When living and operating in a market largely dominated by a vendor that isn’t you, the strategy you must deploy is one of focus. In the early days of Power, IBM tried to take on Intel head to head and that just wasn’t working. You can understand why IBM thought it could do this; it was once the most powerful company in the world. But, like Microsoft, Intel’s strength largely came from providing technology to firms like IBM, and IBM’s decline in the late 1980s and early 1990s not only weakened it substantially, it collectively strengthened other firms. Much like AMD, which has always been weaker than Intel, IBM needed to pick its battles, and given that the company still pretty much owns the market for enterprise-class AI with Watson, and that this segment is slated to become the most lucrative in the industry for servers over the next decade, it chose wisely to make this one of its critical areas of focus.
This week, IBM announced a major move in this direction with its Power System Servers based on Power9 processors. Both Google and the U.S. Department of Energy saw the advantage of this move and backed it immediately.
Let’s chat a bit about what this means.
Dedicated AI Servers
Almost every other company in this space is attempting to adapt some existing technology to address the AI opportunity. IBM, one of the few companies that retains a well-funded research arm in IBM Labs, realized early on that this probably wouldn’t work. Much of the intelligence research focuses on things like neural networks and trying to emulate neurons to build systems that better match what we currently define as intelligence. Yes, there are risks that our definition of intelligence may be flawed, but generally the industry’s success in any area is tied to tying to build against a known model, in this case how we think. This is because, otherwise, you lack either a frame of reference or measurements, so you cannot determine if you’ve really made progress.
IBM has rigor that it has developed over decades that forces it down a path that has a high probability of not only initial success but being able to measure and demonstrate that success. This path created Watson, which has been a pretty impressive, albeit expensive solution (to buy and train). Now the goal shifts to getting the cost out of the solution while improving performance and that path is evident in this latest Power9 Power System Server announcement.
But you don’t want to blend these efforts with normal operations because the unique tuning these servers get optimized them for this workload and should not only allow them to better perform AI duties, but they should be able to do the related tasks more cheaply as well.
Google and Department of Energy
IBM has almost always been very good at selecting reference accounts that showcase the benefits of a new platform, service or initiative. In this case, Google and the U.S. Department of Energy are to point. Google knows that to hold its near monopoly on search, it needs to continue to provide a better service at massive scale and, as it rolls out AI-enabled products like the Home Pod against Amazon, this will also help assure those offerings are successful. So, to Google, AI is massively strategic, and it has one of the most well-funded internal AI efforts in the world. Having the firm bless this effort does an impressive job of validating the value of it.
The U.S. Department of Energy has massive issues managing a diverse grid of aging power-generating and delivery systems that are under attack not only by the elements but by increasingly hostile state-level attackers. In addition, these power grids are increasingly being augmented by new green energy generating technologies like wind and solar power, which can literally change their generating capability with the weather, and that is on top of loading by customers that, as electric cars come into the market, varies far more than ever before. Being able to respond near the speed of light could make the difference between whether people have power or are living in the dark, and people just aren’t fast enough making AI critical to assuring the nation doesn’t go dark. The U.S. Department of Energy validating this move is also incredibly important.
IBM has wisely decided to focus on areas where its server efforts could stand out as superior. One of the most promising server growth areas for the next few decades is AI. IBM currently leads in enterprise AI, so is creating focused servers that provide this AI capability in an ever more cost-efficient fashion. But validation is incredibly important, particularly when you are on the cutting edge of a new market like IBM is, and having Google and the U.S. Department of Energy backing this move makes it more real, even though IBM’s historic strength in this area already gives it a higher level of validity than most.
AI is coming, and it likely will be one of the biggest differentiators between firms that are well managed and provide responsive services at scale over those that don’t. A focused server line targeting this opportunity may be slightly ahead of the curve, but often winners are defined by how well they start. IBM’s starting position remains well in advance of its peers and it appears to be investing to make sure that doesn’t change as enterprise-class AI moves into the mainstream.
Rob Enderle is President and Principal Analyst of the Enderle Group, a forward-looking emerging technology advisory firm. With over 30 years’ experience in emerging technologies, he has provided regional and global companies with guidance in how to better target customer needs; create new business opportunities; anticipate technology changes; select vendors and products; and present their products in the best possible light. Rob covers the technology industry broadly. Before founding the Enderle Group, Rob was the Senior Research Fellow for Forrester Research and the Giga Information Group, and held senior positions at IBM and ROLM. Follow Rob on Twitter @enderle, on Facebook and on Google+