BNSF had a puff piece in Progressive Railroading recently about their new technology division that I couldn’t help but comment on. Sorry BNSF friends, but BNSF tech is a mistake. From the article,
“For a long time, BNSF Railway has considered itself “a company of builders.” The Class I’s leaders use that characterization because throughout the company’s 176-year history, lots of infrastructure has been constructed, much rolling stock has been retooled and myriad new ways to perform tasks have been created…
“But in terms of technological advances, the railroad has not been a company of builders in recent years. Over the past decade, purchasing commercial off-the-shelf products has been the norm instead of developing technologies in house. And dozens of integration projects have incorporated purchased software.
“Created in March as a new company division, bnsf | tech is charged with developing technology solutions to address operational challenges, including various technologies designed to improve safety, on-time performance, terminal functions and the customer experience. The division emphasizes a customer-focused approach and aims to leverage internal expertise and collaboration.”
The article goes on to say BNSF tech (or bnsf | tech - lowercase with that one vertical line thing so you know it’s legit) will develop their own technology in areas such as operations management tools that have been around for decades already, GPS in terminals (? - also been around for decades), AI (the kind that hasn’t been around for decades, of course), and creating prescriptive actions from weather forecasting (i.e. the Weather Channel?).
Although it’s mostly drenched in corporate speak, there are some important details in the article:
“Over the next year, the Class I plans to hire 1,000 employees. The railroad is actively recruiting computer science, data science and software engineers, particularly from colleges in various states, amidst a soft tech recession, says Govind.
“‘We just hired 100 engineers from state colleges in Wisconsin, Illinois, Michigan, Texas, Oklahoma and Florida,’ he says.”
Assuming a conservative average total comp cost of $175k/engineer, that’s $175m/yr BNSF is investing to support this new division. I spoke with someone from BNSF back in the Spring after this was announced and they said they were also going to let this division wear jeans and mostly work from home. Got to get that tech culture right!
While trying to create a romantic connection to a former time as a “company of builders” is heart warming, this article promoting BNSF tech is another one to add to the Class 1’s highlight reel of what not to do. I will first point out why this is a terrible idea and follow it up with what a more effective, pragmatic, but far less sexy approach would be for BNSF and any other Class 1 who still reads this newsletter.
Why is this dumb? Isn’t investing in tech what any company should be doing?
Technology has always been a distraction for management teams since the dot com bubble. Great margins, higher valuations by investors, but almost never done right by any non-tech company for a number of reasons, especially those who don’t already have a bunch of PhD’s churning out patents every year.
To start, there’s a big difference between investing in technology to improve your business and creating technology to improve your business. Unless you’re a technology company, I don’t think any company should do the latter for two reasons:
Comparative advantage - Stick with what you’re good at. There are good reasons why the Class 1 railroads no longer own Madison Square Garden or oil refineries. Those businesses require completely different managerial skill sets and capital investment plans than a railroad. Conglomerates are no longer in vogue because markets have figured out that focusing on one business in one industry can produce greater shareholder returns while other options are available for diversifying portfolio risk.
In one of my earliest articles, I shared examples of how ADM has consistently failed at non-core businesses and why UP’s Cold Connect service failed (Spoiler: you can’t run a business with truck-like margins as a railroad).
Building technology like software, hardware, AI, etc. are not core competencies of a Class 1 railroad. Sorry. You’re special in a different way, railroads. Do railroads need to build LLMs, no. Let the companies that are good at doing that go invest in those products, then build on top of them.
Do you know who is good at creating technology? Technology companies. They have the processes, people, structure, skill sets, correct capital investment strategy that all supports successful technology development. They know the ins and outs of certifications and regulations required to put a product out in the wild. They have relationships with the appropriate regulatory bodies and suppliers. They know how much time and effort nuanced development takes. And they have the scars and stubbed toes to prove it because they had to learn all of it the hard way.
BNSF may be a private company with more “patient capital” than a public one, but that alone doesn’t mean that’s the only constraint preventing a big company from doing technology successfully.
Reinventing the wheel, literally. By closing the shutters and becoming more insular (can railroads really become more insular at this point?), with regards to technology development, BNSF risks wasting capital on reinventing the wheel (i.e. wasting time developing technology that’s already commercially available).
