The AI Supercycle: AI’s Emerging Impact on the Built World

This is a sponsored message from BlackHorn Ventures.

Driving Construction Productivity, Profitability and Safety

Welcome to the AI Supercycle - a time of rapid growth and innovation in artificial intelligence (AI) with far-reaching implications for the $14 trillion global construction industry! Breakthroughs in machine learning infrastructure, vector databases, natural language processing and generative AI have captured the public imagination, but their effects are just beginning to be felt in an industry that makes up 14% of global GDP. Similarly to past “IT supercycles” (personal computers, internet, mobile and cloud), the AI supercycle has led to a surge in new software applications. A McKinsey report predicts generative AI applications could contribute $2.6 trillion to $4.4 trillion annually to the global economy. In order to fulfill its potential to increase construction’s stagnant productivity, AI must build trust, reduce data complexity and demonstrate clear ROI for automation solutions.

AI is Not Just Hype—It’s a Paradigm Shift

With relentless attention focused on AI since ChatGPT launched in November 2022, it’s not easy to separate signal from noise. We believe AI represents a fundamental paradigm shift for the construction industry, given the following macro trends: 

  1. Step function advances in neural networks, natural language processing and vector databases are extending earlier AI applications that could categorize data and process rules to generate semi-custom solutions. Solutions built on these emerging AI platforms can now recognize words, ideas and images, and can auto-generate text, summaries, images and even highly realistic videos using deep learning techniques. These generated outputs - so-called “deep-fakes” - can be indistinguishable from originals.
  2. Innovation within the open-source community and new protocols being more widely adopted are greatly simplifying and democratizing the development of new AI applications; and
  3. Increasing enterprise IT budgets for digital transformation, coupled with accelerated cloud adoption and constantly increasing cellular broadband capacity are extending the delivery of powerful AI applications to mobile phones, tablets and GPS tags to optimize the productivity and safety of construction workers, and equipment they use like trucks, bulldozers, motor graders and cranes. 


Example AI Solutions in Construction
Below, we provide examples of AI-powered solutions currently available for construction based on proprietary data assets owned by projects and companies, and anonymized data from across the industry.

AI Applied to a Project’s or Company’s “Proprietary Data Asset”

  • Document Search: Ingesting and structuring the thousands of documents associated with a major construction project allows the use of large language models (LLMs). The LLMs, combined with data dictionaries and vision AI, can interpret text, sketches or symbols on drawings, and provide instant retrieval of, and relationships among, BIM elements, specifications, submittals (Pype), RFIs, T&M tags, invoices, emails, texts, etc., relevant to a particular component, zone, subcontractor, worker or tools and equipment. This saves countless hours for both project and home office personnel to eliminate conflicts and errors (Iris).
  • Forecasting and Alerting: Machine learning can be applied to a unified pool of project data from the multiple, siloed point solutions used by its workers and managers to forecast time and cost at completion early in a project, or to mitigate safety risks to enable timely intervention and to price insurance coverage (Briq, Foresight, Trusstor).
  • Auto-Generating Detailed Designs, Schedules, etc.: Trade contractors and general contractors that self-perform work spend large amounts of time generating detailed “shop drawings” for their own scopes of work. And project engineers spend hours - sometimes days - poring over these shop drawings to detect errors or spatial conflicts across trades, and in revising schedules to accommodate design changes or to recover from project delays. Modern generative AI design platforms can automate much of this work. They far surpass the performance of earlier rule-based AI auto-configuration platforms (Design Power) to auto-generate 3D designs from text and sketches (Augmenta, Generate), and to create project schedules from BIMs or optimize schedules generated by humans (Alice Technologies). 

AI Applied to Industry-Wide Proprietary Data Assets

  • Proprietary data assets spanning the US and global construction industry have long been assembled and sold by legacy companies like McGraw-Hill and IPA as services for bid opportunity identification and productivity benchmarking. Startups can build intelligent AI solutions with broad applicability by leveraging similar kinds of industry-wide proprietary data assets (Buzz Solutions, Ecomedes).
  • Solutions powered by speech recognition, natural language processing and data dictionaries can now: recognize and enter natural language speech as structured data into the correct fields of an ERP or Manufacturing Management System (Datch); verify that required insurance coverage has been secured (Billy); or identify risky clauses in proposed contracts and generate redline responses to counter with preferred clauses (Document Crunch). Similarly, AI image interpretation and machine learning can be used to detect safety violations, assess damage or identify impending failures in buildings and infrastructure (Buzz Solutions).

Conclusions for Incumbents

For decades, construction firms have been coping with growing skilled labor shortages and rapidly rising financing and materials costs, plus pressures to enhance sustainability of their products and processes. AI offers the potential to digitally transform their businesses, unlocking improvements in labor productivity, resource efficiency and emissions reduction. 

  • Large incumbents will often have enough proprietary data to generate valuable AI solutions based on their own data. 
  • At the same time, the required data science and AI expertise to build AI solutions has been greatly reduced by the availability of new foundational AI platforms.
  • Incumbents can now rapidly and cost-effectively build solutions based on data from their own past designs, bids, construction projects, asset operations or consulting services without needing to assemble large AI teams. 
  • In the next decade, this is likely to become “table stakes” for incumbents to defend against disruption from more agile competitors or startups.

Conclusions for Startups

After years of underinvestment, the construction tech space is evolving rapidly. Vertical construction platforms and operating systems will deliver significant value through the integration of AI with a general intelligence layer that acts on firm-wide and/or industry-wide proprietary data assets. Significant opportunities exist for new infrastructure and tooling platforms that enable companies’ data science and product teams to build and scale AI and ML models. 

Blackhorn is actively investing in founders building AI-enabled startups that embrace the following principles to drive the next chapter of construction productivity and profitability:

  • As the AI layer gets increasingly commoditized, startups building vertical industrial AI applications need to focus on defensibility through their use of proprietary "foundational data assets" -  the critical data unique to a market, platform or business outcome. The companies that control the foundational data asset will have outsized competitive advantage in model refinement and training to drive operational savings, labor productivity, safety and emissions reductions and create a highly defensible moat.
  • Portability across cloud, hybrid and on-premise environments paired with a “walled garden” model for protecting proprietary data will ease adoption barriers for customers. 
  • Integrating the AI solution into ERP, PM or other key business systems makes the AI solutions sticky.
  • Positioning needs to be centered on AI’s radically enhanced value (ROI) to users’ employees and/or customers, rather than “AI” itself.


Dr. Ray Levitt is an operating partner at Blackhorn Ventures, along with Stephan Cizmar, partner; Micah Kotch, partner; and Omar Smith, senior associate.