Construction · Deep dive

Higharc

The homebuilding cloud that treats a house as a database, not a drawing — and is now trying to push that database into the lumber yard.

emerging

The question that decides it: Will distributors and builders accept AutoTranslate's AI-derived material quantities as the contractual basis for a bid — or keep re-checking every takeoff by hand, leaving Higharc a drafting tool rather than the supply chain's system of record?

Snapshot

Higharc sells production homebuilders a single, structured representation of a house — geometry, options, construction standards, code constraints — and then generates every downstream artefact from it: permit-ready construction documents, material takeoffs, purchase orders, photoreal renders and a browser-based buyer configurator. Founded in Durham in 2018, it raised a $95M Series C led by Insight Partners announced 30 June 2026, taking total funding past $170M. Its customers build roughly 39,000-40,000 homes a year representing about $19.1B of new-home sales volume (company figures, mid-2026). The Series C came paired with an agreement to put Higharc’s new AI estimating engine inside US LBM, the largest privately held full-line building-materials distributor in the country — which is the actual news. Higharc is trying to stop being a design tool and start being the data layer the supply chain runs on.

Founding story

Marc Minor is a producer, not an architect. He ran production at Apple, was a group executive producer at BBDO, worked at R/GA, and then handled the launches of Carbon3D and Desktop Metal — two companies whose entire thesis was that generative software should drive physical manufacturing. In 2018 he tried to build his family a home in Durham and found the design process slow, opaque and analogue in a way that seemed impossible for an industry that produces a million homes a year. That mismatch — generative design had already reshaped how objects get made, but not houses — is the founding insight.

The technical co-founders make the thesis credible. Michael Bergin ran Autodesk Research’s generative design work for the built environment after graduate research on mass-customised housing. Peter Boyer was a principal engineer at Autodesk and helped build Dynamo, the computational-design layer grafted onto Revit. Between them they know precisely why the incumbent’s data model cannot do what Higharc claims. Thomas Holt came from video games — seven studios, fifteen-plus titles — because the buyer-facing 3D experience is a real-time rendering problem, not a CAD one. Carl Bass, Autodesk’s former CEO, invested in 2023 and joined as an advisor, which is close to an endorsement from the incumbent’s own alumni network.

How it works

The core object is not a drawing. When a builder onboards, Higharc encodes the plan library as structured spatial data: rooms, walls, openings, assemblies, framing rules, option logic, and the builder’s own construction standards. A wall is not a line on a sheet; it is a typed object with a stud spacing, a sheathing spec, a fire rating and a supplier SKU behind it.

From that one model, every artefact is a projection. Move a wall two feet in Studio and the elevation, the framing, the construction document set, the quantity takeoff, the purchase order, the photoreal render and the Showroom configurator all update together. This is the whole argument: in a conventional builder, that same change ripples through CAD, a spreadsheet, a rendering vendor, an ERP and a field system by phone call and markup, and each hop is a chance to be wrong.

The June 2026 addition, AutoTranslate AI, runs the pipeline backwards. Computer-vision models trained on residential architectural drawings read a flat 2D plan image — walls, rooms, fixtures, windows, structural elements — and rebuild it as a 3D spatial model. That model is then passed through what Higharc calls deterministic construction logic: rule-based constraints drawn from building codes and real construction practice, with layered models cross-checking each other and human experts in the loop. The output is a material takeoff mapped to purchasable products, and estimators can click a line in the spreadsheet and see exactly where in the house that material lives. That bidirectional link is the point. It is also the thing a hallucinating model would break most expensively: a wrong sentence is embarrassing, a wrong lumber quantity is a delayed start and a change order.

Product and business overview

Six named modules on one model. Studio builds and manages plans generatively. Config centrally manages communities, options and pricing. Showroom is the buyer-facing online configurator. Estimate (formerly Procure) automates purchasing and estimating. Build produces permit-ready construction documents and CAD deliverables. Photoreal generates data-driven visualisation from the same model rather than from a separate rendering pipeline. Higharc AI shipped to select customers in beta in early 2025; AutoTranslate AI, the distributor-facing product, arrived with the Series C.

The strategic shape is a land grab across the org chart: architecture, sales, purchasing and field all end up working off one file, which is how a drafting tool becomes an operating layer. The US LBM deal extends the same model to the other side of the transaction — a distributor generating takeoffs directly from the builder’s plan set.

Business model and pricing

Annual platform subscription with unlimited users, sold as a custom package per builder: implementation and training up front, then a recurring licence covering Studio, Config, Showroom, Estimate, Build and Photoreal, plus services such as exterior renders and dedicated support (Higharc pricing pages and third-party listings including G2 and SoftwareFinder, accessed July 2026). There is no published price list and no per-seat rate — the pricing page is a demo request.