Recently a Class 1 railroad touted it’s new whiz-bang drone launch system that it had developed. The message to investors was, look at all this cool innovation we can do! Except, that’s the exact capability that’s already been on the market for years.
As another example, NS built a car inspection portal. CSX, CN, and CPKC have them too, except CSX and the Canadians just went out and bought one in the market. Why? Sure, NS cut out the manufacturer’s margin (which probably isn’t great to begin with when you’re selling so much hardware and moving so much data), but what a waste. They’ve also closed the door on upgrades that CSX, CN, and CPKC can receive because the manufacturer’s technology is seeing so many more cars. (Sidenote: Class 1’s aren’t even using all the data and AI capabilities these provide. More on that below).
Fixing the Real Problem
The oft-mentioned Matt Rose quote that Tony Hatch likes to use is informative, “Silicon Valley isn’t going to come and save us.” This myopic view of the world leads to the conclusion that railroads have to build their own tech to advance. Matt, and all the other Class 1’s, make the mistake of discounting the ability for the market to build technology for the railroads by not understanding exactly why Silicon Valley isn’t beating a path down to Fort Worth.
The core issue is that railroads reject any new innovation like a virus. The host rejects the implant like a foreign body. As a technology company CEO, let me state the obvious: We will build products for people who pay us! Novel idea.
Take, for example, what we’ve already encountered at Hum as we’ve tried to introduce the very technology that could have prevented the East Palestine derailment. One Class 1 approached us with the idea of offering shippers and car owners rate discounts if they equipped their railcars with our bearing and wheel condition monitoring safety technology. The railroad offers the discount, the shipper equips it and shares the data, the railroad can then perform targeted, condition-based maintenance to not only prevent derailments, but also to avoid bad ordering cars or stopping trains due to faulty wayside sensors. Perfect win-win for everyone involved.
After 9 months of playing matchmaker with the Class 1 and various shippers on this concept, educating all parties, and helping to work through any questions or issues, we finally get everyone on the Class 1 commercial side to sign off on it. All systems go, or so we thought. We then reach out to the operations side of the house to begin talking about the nuts and bolts of how to integrate this new data stream into their operation so that it could actually be used. Maybe this was our mistake as an engineering company that obsesses over the details. The response: “We already pay for safety systems, we don’t need to pay for any more safety systems.” This is why the railroads do not innovate.
Rather than spend hundreds of millions on reinventing the wheel, BNSF (and every Class 1) should instead try to fix its corporate and operating culture that is stifling innovation today. Three ideas:
Perform post-mortems on past technology implementations and what worked/didn’t work (no organization learns from past mistakes). Then share with the rest of the organization. Educate senior management. Identify the core issues that killed prior projects and the key ingredients that made others work.
Get smart and jump in AFTER the development has happened - Have a small team of 2-4 people with a 2000 sq ft office in Silicon Valley, Austin, New York, Chicago, or any other tech hub to just go out and have coffee with founders and engineers at technology companies. You will be shocked at just how much you can learn by asking questions in this manner. (This, cold calls, and reading endless research papers on railcar engineering were about all I did the first year I started Hum). And this is super cheap. You should be able to spend $300-400k/yr or less on this.
You can (and probably should) even take this further by buying or investing in Series A+ startups to take technology fresh off the line and begin having the tech company fit it to your needs. Some of the Class 1’s are kind of involved in Plug and Play ventures which does something similar to this, but that’s not what I’m talking about. That’s lazy. Don’t do lazy. Do it right.
Tie executive and senior management compensation to technology implementation objectives. This may be the most important one, especially with respect to AI. Innovation groups, like safety groups, are borderline useless if they aren’t given direct P&L responsibility. Companies like to pay lip service to them, throw a few mil there way, and call it good. This is why companies are horrible about things like innovation. Without tying it to management compensation, absolutely nothing will happen. Especially the COO and his/her lieutenants for a railroad.
Credit to BNSF for trying to implement more technology. This is good. Maybe they’ll get at least part of the way there in the end, but they’re going to waste a lot of money doing it. If a Class 1 truly wanted to become an innovative powerhouse, it needs to put down the shiny object and do the hard, boring work of creating the right organization that can execute on innovation.