protect. What to cut bait on. I don’t have information from the VCs, but Polaris is not known to do this. They are one of the better ones. The team SiCortex got capital from are a pretty good team. VC investment has dropped something around 30-40% nationally over the last 2 quarters. Many are in protect the assets mode. Their rationale may be inscrutable. But I doubt it was nefarious. They had to make hard choices. As Jeff, and Matt pointed out, they made a run at it, ran hard, and proved a point. They have altered the landscape of HPC. The risk for HPC buyers is significant, but its not just in computing. Its in storage. In fabrics. Quadrics is dissolving. Any Quadrics shops out there? Big one at Sharcnet in Canada. Others? Probably. Quadrics fabrics are bricked. The business risk in HPC, in storage, is the risk of bricking. The performance risk is always there. If you build a business dependency upon a brickable technology, you need a plan “B”. We made this point in cloud computing. There are lots of fans of cloud computing, and we like some of the concepts. Reduce the cost to stand up new resources. Reduce the time to do this. The bricking risk is more of a data/capability lossage here. There are access/control risks, jurisdictional risks, and other risks. But the point is that you can’t claim that working with “conservative vendors” will somehow ameliorate risk … Network.com? The S3 outage? No. Bricking happens. How you can continue operating when a critical aspect has been knocked out from under you? How long can you operate on that platform knowing it is unfixable? This is sadly why all those customers of conservative Sun, are scrambling to find alternatives. Because “conservative” does not mean safe, nor does it mean risk free, or even reduced risk. It is a mistake to think otherwise. SiCortex is gone. People with their boxes are worried. Use them until they break. Maybe we will have options for them. The market is many things. Unforgiving is one of them. But the market is also eminently creative. A need exists for low power consumption high performance tightly coupled (but not SSI) HPC computing systems. SiCortex proved this. They not only demonstrated the market … they serviced and grew it. That market isn’t going to go away just because they went away. Bricking happens. Risk is a function of decisions made by those in control. Not by the size of the organization. Bricking is a risk when decisions are made. Being conservative doesn’t reduce risk.