Liability, not Legacy: why a16z is wrong about construction

Last week, Andreessen Horowitz (a16z) released a widely circulated piece titled “Every Building You’ve Ever Been In Was Designed By Software Built in 1997.” On the surface, it’s a compelling narrative: the $13 trillion construction industry is a sleeping giant, shackled by archaic software, waiting for the “AI-native” knight in shining armor to save it.

But for those of us in the BIM trenches, the article feels less like an industry analysis and more like a high-gloss brochure for their own portfolio. While we should welcome the influx of capital into AEC, it’s time to call out the “Silicon Valley Reality” for what it is: and highlight the incredible, unique way our industry actually innovates.

The “Thought Leadership” as a pitch deck

Let’s be honest: a16z isn’t writing for architects or general contractors. They are writing for their Limited Partners and prospective founders. By framing the industry as “stuck in 1997,” they create a vacuum that only their portfolio companies (like Snapshot or others in the AI-design space) can fill.

The “Rip and Replace” narrative is a classic SaaS playbook. If you can convince the market that the incumbent is a dinosaur, you can justify the massive valuations required for “Revit-killers.” However, in our world, software isn’t just a tool; it’s a legal and professional infrastructure. You don’t “disrupt” a building code or a professional liability insurance policy with a better UI.

Innovation is happening; it just doesn’t look like a “unicorn”

The a16z narrative suggests a stagnant industry devoid of progress, but that is a fundamental misunderstanding of how construction actually evolves. Innovation in AEC isn’t absent; it is distributed, specialised, and submerged. The article treats the industry as if it begins and ends with Revit, ignoring the reality that a modern project is powered by a massive, heterogeneous ecosystem of tools; from Rhino and Grasshopper for complex geometry to highly specialised structural analysis and thermal simulation engines.

More importantly, the industry is currently fuelled by a thriving open-source community and an increasing integration with GIS. While VCs look for the next proprietary “king” to invest in, industry professionals are building the future from the ground up through vendor-neutral standards like IFC and open-source projects like Bonsai, IfcOpenShell, and the Ladybug Tools suite for environmental analysis.

This isn’t just about software on a screen; it’s about 2026-level hardware-software integration. We are seeing an explosion in field-first tech, process mining, 3D laser scanning, and robotic total stations, that has fundamentally changed site precision. From Grassroots Automation, where BIM managers use Dynamo and Python to solve daily bottlenecks, to the shift toward Industrialised Construction and pre-fabrication, innovation in AEC grows from the bottom up. It is driven by solving specific, high-stakes problems on-site rather than imposing “top-down” platforms from a Silicon Valley boardroom.

The “move fast and break things” cultural clash

You are likely remembering Katerra, which is the “poster child” for the Silicon Valley construction failure. It was the ultimate embodiment of the “move fast and break things” ethos applied to a world where breaking things is incredibly expensive and legally perilous.

The Katerra collapse (2021)

Katerra was founded in 2015 by Michael Marks (the former CEO of Flextronics) and was backed by billions from the SoftBank Vision Fund. It followed exactly the narrative in the a16z article: that construction is a “broken” industry that needs a tech-first, vertically integrated “Amazon of Construction” to fix it.

• The Vision: They wanted to own the entire value chain. They bought architectural firms, engineering companies, and built massive factories to manufacture cross-laminated timber (CLT) and modular components. They even developed their own proprietary software to manage it all.

• The Reality: They learned the hard way that construction is hyper-local. A “standardised” module designed in a Silicon Valley lab often didn’t meet local building codes in Seattle or weather requirements in Texas.

• The Failure: By trying to “boil the ocean” (doing everything at once), they created an astronomical overhead. When projects were delayed—as they always are in construction—their massive factory costs didn’t stop, leading to a spectacular bankruptcy in June 2021 with over $2 billion in capital lost.

Veev (2023)

If 2021 feels too long ago, you might be thinking of Veev.

• Veev raised over $600 million to build high-tech modular homes using laser-cut steel and proprietary digital walls.

