Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

Strategic Investing Trends: Autonomous Computing

|Includes: Caterpillar Inc. (CAT), KMTUF

Autonomous computing is a long term investing trend which will fundamentally change almost everything, nonetheless the question remains: Is there profit in this trend for the home investor? After all, flying machines also changed everything, and as Warren Buffett likes to joke investors would have been money ahead if someone had shot down the Wright Brother's first flight.

First, a quick definition: autonomous computing encompasses a continuum of programmable tools which mimic and enhance our ability to solve complex problems. Think of a spreadsheet that solves the problem for you, after setting up the problem for you, after identifying the problem for you. Think of computers killing at Jeopardy, assisting in medical diagnoses, and targeting ever more annoying ads at already annoyed online shoppers. The autonomous computing trend will evolve over the next ten+ years until it becomes pervasive in everything from surgical instruments to automobiles to excavators. Today's simple industrial robots will mature, drop their spot welders, and grab a seat in front of every type of complex problem solving and pattern recognition task.

Some examples of autonomous computing have been under development for over 10 years and are already being deployed. Audi was granted a license to operate autonomous cars in Nevada in January of 2013, Google received a similar license in May of 2012 and Continental, an automobile supplier, in December, 2012. Trivial, though useful, current examples of autonomous cars include self-parking cars for nervous urban drivers. More profound autonomous vehicle technology may largely eliminate the number of "texting" drivers who end up in the wrong lane or upside down in a ditch. What parent wouldn't want to buy an autonomous vehicle as their teenager's first car with the accelerator set to legal speed limit?

Caterpillar's MineStar™ system is developing an integrated set of mining tools which includes autonomous mine vehicle tracking. Cat is deploying this technology with Fortescue and began experimenting with autonomous vehicles in the mid-1990's. Rio Tinto is piloting Komatsu's FrontRunner® technology for semi-autonomous explosives loading at their "Mine of the Future" test site. So eventually we'll have autonomous mining systems barking directions at autonomous excavators fighting with an intelligent drilling machine over who gets the parking spot closest to the office. All complex, dangerous, and lucrative physical processes will be automated along these lines. Do we really need people to fly cargo planes, drill for oil, process sewage, mine diamonds, route electrical power or match ask and bid prices?

It becomes pretty obvious that control centers, probably thousands of miles away, will soon monitor/control (or already do control) physical systems of highly automated, autonomous machines carrying oil around the horn, producing earnings reports, or replacing kidneys in Shanghai.

Autonomous computing is unlikely to replace human intelligence in the near term, but these tools will greatly extend human capabilities while also pushing technological capabilities forward. The limitation at this point is that all these autonomous systems are using primarily conventional technology (with some specialized chips) with highly unconventional (and grossly expensive) algorithms. Our brains, as slow as they are, are still more advanced general purpose pattern recognition engines than our current chip technology can duplicate cost-effectively. Still, it is only a matter of time.

The practical deployment of autonomous computing from the R&D labs into production environments over the next ten years will produce winners and losers, just as the deployment of Arpanet and its evolution into the Internet produced winners and losers. Big winners (think Google, Facebook, Twitter, maybe Amazon) and big losers (every company I invested in from 1997 to 2000).

For autonomous computing to be effective the capabilities of current pattern recognition technologies need to break through current performance barriers. Pattern recognition systems will evolve rapidly in response to the demands of autonomous computing and extend the capabilities of current vision systems, natural language recognition engines, unstructured data analyzers and contextual parsers.

The technologies used to develop these pattern recognition breakthroughs will enable correlative breakthroughs in expert systems, artificial intelligence systems, neural net training systems and similar tools required for autonomous systems to react to and learn from unexpected changes in their physical or quantitative environment.

Cloud computing will be pushed, during the near term, because these applications will be incredibly compelling, but sophisticated (read expensive) programmers will be required to create and maintain the early versions of these tools. Eventually most of the programming will be largely initiated and extended by the tools themselves.

There are a few industries where it seems obvious autonomous machines will introduce disruptive changes.

Medicine: automated diagnostics, automated surgery, drug interactions, knowledge management, drug development, prototyping and evaluation, remote collaboration.

Architecture: automated design, automated construction, automated building management.

Finance/Accounting: purchasing, invoice processing, pricing, reporting, variance analysis, risk management.

Marketing/Advertising: ever more obnoxious and intrusive individual and life cycle targeting.

Energy: grid management, monitoring, exploration, drilling, refining, pricing, hedging, arbitrage.

Machinery (Autos, Ships, Trains, Foundries, Factories, etc): Inventory management, order processing, machine tool management, automated design, global resource routing.

Brokerage and Capital Management: Investing could become more about identifying strategic trends and less about data analysis. Most of the data analysis or underlying regression algorithms will be massaged by these autonomous computing engines and the human analyst will be left to identify broad themes which are populated with appropriate stocks by the non-human research engine. The quants would be recognized as a rather primitive attempt to apply linear algorithms to an environment of fundamentally hyper-causal linkages. As central banks introduced hyper-causal effects on world economies in the past few years the quants have suffered predictably. The good news is that the stock market, once autonomous market-makers are involved, could decide to return to rational analysis and longer term investing as short term price movements are seen to be largely random-unless inside information is obtained-and the underlying capabilities and decisions within companies could actually drive long term performance. The markets themselves will become entirely emptied of people shepherding day-to-day trades and the current cumbersome system of bond trading will be pulled into the electronic trading system currently used for equities.

Given that autonomous computing will drive widespread and comprehensive change across many sectors we should begin the process of identifying winners and losers, placing our bets and assessing our investments against this strategic backdrop. Future articles will explore individual programs by specific companies in an effort to find suitable investments.

Disclosure: I am long CAT.