Somewhere in America right now, a paving company is bidding work with a back office built around AI from day one. One person in the back office; millions in revenue.  Small. Fast-moving. You haven't heard of them. They aren't trying to disrupt anyone. In the words of the famous 1997 Apple advertising campaign: they “Think Different.”  

There will be more of them. Some are not even companies yet — they are an idea in the head of someone who has not filed an LLC. They will not ask for permission. They will simply show up at the letting with a lower number.

This essay is about why we think they will win, and what the rest of our industry — our industry, the one we love — should do about it.

I. The Pace Is Different This Time

"Every day I go to sleep, there's an AI breakthrough. I wake up, there's another."

That's Elon Musk. You can disagree with Elon on a hundred things. You shouldn't disagree with him on this one. The pace is real, and it is unlike anything we have lived through.

Most heavy civil contractors don't feel it yet. The plant runs. The crews show up. The bids go out the door. Monday looks like the Monday before it. We get it — we live in this industry too. From the inside, the world looks unchanged.

It isn't. And it isn't going to look unchanged for much longer.

Construction has heard this technology pitch before. ERP systems. BIM. Procore. Drones. Each one was supposed to be the inflection point. Each one, in the end, mostly organized information differently. None of them did the work.  In fact, we essentially work the same way we did fifty years ago.  

The industry’s numbers do not lie. Construction labor productivity in the United States has fallen by more than 30 percent since 1970, while productivity in the overall economy doubled and manufacturing productivity expanded ninefold. In 1930, the Empire State Building was constructed at 4.5 floors per week in 410 days.  What happened? This is not because the people in our industry got worse. It is because everything around them got more complex faster than the tools could keep up. How will we ever catch up?

AI is different from what came before. Not in the marketing sense — in the literal sense. AI doesn't just skim the spec. It reads the 300-page spec and flags every risk. It doesn't just store competitor bid history. It analyzes pricing across hundreds of historical jobs and tells you where your number should be. It doesn't just hold a draft submittal. It writes the submittal. It does this twenty-four hours a day, seven days a week, without overtime.

We can “hear” what you are thinking.  “That’s all in the back office.  What about the work?”  Robots are entering our sites and work zones at an exponential pace.  Self-driving is here and it is real.  Do we really think our paving crews will look the same five years from now?  Three?  In the 1920’s a 1000 acre farm required a work force of at least 50 men.  Today that number is one or two.  

And the curve is bending. The head of Claude Code at Anthropic — one of the three leading AI labs in the world — hasn't edited a single line of code by hand since November 2025. He writes the equivalent of a full engineer's weekly output before lunch - all with AI. The CEO of Anthropic predicts 50% of entry-level white-collar jobs will be eliminated in the next one to five years. This is a big deal. It is not five years away. It is here, now, and it is accelerating.

II. What AI-Native Looks Like

The typical heavy civil contractor is organized around a structure that was designed in the Eisenhower administration. Estimating. Project Management. Accounting. HR. Compliance. IT. Admin. Business Development. Each function staffed with people doing knowledge work — reviewing documents, entering data, writing reports, tracking schedules, managing paperwork.

The contractors built around that structure are excellent. They built America. They are also about to face competitors who never inherited that structure in the first place — and that is what changes the math.

Run the numbers. Back-office overhead at the leanest public heavy civil contractors runs about 8% of revenue. On a $500 million operation, that is roughly $40 million a year. That is not concrete. That is not asphalt. That is not steel. That is paper, process, and people managing information.

What happens when a competitor shows up running half that overhead with AI? That is a structural cost advantage that flows directly into more competitive bids, higher margins, or both. Every year. Compounding.

The shift outside the office walls is even more dramatic. A paving crew today runs about twelve people. With autonomous compaction, semi-autonomous paving, and AI-coordinated logistics, that crew could be three. Your best operators stop running a single machine and start running a fleet — managing autonomous equipment that doesn't get tired, doesn't take breaks, and doesn't make fatigue errors at hour thirteen of a summer shift. Inside the office, your best estimator/project manager stops grinding through jobs alone and starts directing twenty AI agents — pulling data, running takeoffs, flagging risks, analyzing competitor pricing, comparing specs, and updating crew schedules in real time. The estimator/project manager becomes a coach, not a clerk.

And it is not just software. In January 2026,  Musk said what most people in our industry have not yet heard: "My prediction is that there will be more robots than people." Tesla plans to deliver billions of humanoid Optimus units in the coming decades. You don't have to believe his timeline to take the trajectory seriously. Bluelight Machines has autonomous rollers that have already compacted over one billion square feet of soil — injury free — scanning their surroundings five times per second. Autonomous excavators, dozers, and general-purpose construction robots have collectively raised more than a billion dollars in the last eighteen months — from investors including Bezos, NVIDIA, Temasek, and the U.S. Air Force. Smart money does not bet on gradual improvement. It bets on disruption. And it is pouring into our industry.

