For half a century, the smart money bet on finance over building. That’s reversing. Capital is pouring into five kinds of construction demand at once, and the thing standing between that money and what it wants built isn’t more money. It’s whether the industry can supply the labor, materials, and machines fast enough. On the latest Bricks & Bytes, Patric Hellermann lays out why supply, not capital, is now the real constraint, and what that means if you build for a living.
Everyone’s got a take on how AI and software will remake construction. Better workflows, smarter scheduling, slicker risk tools. Some of it helps. But pull up an actual project’s profit and loss and software barely registers. The story that matters is happening one level down, in the atoms.
Right now capital is moving into construction harder than it has in a generation. Reshoring factories, data centers, energy, civil infrastructure, defense. Five distinct demand drivers, hundreds of billions of dollars, and demand that’s unusually clear about what it needs.
The catch is that money isn’t the hard part anymore. Supply is. The industry’s ability to actually keep up with all that demand is the bottleneck, and that quietly changes where the smart bets sit.
When every customer wants roughly the same thing at roughly the same scale, you can finally build real solutions against it. Patric calls that standardized demand. The winners will be whoever moves cost, speed, or quality at scale, and touches one of the five drivers pulling capital in.
The constraint was never money
Why capital is chasing atoms, not pixels
For about 50 years, the smart money said the future was financial. Pixels over atoms, derivatives over factories, trading over building. If you wanted to get rich, you went into finance, not concrete. That bet is finally turning. For the first time in half a century, capital is flowing back toward the people who actually make physical things.
And it isn’t spread thin. It’s concentrated in five demand drivers: re-industrialization and reshoring, data centers, energy, civil infrastructure, and defense. Populations are consolidating into fewer, bigger cities, which means water systems, sewage, roads, and power grids all have to scale at once, with climate resilience layered on top. Europe is rebuilding its defense base. Data centers are going up everywhere. Each of these needs labor, materials, and machines moved at massive, standardized scale.
That rotation from bits to atoms is the same shift a16z and others have been circling, and construction sits right in the middle of it. If you want the long-run data behind why the industry is such fertile ground for this, our breakdown of construction’s 75-year productivity problem is the backdrop to this whole thesis.
Where the money actually goes on a project
Labor, materials, machines, and software’s thin slice
- Labor. Around 35 to 40% of total construction cost.
- Materials. Another 35 to 40%.
- Machines and equipment. Roughly 10%.
- Financing and working capital. Spread across all of it.
- Software. About 2% of the total.
You hear it constantly. AI, automation, design software, risk platforms, all set to remake how we build. Useful tools, some of them. But when you look at the actual profit and loss of a project, software lands at around 2%. The costs that move the needle are labor, materials, and machines. Atoms.
That’s why the productivity gap doesn’t close with better software alone. To change what a project costs, or how fast it gets built, you have to touch the 90-plus percent that software doesn’t. That’s the uncomfortable part for a lot of tech pitches, and it’s the opening for anyone willing to work on the physical layer.
Get the construction briefing execs actually read
News, deals, and deep takes on the forces reshaping construction – from 300+ issues and counting.
Join 3000+ ReadersStandardized demand is the real tell
Why five drivers change the math for builders and founders
The reason capital can move this aggressively is that the demand is unusually uniform. When every customer wants roughly the same thing at roughly the same scale, you can build a solution against it and sell it more than once. Patric calls this standardized demand, and it’s exactly what investors are hunting for.
So the filter is simple. A solution has to address a real supply constraint, it has to touch one of the five drivers, and it has to move cost, speed, or quality at scale. That might be site robotics, prefabrication, logistics, materials innovation, or workforce solutions. SoftBank’s $100 billion bet on robots that build data centers is the loudest version of this exact logic, and you can read our take on why Roze AI treats data center construction as the bottleneck to break.
For 50 years, the people moving atoms were an afterthought. Now they’re the constraint. And the constraint is exactly where the money goes.Patric Hellermann, Bricks & Bytes
What this is not
A tailwind, not a cure-all
Worth being clear-eyed. This isn’t a boom that fixes everything. It won’t solve unemployment or repair the wider economy. Data centers and reshoring create jobs, sure, but they’re aimed at five specific demand drivers, not the world’s biggest problems.
For anyone who builds, though, it’s about the best tailwind you’ll get in a career. The money is here, the demand is clear, and everyone knows it, which is precisely why the window is narrow. The ones who win will be the ones who move while it’s open, and who address the fact that we can’t keep up with demand in a way that actually moves cost, speed, or quality.
| Demand driver | What’s pulling capital in | Supply constraint it hits | Where builders and founders can move the needle |
|---|---|---|---|
| Re-industrialization / reshoring | Manufacturing moving back onshore | Skilled trades, buildout speed | Prefab, modular factories, workforce solutions |
| Data centers | AI compute buildout | Electricians, crews, schedule | Site robotics, off-site assembly, logistics |
| Energy | Grid and generation demand | Specialist labor, materials | Materials innovation, standardized components |
| Civil infrastructure | Urban consolidation, climate resilience | Labor at scale, long timelines | Industrialized delivery, logistics |
| Defense | Europe rebuilding its defense base | Secure capacity, materials | Rapid buildout, hardened supply chains |
That money is no longer the hard part in construction. Capital is flooding toward five big demand drivers, and the real bottleneck is supply, the industry’s ability to move labor, materials, and machines fast enough to keep up. (Source)
Re-industrialization and reshoring, data centers, energy, civil infrastructure, and defense. Together they represent hundreds of billions in capital and demand that’s clear and standardized about what it needs built. (Source)
It matters, but the point is that software alone won’t close the productivity gap, because it’s a tiny slice of the P&L. Labor, materials, and machines are where cost actually lives, so tools that reach those atoms are the ones that move the numbers. Our look at construction’s productivity problem covers the data behind that gap. (Source)
It’s Patric’s term for the moment when every customer wants roughly the same thing at roughly the same scale. When demand looks like that, you can build a repeatable solution against it and sell it many times, which is what makes these five drivers so investable. (Source)
Anything that addresses a real supply constraint, touches one of the five drivers, and moves cost, speed, or quality at scale. Think site robotics, prefabrication, logistics, materials innovation, or workforce solutions. SoftBank’s data center robotics play is one high-profile version of that bet. (Source)
No, and the episode is careful about that. Data centers and reshoring create jobs, but they’re pointed at five specific demand drivers, not the economy’s biggest problems. It’s a powerful tailwind for people who build, not a cure for everything. (Source)
Because everyone can see it happening. The capital is moving, the demand is clear, and the need is understood, so the advantage goes to whoever acts while the opening is still there rather than waiting for certainty. (Source)
Related Articles
Construction’s Productivity Problem: 75 Years of Data That Should Bother You
SoftBank’s $100B Robotics Bet: Why Roze AI Wants to Build the Data Centers AI Runs On
