current leading-edge in data centers will produce its own hot startups and segments, and it will also push many existing segments and startups off the [hot] list.
Fair enough, needs drive innovation. The “must have” phenomenon. Build something that remarkably alters the economics to be strongly positive for customers to acquire and use over their existing technologies, or reduce their pain points so that they can save lots of money due to the secondary effects of workload reduction/automation.
I agree with this. Quite strongly.
Then his second point was what I had latched on to
HPC won???t be a ???hot??? segment anymore because ???today???s leading-edge players ??? eBay, Google and Amazon – are not relying on commercial HPC at all???.
My response was that they weren’t in that market space in the first place, so to some significant degree this is a non sequitur. That is, we can say anything that the union of these three (or N if you prefer) for products/projects outside their scope is not hot, as they are not doing it, and be just as (in)correct. Specific example. Linux. “Linux is not hot as todays leading edge players ??? eBay, Google and Amazon – are not relying on commercial Linux at all???. This is false for several reasons. First, Linux is hot in the market, open source in general is. Second, Google appears to be using a non-commercial version. Ok, we can plug in things that they are not using, which are “hot”, and come up with other logical fallacies. I am not trying to set up a strawman argument. I am trying to show that I don’t agree with the reasoning. Just because these three, or your N favorite large companies are not using it, doesn’t indicate it as hot/not.
On the contrary, the larger companies often take longer periods of time to assess “risk” of newer technologies and methodologies. Sometimes to the point of paralysis. It is rare that you find a large nimble innovator. Very rare.
The flip side of the above, and sadly, the usual case is when the organization effectively makes a decision by failing to decide and act in a reasonable period of time. Leadership means sometimes taking calculated risks, and sometimes in those calculated risks, you have to have elements of the unknown. It is impossible to clarify all risks ahead of an action, and to delay action until all risks are quantified often means you accept that you are paying the opportunity cost of alternative decisions.
This is where disruptive innovation comes in. These higher risks have potential to drastically reduce the costs of alternatives. Larger organizations seldom will embrace these risks and simply accept them. Smaller startups will both produce and consume these. At some point the larger risk averse consumer and vendor will start to notice the upstart. That’s pretty close to the point where it is too late for the vendor. Their hand has been forced at that point.
Disruptive innovation abounds, in HPC, less so in the data center space. My own take on what is perceived to the disruptive in the data center space include blades and virtualization. SaaS (see the sidebar on the orbit change blog, which is where I got the impression that Sharad’s focus was on SaaS and related) is being touted as a disruptive technology, though it has not, as of yet, really demonstrated this disruptive nature. I tend to think of disruptive nature as an 80-20 type rule. You can do 80% of the functions for 20% of the cost. Or put another way, a discontinuity in the first derivative of the utility versus cost curve. You provide more power for less cost, in such a way that it is very hard for your competitors to follow.
This is part of how we have positioned JackRabbit, but that is a talk for another time.
The point being is that disruption favors those who are less inclined to be conservative, and those willing to take a risk. Sharad also indicates that this is a point that he was making
Essentially what I am saying is that a leading edge of any innovation wave is tilted in favor of startups.
The sense I have is that Sharad and I agree probably on far more than this, though what we had might be termed “a failure to communicate.”
Curiously, HPC has had several disruptive technologies emerge. In the late 80s early 90s it was the rise of the so called supermicros, which could do 80% or more of the processing for 20% or less of the cost, the cost of cycles dropped precipitously. The HPC market went from $40-80M in size to about $1B during this time. In the late 90s and early 00’s we had the rise of the clusters. Some of us had predicted this. Again, the cost of cycles dropped precipitously. Again, this market went from $1B to about $10B today.
The question of hotness of a market shouldn’t matter. Its all about disruption. Adding lots more value at a constant cost, or reducing the cost of the value.
In 02 and beyond, some of us started talking about that next disruptive event. APUs. In 05 and 06 they started gathering a great deal of interest. Now in 07 they appear ready to disrupt.
I have a $3000 board attached to one of my $4500 machines which is 5-8x faster on some of the calculations I have run on it. Others, realizing the incredible possibilities here are pushing to run calculations on GPUs. There are other vendors, but largely, if your cost is above $10k, you have marginalized yourself.
There are communities forming around these technologies. There are ISVs who see competitive advantage in delivering their value upon them. The market is poised yet again, to grow larger as the cost of power drops precipitously, the ease of use of this power increases drastically, and you no longer need an expert to drive this stuff. You are not talking about installed base of thousands. You are talking millions.
Yeah, the market is not hot in the eyes of people with the money to accelerate this.
Just remember that Linux Networx, Appro and other “large” vendors weren’t really on the radar 5 years ago during the last disruptive change. This change also had little in the way of external capital driving it. It happened none-the-less, and the entrenched vendors were left scratching their heads for a while. Took some thought to regroup, and some of them have recovered, some are still fighting (losing battles with) what the market has become.
So in this sense I disagree with Sharad about reasonable neglect by VCs of “non-hot” markets. There is value, there is disruption, there is non-linear growth. They don’t want to follow fads (e.g. hot-markets). They want to identify disruption before it happens, enable it to happen, and collect afterword.
Unfortunately this is not how they operate. And it means that markets with real disruptive waves are being ignored. Such is life.