HPE: Getting the Boat Out of the Basement

Recently, HPE announced that it was committing to becoming a leader in intelligent edge computing and to substantiate its intentions it announced it was going to put $4b towards this initiative over the next four years. The details concerning how this money would be invested weren’t released, however, it did intimate that it’s “memory-driven computing”, aka “the Machine”, would be involved. This is slightly reminiscent of HPE’s previous commitment to cloud computing, Helion, which lasted nearly four years as well. Come late and leave early. Works for me.

You may recall that “the Machine”, HPE’s memristor based computer, was the subject of a post here back in 2016 (see HPE Builds a Boat in its Basement – December 2016) in which we “suggested” that HP withhold the introduction of the technology until it could be fitted for emerging edge applications. We also “suggested” that the challenge for HP would be to exploit application “sweet spots” that lend themselves to the advantages inherent in memristor technology.  More easily said than done.

Edge is often conflated with a lot of the noise surrounding IoT. But there is one defining distinction, intelligent edge devices are autonomous agents that act independently and IoT devices are not. Apart from capturing an endless panorama of sensory, visual and aural data of empty parking lots and idle washing machines most of the economics of IoT is predicated on removing latent inefficiencies that result from common, every day processes. It’s a bit like making the case that recovering sawdust from mills will eventually pay for itself by eliminating the need for lumber. And at some point in the history of our species we will be glad it did. But until then, my wager is that the power law for extracting value from the information produced by these little buggers is shaped like an “L” with the bottom part extending infinitely towards some distant planetary intercept. (see Innovation: Power Laws, the Adjacent Possible and the Accumulation of Continuity – March 2017)

Source: IOT Analtics

By 2020, with estimates of some 50 billion IoT devices deployed, contrary to existing economic forecasts, chances are the value of the power consumed to run these devices will far exceed the value that their data individually and collectively produce. So, the value extracted from autonomous “memory-driven, intelligent edge” devices will need to be predicated on the quality of their decisions, not the quantity of their data. And the quality of their decisions is often a factor of the liability they may or may not incur, a conversation that is stalking the board rooms of nearly every high tech company tinkering with self-driving vehicles. This one factor alone influences the prospect pool enormously.

But it can also serve as an inspiration.

For instance, actuarially speaking, a well-booked Boeing 777 is a $1.5b flying liability. US carriers pay on average $4.5m per deceased passenger and the 777 carries about 350 passengers. Lose one and it adds up. Just ask Southwest whose CFM56-7B engine lost a fan blade mid-flight. Anticipating fan blade loss is an enormous problem but it can be done employing an autonomous, on-board, analog sensing, ML capable, memristor like computing facility. In some cases, predictive analysis can get the plane on the ground before a catastrophic failure occurs. There are approximately 30,000 commercial jets in service so, affordably priced, this application would amount to a $10b potential opportunity. Give or take a couple billion depending upon execution.

Rotor Imbalance Impact

Source: “Dynamic Behavior of Aero-engine Rotor”

C.Wang, D. Zhang, Y. Ma, Z. Liang, J. Hong

So, back to the boat in the basement.

If HPE really wants to get it out it might consider the following:

Frist, buy Predix from GE. They don’t know what to do with it anyway. After turning it into a captive software division of other operating units (GE Digital) it’s going to be a wasting asset. And GE already has too many of those. Besides they’re right down the street from you guys so you can screw things up even faster.

Next, develop and prefect the fan blade loss application. I understand that Harris Corp has some intellectual property that might fit well with this. Lots of other interesting ideas are bouncing around academia. Find them. Acqui-hire everything you can lay your hands on.

Finally, exploit high-value adjacencies and partnerships. As Bloomberg recently pointed out, this fan blade issue has got the entire commercial jet industry wrapped around its own axel, so to speak.

Turbines, turbines, turbines.

Now get on this. Daylight’s wasting.

Graphic courtesy of the Viking Ship Museum, Oslo, Norway all other images, statistics, illustrations and citations, etc. derived and included under fair use/royalty free provisions.