Every construction AI pitch of the last two years has assumed a layer the industry has not finished building. Ask a model which products meet a project’s fire, acoustic and embodied-carbon requirements and it needs to know what those requirements mean, what the manufacturer’s datasheet means, and that the two describe the same property. Architecture standardised how it models geometry decades ago; the people now trying to fix the next layer argue it never did the same for what goes inside the geometry. They are quoted below making that case.
On 30 September the Innovation Design Consortium — a body of large American architecture and engineering firms — named Acelab its co-development partner on an open material data standard. Acelab gets a seat in the working groups, contributor credit on the standard, and reference implementation rights. The standard itself, per the consortium’s announcement, is to be built on an open ontology that gives manufacturers, firms, owners and contractors a shared structure for material data, so that product information is published once and reused without re-entry.
The consortium’s technical product manager, Jim Smell, framed the problem in one line worth quoting exactly: “Architecture never agreed on what its data means, so the data moves but the meaning doesn’t.”
Two things to get right: the name and the domain
The organisation is the Innovation Design Consortium, abbreviated IDC, at idc-aec.com, and
that is the form its own site uses. The naming in the coverage is inconsistent enough to send a
reader to the wrong place.
ENR’s piece
is headlined “Acelab Named First Materials Partner of AI Group—the Innovative Design Consortium”.
The phrase “AI Group” appears in that headline and in the page’s metadata, and nowhere in the
article’s prose, which opens with the correct “The Innovation Design Consortium”. The “Innovative”
rendering also appears in the photo credit and caption above the body. Search for “Innovation”.
There is a second trap worth flagging, and it is the kind that has produced published errors at this
publication before. Acelab, the materials platform in this story, is at acelabusa.com. The domain
acelab.com resolves and belongs to an unrelated business — ACELAB × MakerMex, which sells
large-format robotic 3D printing. They are different companies, and only the first is in this
story.
Who the consortium actually is
IDC’s own About page describes it as a public benefit corporation, governed by a seven-person board drawn from member firms, with a steering committee carrying one representative per founding firm and meeting monthly, and a six-member executive committee. It says the consortium was established with funding from thirty-nine firms, all of them members of the AIA Large Firm Round Table, ranging in size from about 150 people to over 2,000. Membership is planned to open to all firms on a subscription basis after an initial organisational year.
The Acelab announcement puts current membership at 46 firms, employing an estimated 30,000 design professionals and delivering more than 10,000 projects a year. IDC’s homepage describes “over 40 founding firms.” These are three different counts of three different things — initial funders, current members, and founding firms — stated in three documents, and this publication reports each scoped to the document it comes from rather than picking one.
The named members are the point. ENR’s account lists SOM, HOK, HKS, Leo A Daly, NBBJ, ZGF, CannonDesign, Corgan, DLR Group, SmithGroup, The SLAM Collaborative, Page, Gresham Smith and HLW among them, and describes the member firms as having pledged to advance shared standards, interoperability and AI readiness. That is a meaningful share of American institutional and commercial practice agreeing to work on a common vocabulary — which is the part that has stalled before, and not for want of good schemas.
The number that makes this worth doing
The economic case, as reported by Building Design+Construction on 2 October and set out in IDC’s own announcement, rests on one figure: materials account for roughly 40% to 50% of the construction cost of every building. Both documents cite research from Autodesk and FMI estimating that bad data cost the industry $1.85 trillion in 2020 and drove 14% of all rework. Those are the cited researchers’ figures, reported here with attribution rather than independently reproduced.
Dave Lemont, Acelab’s executive chairman, has the standing to make the comparison and made it in the consortium’s announcement: “I watched this industry standardize how it models buildings. It never standardized the materials that go into them, and materials are half the money.” Lemont led Revit Technology through its 2002 acquisition by Autodesk, so the modelling standardisation he is describing is one he was commercially inside. The same statement has him calling IDC “the first body with the scale and the mandate to fix that”.
Acelab’s side of the ledger is its Material Hub platform, which its site describes under the heading “AI powered material intelligence” and quantifies as 250,000-plus building products, 10,000-plus brands, 2,000-plus certifications tracked, 30-plus building and energy codes, 200-plus performance metrics and 50-plus aesthetics attributes. Those are the company’s figures. The relevant asset for a standards effort is arguably less the count than what producing it required: somebody has already had to decide, product by product, which field a given manufacturer claim belongs in.
On “first”, which deserves scoping
IDC’s announcement describes this as a first-ever open material data effort, and trade coverage has filed it as the consortium’s first open data standard for materials. That framing belongs to IDC, and a reader should know what else is already in the field before accepting it.
The nearest existing work is the Common Materials Framework, run by the 501(c)(3) mindful MATERIALS. Its CMF Implementation Toolkit publishes a “CMF Prioritization v1.0” that, in its own words, organises “certifications, data points, and impact areas into one clear structure,” alongside a “Data Ecosystem v1.0” that “links manufacturers, standards, and workflow tools so sustainability data flows seamlessly across the value chain,” with role-based toolkits for AEC firms and owners, manufacturers, technology vendors and ecolabel standards bodies.
The CMF is not the only neighbour, and the differences are not all of scope. buildingSMART’s bSDD — the buildingSMART Data Dictionary — describes itself in its own official repository as “an online service for hosting data dictionaries containing classifications, their properties, allowed values, units, translations, etc,” reachable by software through a public API and a search portal. That is general construction property data rather than sustainability data, which puts it closer to the IDC description than the CMF is. OmniClass, MasterFormat and IFC property sets have occupied adjacent ground for years.
Whether a new effort is the “first” therefore depends entirely on where the line around “material data standard” is drawn — and that is IDC’s line to draw, not this publication’s to ratify. So the claim is reported here as IDC’s, and not repeated unqualified. The question worth putting to the consortium is not whether somebody got there first. It is whether its standard will map onto what already exists, or whether specifiers will end up maintaining several.
Why a standard, and why now
This section is analysis. The timing is not an accident of good intentions. IDC’s other recent move was taking on Autodesk as a strategic partner in March, explicitly to advance data leadership and AI readiness, and to put the consortium’s work on a platform where it can be implemented at scale. A standard with no reference implementation is a PDF. A standard with a vendor that already catalogues a quarter of a million building products, and a platform partner that owns the authoring tools, is a different proposition.
It also fits a pattern this publication has been tracking all year, in which the most consequential construction AI work is unglamorous plumbing. Virginia Tech’s coalition is building a digital commons so builders can compare notes on tools they cannot individually evaluate. Neuron Factory raised strategic money from Trimble and Suffolk to build a construction knowledge graph. ONESTRUCTION raised roughly $58 million in Japan to turn unstructured construction data into something a model can use. Revizto shipped an MCP server to point outside models at live project data. Four unrelated bets on the same conclusion: the models are fine, and the inputs are the problem.
What is not yet known
A good deal. IDC’s announcement gives no timeline beyond the partnership itself, and no document this publication fetched — the announcement, IDC’s About page, the ENR piece or the BD+C write-up — sets a version number or a date for a first draft. The announcement names what Acelab receives — working-group seat, contributor credit, reference implementation rights — and describes the ontology without saying what any manufacturer would be obliged to publish into it. That obligation is where product-data standards have historically either worked or quietly stopped.
The question to hold IDC to is not whether 46 firms can agree on a schema. It is whether a schema 46 firms agree on can compel the manufacturers who actually hold the data to populate it. Materials may be half the money, but the specification is written by one side of the transaction and the datasheet by the other.