• Despite having a finished product that people actually liked, they couldn’t survive the high interest rates and the sheer “lumpiness” of construction cash flow. They shuttered their operations in late 2023 and were eventually sold for parts.

Both of these companies fell into the predictable trap: they tried to treat a fragmented, relationship-based industry like a software stack.

• Coordination over control: In the industry, “fragmentation” isn’t a bug; it’s a way to spread risk across 20 different specialised companies. Katerra tried to take all that risk onto one balance sheet as a vertically integrated solution. Vertical integration is effective in some markets, but it is not a solution that works everywhere. If there is one thing we learned from capitalism is that specialists are more efficient and effective. Strange that silicon valley keeps forgetting.

• The hubris of the platform: Both companies believed that “better software” and “better manufacturing” would override the need for local expertise.

The Oracle vs. the Unicorn

This does not mean our industry is ‘un-investable.’ While VCs chase the next “disruptor,” Warren Buffett’s consistent stake in homebuilders proves that the industry’s tangible assets and essential demand make it a valuable investment for those who prioritise reality over hype. Buffett understands that in construction, the moat isn’t the code: it’s the land, the permits, and the specialised expertise. On the other end of the longterm/short term investment spectrum, Michael Burry just announced he bought more ADSK.

Fragmentation is a feature, not a bug

It is a strange irony that Silicon Valley, the supposed vanguard of modern capitalism, frequently ignores the system’s most basic principle: the division of labor. Since the days of Adam Smith, we have known that economic progress and efficiency are driven by specialisation, yet tech-disruptors continue to view the AEC industry’s “fragmentation” as a failure of coordination rather than a triumph of specialised expertise. The Silicon Valley mindset views this fragmentation as a technical bug to be solved, but in the built environment, it is actually an essential feature of systemic anti-fragility.

While articles from firms like a16z lament the $177 billion lost to rework, industry veterans recognise that this “waste” is often a byproduct of specialised risk management; in a $500 million hospital project, we don’t have thirty different firms because we lack the tech to unify them, but because they provide a vital liability buffer that decouples risk across multiple professional licenses and insurance policies.

A structural engineer needs their own “silo” to protect their professional license, and an MEP consultant requires their own environment to validate calculations, ensuring that a single catastrophic error doesn’t bankrupt the entire project or its stakeholders.

Silicon Valley views this as a “data silo” problem that AI can bridge, but we know it is a liability and incentive problem: until a software company is willing to sign off on the structural integrity of a beam, “unified cloud platforms” will always hit the wall of professional accountability. By attempting to vertically integrate every discipline into a single, centralised “stack,” VCs are effectively betting against the very specialisation that drives precision and innovation in the physical world, inevitably diluting the deep, narrow expertise (from medical gas piping to seismic reinforcements) that ensures a building is safe, compliant, and buildable.

Ultimately, this ecosystem of specialists is exactly what makes the industry resilient, proving that the efficiency of a node of peak competence will always outperform the centralised control of a generalist platform; a reality that the “move fast and break things” crowd continues to overlook at their own multi-billion dollar expense.

The bottom line

To be fair, it is entirely possible that some of the startups currently tucked away in the a16z portfolio will eventually find their footing and succeed. But in an industry defined by physical longevity and legal accountability, a massive valuation is not a guarantee of a lasting legacy. Selectively showing graphs from questionable sources will resonate with skeptics, but does not make it a success.

The a16z article is right about one thing: the potential for growth is enormous. But the next great leap in construction won’t come from a startup that tries to ignore the industry’s complexity.

The winners will be the tools that embrace our fragmentation, respect our liability models, and enhance the incredible expertise already present in our workforce. We don’t need a “Revit Killer” designed in a Palo Alto coffee shop; we need tools built by people who have actually spent a Tuesday afternoon in a job-site trailer trying to coordinate a clash between a 24-inch duct and a structural brace.

Is Silicon Valley actually interested in fixing construction, or are they just looking for the next place to park their AI capital? Let’s discuss in the comments.