Beyond design software: can AI actually accelerate infrastructure development?
The Build founders: AI researcher Ben McCClusky and architect James Stirrat-Ellis. The latter previously worked on projects including the world-renowned Changi Airport T5 in Singapore,
London and New York-based Build announced on June 30. that it has raised $8.5 million in seed funding in a round led by Index Ventures to expand an artificial intelligence platform designed to accelerate infrastructure development. The company says the platform has already been deployed across more than 100 projects in 15 countries, helping governments, developers and investors shorten processes such as site selection, technical due diligence, power assessments and early-stage design.
At first glance, the proposition sounds familiar. Architecture, engineering and construction firms have spent years experimenting with generative design, digital twins and machine learning tools intended to improve efficiency. Software vendors routinely promise shorter timelines, lower costs and better decisions.
Build argues that it is pursuing something different.
Rather than selling software licences, the company says it delivers completed infrastructure work itself. Its AI systems analyse information from more than 1,600 datasets and carry out assessments in parallel, while human reviewers validate outputs before they are delivered to customers.
The distinction is important because infrastructure development remains heavily dependent on manual processes. Before construction begins, developers typically evaluate planning restrictions, environmental conditions, utility connections, transportation links and political considerations, often through multiple consultants operating independently.
Build claims those activities can be accelerated dramatically. The company says project timelines can be reduced by more than 95%.
It is also a claim that deserves scrutiny: how were those reductions measured?
AI has already reshaped architecture and urban planning
Build is far from the first company seeking to apply artificial intelligence to the built environment.
In 2020, Autodesk acquired Spacemaker, a Norwegian startup specialising in AI-assisted urban planning. The software enables architects and developers to evaluate factors such as density, sunlight exposure, wind conditions and noise levels before detailed designs are produced.
Autodesk subsequently integrated the technology into Autodesk Forma, its cloud-based platform for early-stage planning and conceptual design.
Companies such as TestFit have focused on automating real estate feasibility studies, generating residential and mixed-use layouts based on zoning rules, parking requirements and site constraints.
Meanwhile, Cove.tool uses machine learning techniques to support building performance analysis and energy modelling, helping architects assess sustainability considerations earlier in the design process.
The use of digital twins has also expanded considerably over the past decade.
Infrastructure software providers such as Bentley Systems and Dassault Systèmes have developed platforms that allow operators to create virtual representations of airports, industrial facilities and transportation networks. These systems combine design information with operational data to improve maintenance, planning and asset management.
In other words, Build enters an architecture, engineering and infrastructure market that is already experimenting extensively with artificial intelligence.
Its claim to uniqueness appears to lie elsewhere.
Selling infrastructure outcomes rather than software
The most notable aspect of Build's proposition is not necessarily the technology itself but the business model surrounding it.
The company describes its approach as "sell the work, not the software".
That moves Build closer to a consultancy model supported by AI rather than a conventional software provider.
Its platform evaluates planning constraints, environmental risks, power availability and site characteristics simultaneously, rather than through the sequential processes that traditionally define infrastructure development.
According to the company, customers receive completed assessments rather than software tools they must operate themselves.
Why AI infrastructure is creating new bottlenecks
Build's focus on infrastructure also reflects changing market conditions.
Artificial intelligence is increasing demand for data centres, power infrastructure, industrial facilities and transmission capacity at a pace that developers and utilities are struggling to match.
New data centres require access to power, water, fibre connectivity and suitable land. Electrification strategies are driving investments in grid infrastructure, battery production and industrial facilities. Governments are simultaneously attempting to accelerate housing construction and expand renewable energy capacity.
Each of those projects begins with a series of assessments that can consume weeks or months.
Site searches involve examining land ownership, transmission availability, environmental regulations and local planning frameworks. Utility capacity can determine whether a project proceeds or stalls. Permitting timelines vary significantly between jurisdictions.
Build argues that these constraints represent an opportunity for AI.
The company says its systems analyse information from more than 1,600 sources and orchestrate thousands of tasks simultaneously.
What remains unclear is how much of this capability depends on proprietary technology and how much reflects better integration of existing datasets.
Many planning authorities, utilities and environmental agencies already publish large amounts of information. The challenge is often not the availability of data but the ability to combine disparate sources into workflows that support decision-making.
That integration challenge has long been one of the least digitised aspects of infrastructure delivery.
Does human expertise remain part of the process?
Infrastructure projects frequently involve billions of dollars in capital expenditure, long approval processes and extensive public consultation. Political priorities change. Regulations evolve. Community opposition can alter timelines regardless of technical feasibility.
Build acknowledges this reality.
The company says every customer deliverable is reviewed by experienced operators before it reaches clients.
Its broader strategy is to automate increasing portions of the development lifecycle, from site selection and due diligence to engineering, permitting and asset management.
The company refers to this vision as "agentic real estate".
Whether infrastructure owners are prepared to delegate larger parts of the development process to AI systems remains to be seen.
For now, many organisations appear more comfortable using AI to compress analysis than to replace judgement.
That approach may prove more practical.
Infrastructure projects are rarely delayed because engineering expertise is unavailable. More often, delays emerge from fragmented information, sequential workflows and competing regulatory requirements.
If AI can shorten those processes in a meaningful way, companies such as Build may find themselves addressing a problem that has persisted for decades.
Its funding round suggests investors believe there is room for experimentation.
In any case, the company offers an interesting test case for a sector where the challenge has rarely been deciding what to build but determining whether it can be built at all.
Liked this article? You can support our independent journalism via our page on Buy Me a Coffee. It helps keep MoveTheNeedle.news focused on depth, not clicks.