I have been thinking a lot about transportation recently. This is not really my field, but I’ve been travelling a lot, so it’s on my mind. I’ll spare you a rant about air travel in the US, what really has me intrigued is the self-driving car.

When I first heard about Google’s self-driving car, I was definitely a skeptic. It struck me as just one more science project from a company with too many distractions.

This all changed over a conversation I had with Rob Coneybeer at the Venture Beat Mobile Summit a couple years ago. He convinced me that the autonomous car could be the next major software platform. Also, with my new job I have been driving to work for the first time in my career.

There is some very compelling math behind this idea. So far, most of the analysis of self-driving cars has focused on the cost of upgrading a vehicle to drive itself. The Journal did an article last year where they calculated that the additional sensors and computers needed for such a vehicle cost about $15,000 to $20,000 per vehicle today. But those figures are for prototypes, and much of the cost is the sensors. When these become mass-produced, the self-driving components will probably end up adding about $5,000 to the cost of a car. Not a big number, but it enough so that this is not an obvious upgrade for everyone.

However, I think there has been a lot less understanding of the benefits. The Journal seems to really like this idea, and has written a lot about driverless cars. They recently did a piece focusing on the safety benefits, which are meaningful. But I think there is a lot more to the idea. So I hit the census bureau for some data on US driving patterns.

The average American drives 24 minutes to work every day, or 48 minutes per day. That works out to 206 hours per year per worker. The average American salary [PDF] is $50,054 per year, which equates to roughly $25 per hour. With 119 million Americans driving themselves to work every day, all of this works out to $616 billion per year in time spent commuting to work, or $5,162 per worker.

By this math, an extra $5,000 per car to upgrade to self-driving frees up enough time for the upgrades to pay for itself in one year.

These are big numbers, the US economy spends more than half a trillion dollars in commuting time every year. This is a huge figure, and a huge opportunity for the company that can winnow that down.

Obviously this is a complex ecosystem. There will be many companies vying to control this market. However, they key piece of the puzzle is the software that goes into making the car autonomous. And so far, the company that seems to be miles ahead of everyone else is Google. As with so much of computing, the value rests in the software, and for us to move to driverless cars, it is reasonable to think of the algorithms that drive that can capture much of the upside of the ecosystem. To put it into a simple framework, those algorithms will form the basis of the car’s operating system. And the owner of the OS will drive many of the other decisions that go into building such a vehicle.

By that math, Google’s investment looks very smart. This also explains the reports this week from former Journal reporter Jessica Lessin that Google is looking to build its own cars, circumventing the auto OEMs. This is very similar to the smartphone market, where Google has sought to build its own hardware to push innovation for its Android OS. Google is doing something similar in the data center, building its own switches and servers. The auto OEMs probably recognize that Google can drastically alter the economics of the industry and are reluctant to let Google to get too far ahead, and so in typical Google fashion, they are forging their own path.

So the next time you are stuck on the 101 going 10 miles an hour, think about all the better ways you could be spending your time, multiply that by everyone and you get a sense of this opportunity.