Two things follow. The unlimited-user structure is deliberate: Higharc wants the estimator, the sales counsellor and the purchasing manager in the model, not just the drafter, because that is what makes it hard to remove. And revenue scales with builder size rather than seats, so the motion is enterprise, top-down and slow, with no self-serve floor to fall back on when housing turns.

Traction over time

Higharc discloses no revenue, no ARR and no customer count. The Deloitte placement is the most useful proxy available: rank 28 of 500 on a list scored purely on FY2021-2024 revenue growth, in a cohort whose overall growth ranged from 122% to nearly 30,000%. That implies a very steep multi-year curve off a small base — which is exactly what a 2021 base year would produce. Treat it as evidence of velocity, not of scale.

The customer evidence is thin but real: named builders include Epcon Communities, Buffington Homes, Jagoe Homes, Highland Homes and Signature Homes — regional and mid-size production builders, not the national top ten. Company-reported outcomes (July 2026, self-reported and unaudited): product development compressed from months to weeks, time-to-community-opening cut 25-50%, margins up 10-15%. Notably, growth continued through 2025’s poor housing cycle, which is the more interesting datapoint: builders bought efficiency software while starts were soft.

Market analysis

Single-family starts were running at roughly a 1M annualised rate as of March 2026 (US Census), and the NAHB estimates custom homes were only about 20% of 2025 single-family starts — the rest is repeatable production building, which is Higharc’s exact target. That is the demand backdrop.

The software TAM is much smaller than the construction flow it sits on, and honest pages should say so. The US home construction design software market was about $1.9B in 2024, projected to roughly $3.3B by 2033 (Verified Market Reports); the broader US construction and design software market was around $2.4B in 2023, growing about 8% a year through 2030 (Grand View Research); residential construction estimating software specifically was about $504M in 2025 (Market Research Intellect). Higharc’s addressable slice is a few billion dollars, growing at high single digits — respectable, not enormous.

Which is why the US LBM move matters more than the funding. Distribution attaches Higharc to the materials channel — US LBM alone runs 450-plus locations and was doing north of $3.5B of revenue when Bain Capital bought it in 2020 (Bain and Platinum Equity have shared ownership since October 2023). Software priced against a drafting budget is a modest business. Software priced against material spend is not.

Competitive intel

Autodesk is the wall. Revit and AutoCAD are the default, they are already in the building, and the architect of record already stamps their output. Higharc’s counter is that Revit was built drawing-first and cannot keep a thousand-home plan library, its options and its estimate in sync. That is true and it is also a hard sell to a director of architecture whose team has fifteen years of Revit families.

Hyphen Solutions is the underrated threat. Its BuildPro/SupplyPro rails already connect most large builders to most of their suppliers — the exact builder-to-distributor lane Higharc just entered with US LBM. Hyphen has the distribution; Higharc has the model. ECI, whose MarkSystems ERP Higharc integrated with in February 2026, owns the financial system of record and is a partner only for as long as that is convenient.

Below them: Cedreo and the design-only tools, which are cheaper and faster but stop short of permit-ready documents; and a swarm of AI takeoff startups — Togal, Kreo, Beam, CountBricks, STACK — attacking the estimating wedge without asking anyone to replace CAD. And the true incumbent, always, is the builder’s own drafting department plus an offshore CAD vendor.

History and evolution

• 2018 — Marc Minor’s own stalled home build in Durham; Higharc founded. Bergin, Boyer and Holt join after the first cheque.
• May 2019 — $4.7M seed (Pillar VC, Lux, Javelin, Vertex).
• Apr 2021 — $21M Series A led by Spark Capital.
• Jan-Jul 2023 — A strategic financing assembled in public: Simpson Strong-Tie (January), Ferguson and Starwood Capital (March), then $15M from Standard Investments, Home Depot Ventures and Carl Bass (July), taking total raised past $40M. Building-products companies buying into the design layer is the tell of what the company would later become.
• Feb 2024 — $53M Series B led by Spark and Pillar, with SE Ventures, Home Depot, Fifth Wall, Ferguson and Suffolk Technologies.
• Feb 2025 — Higharc AI enters beta with select customers.
• Dec 2025 — Ranked #28 on Deloitte’s Technology Fast 500, growth achieved through a bad housing market.
• Feb 2026 — ECI Software Solutions partnership; joint MarkSystems integration.
• Mar 2026 — Fast Company Most Innovative Companies.
• 30 Jun 2026 — $95M Series C led by Insight Partners; AutoTranslate AI and the US LBM estimating agreement announced the same day.