This is not a story about wishing away the men and women who lay asphalt, pour concrete, run equipment, and build roads. Construction is short 350,000 workers today and will need 500,000 more by 2027, and that gap is real. But it is also true that, over time, AI and automation will change a lot of jobs in this industry — including jobs in the back office and jobs on the crew. Pretending otherwise is not respect. It is avoidance. The honest framing is this: every contractor in America now faces a choice between partial displacement with a future and full displacement with no future. The companies that adopt early keep most of their people, redeploy the rest into higher-leverage roles, and stay in business. The companies that wait lose the bid, lose the backlog, and eventually lose every job in the company. The next generation will not work for a dying business. They will work for the one that is building the future of how this industry operates.  This is about a tectonic shift in productivity.  It is about our industry finally realizing the exponential productivity gains that have eluded us since the Empire State Building went up.  

III. The Trapdoor

Almost all government highway work in America goes to the lowest bidder. That is not a bug. That is the best feature of the entire system. You don't need permission to disrupt construction. You don't need to convince a procurement committee. You need a number.

If your number is lower because AI made your operation leaner, faster, and smarter — you win. Every letting. Every time. Low bid is the trapdoor.

And on the work that doesn't go strictly to the lowest number — design-build, alternative delivery, anywhere the owner picks on best value — the same logic flips in your favor. If you have a faster way to design, a better way to manage risk, a smarter way to sequence the work, you win there too. AI-native contractors win the low-bid lettings on price and the best-value awards on quality. The trapdoor opens on both.

The last time a technology-first competitor went up against a legacy institution, it didn't end well for the incumbent. NASA spent $24 billion developing the Space Launch System. It costs over $2 billion per launch. The rocket crashes into the ocean after a single use. SpaceX developed the Falcon 9 for roughly $390 million — and it costs $67 million to launch. The rocket lands itself and flies again.

Same workforce pool. Same laws of physics. Different model. Different result.  Better result.  Better for every stakeholder. 

SpaceX did not win because NASA's engineers were bad. SpaceX won because the structure around them had calcified around an older way of doing things. The AI-native contractor will win for the same reason. And before anyone in our industry tells themselves this kind of disruption couldn't happen here, it already has. Clayton Christensen built his theory of disruption around a case our industry knows well: the cable shovel versus the hydraulic excavator. The cable-shovel manufacturers were excellent. They listened carefully to their biggest customers, who wanted bigger buckets and longer reach. Hydraulics couldn't match them on those specs at first, so the incumbents passed. Hydraulics improved anyway — starting in markets the incumbents didn't care about, small contractors digging residential trenches and landscaping work — until one day they could do everything cable could, only better. Today nearly every excavator on every job site in America is hydraulic, and almost none of the original cable-shovel manufacturers still exist. The companies that lost weren't run by bad people. They were run by good people whose org charts, customer relationships, and capital structures were calibrated to a way of working that was about to end. A handful of incumbents in our industry will see this clearly enough, and move fast enough, to lead the next chapter. Many will not. The honest question every leader needs to ask is which group they intend to be in — and whether the answer is supported by what they actually do on Monday.

Let us be clear — we are not cheerleaders for disruption. The men and women we have worked alongside and genuinely care about are the backbone of large civil contractors today. We are deeply committed to the success and longevity of the companies that built America. That said, we do see a significant inflection point approaching, and we passionately want to see these contractors navigate it successfully. We believe the incumbents not only can make this shift — for now, they hold the advantage to do so. 

IV. What To Do Monday Morning

We will end with what we hope you do, not what we hope you feel.

If you are reading this article you are most likely our colleagues, friends, family, or work in the construction industry.  Start to teach yourself how to use these tools.  You don’t need a class, just ask the AI to teach you.  Spend 30 minutes per day challenging yourself to learn a new skill and open your mind.

If you lead a heavy civil contractor today: pick one part of your back office — estimating, takeoff, submittals, AP — and put an AI tool next to your team for ninety days. Measure. The companies that move first will compound. The companies that wait will be racing to the trapdoor.

If you are a founder launching a firm from scratch: build around AI from day one. Don't hire the org chart you inherited. Hire one excellent person per function and put agents behind each of them. Be honest with yourself about the hard part. The bottleneck is not the agents. The bottleneck is finding operator-managers who can actually orchestrate them — people who understand the work well enough to direct AI, judge its output, and intervene when it's wrong. Nobody has trained anyone to do this job. Whoever figures out how to identify, hire, and develop those people first wins the next decade. Then show up at the letting with the number that has to win. Run that play until you have enough work to matter. Then keep running it.

If you are the next generation looking at this industry from the outside, wondering whether road-building is where you'd want to spend your career — the answer is yes. The next fifty years of American building can be the most ambitious in our history. More highways, more bridges, more ports, more energy, more water, more data centers, more of everything this country needs to lead. The pie gets bigger. America gets built faster. There is a seat at the table for you, and the work has never mattered more.

One last thing. AI disrupts American civil construction. The only question is who runs the firms that win. Infrastructure is how nations compete. The cable-shovel guys had a fifty-year head start. It didn't save them. The trapdoor is open. Let’s walk through it. 

Pat Nelson is Vice President at Granite Construction, former President of Lehman-Roberts Company, and the 2025 Chairman of the National Asphalt Pavement Association. 

Tristan Wilson is the CEO and Founder of Edgevanta and a fourth-generation contractor. His family built and led Barriere Construction.

Both grew up in the road-building business and share the conviction that AI will transform how America builds its infrastructure.

Enjoyed this newsletter? Forward it to a friend and have them sign up here!