What people say

The case for. Reviewers on G2 — where Higharc took 23 badges across five categories in the Spring 2025 report — consistently praise ease of use relative to Revit, the speed of the design tooling, and, unusually for enterprise construction software, the implementation experience: builders describe communication and timelines being met, and at least one credits Higharc with its fastest-ever community launch. The recurring structural theme is single-source-of-truth: reviewers report that having customer selections and construction documents in one place removes the reconciliation work between sales and construction. Trade coverage is warm — HousingWire’s Builder’s Daily (July 2026) framed Higharc as the first credible test of whether builders will trust AI outputs in live production, and Insight’s own investment note (July 2026) argues the platform compounds because each builder’s standards and construction logic get encoded into the model, making it progressively harder to rip out. On the employee side, Glassdoor shows 4.5/5 across 28 reviews with 91% recommending the company (accessed July 2026) — a small sample, but a good one.

The complaints. They are specific and they cluster on maturity. G2 reviewers report a real learning curve; missing features including fixed interior dimensions and incomplete construction-document sets; weak casework — cabinets and trim; browser lag during training; and a drafting methodology built on rooms and logic rather than the geometry-first habits people learned in AutoCAD and Revit. More than one reviewer says plainly that Higharc is a work in progress without the decades of development behind Revit. For a product whose entire promise is permit-ready documents, “incomplete CD sets” is not a cosmetic gripe. Above that sits the industry’s structural objection: liability. A professional stamp is a licensed human taking personal responsibility for a drawing’s safety, and design-liability commentary (Common Edge, 2024-2025; construction-law practices writing on AI in design, 2025) is consistent that when an AI-generated drawing fails, the exposure lands on the architect or engineer of record, not on the software vendor. Minor himself concedes technology is not a panacea. Employee reviews, even positive ones, flag initiative fatigue and unclear, still-forming processes. And the loudest signal is a silence: for a company positioned as the AI operating system for residential construction, there is essentially no organic discussion of it on Reddit’s construction and homebuilding forums — no crowd of drafters arguing about it. That is what pre-mainstream adoption looks like.

Outlook: the open question

For Higharc to work, an estimator at a US LBM branch has to look at an AI-generated takeoff and cut a purchase order without re-counting it; for Higharc to stall, that same estimator keeps a spreadsheet open beside it, and the product stays a very good drafting tool with a $170M price tag. Everything else is downstream of that.

The bull case is coherent and unusually well-financed. The data model is the right primitive — a house is a repeated manufactured product, and representing it as a drawing rather than a database is a category error Revit is architecturally unable to fix. The founders came out of Autodesk Research and know where the incumbent breaks. Growth continued through a bad 2025 housing cycle. Home Depot, Ferguson, Simpson Strong-Tie and Schneider are all on the cap table — the materials industry paying for a seat next to the model. And US LBM, if it works, changes the pricing reference from a drafting budget to material spend.

The bear case is equally clear. Higharc has never published revenue, ARR, customer count or retention — for a Series C company claiming to be infrastructure, that is a conspicuous omission, and the Deloitte ranking is a growth-rate proxy off a tiny base, not evidence of scale. Its named customers are regional builders, not the national top ten, and the top ten are exactly who Hyphen already owns. The product complaints go to the heart of the claim: incomplete CD sets and weak casework are what a builder discovers three months into an implementation. The liability regime has not moved an inch — a licensed human still stamps the drawings and still eats the error — which caps how autonomous any of this can get. And distributor estimating is a knife fight with a dozen cheaper AI takeoff startups that never asked anyone to replace CAD.

Watch three things over the next twelve months: whether US LBM expands the AI estimating agreement past a pilot into branch-wide deployment; whether a top-ten national builder signs; and whether Higharc ever discloses a retention number. The first tells you the takeoffs are trusted. The second tells you Hyphen and Autodesk can be displaced. The third tells you the operating layer is real.

Sources and further reading

• Insight Partners — Behind the investment: Higharc and the AI-native future of homebuilding (Debayo-Doherty, Fredberg, Leibowitz, Matus, 7 Jul 2026)
• HousingWire / The Builder’s Daily — $95M funding, US LBM deal back Higharc AI push in homebuilding (John McManus, 6 Jul 2026)
• PR Newswire — Higharc raises $95M Series C to scale AI for homebuilding (30 Jun 2026)
• Higharc newsroom — Named #28 on Deloitte’s 2025 Technology Fast 500 (2 Dec 2025)
• Higharc newsroom — $53M Series B for the Connected Homebuilding Cloud (Feb 2024)
• Inman — Home Depot Ventures backs Higharc; $15M strategic round with Standard Investments and Carl Bass (18 Jul 2023)
• Inman — Digital home designer Higharc raises $4.7M seed round (16 May 2019)
• G2 — Higharc reviews: pros, cons and user dislikes (accessed Jul 2026)
• Glassdoor — Higharc employee reviews, 4.5/5 across 28 reviews (accessed Jul 2026)
• Common Edge — The liability question: when AI designs buildings, who’s responsible? (2024)
• Hyphen Solutions — BuildPro scale: 22 of the top 26 US builders, ~1 in 3 US homes (accessed Jul 2026)