ConTech Predictions for 2026
What could shape construction technology in the year ahead
Note: This is part 2 of the ConTech Funding Report: 2025 where we make predictions based on the funding data.
In the current political and global climate, anticipating the future is incredibly challenging.
Over the course of 2025, there was a rise in macroeconomic uncertainty driven by regulatory and policy shifts in the US due to the election of the new administration. This included tariffs, the acceleration of nearshoring and the war against wind energy. On the other side, there’s been tailwinds for AI data centers, nuclear energy and domestically manufactured goods.
This environment is unlikely to normalize in the near term and it is therefore important for participants in the industry to understand both the wider geopolitical context and the shifting policy landscape as well as deeply understanding the construction industry’s pain points when reviewing themes.
Drawing on the 2025 funding data, investor and founder conversations, and the themes we track weekly, we are interested in the following areas for 2026:
Note: For each of the following themes we will provide an overview and our hypothesis on why we are interested and tracking them. This is not investment advice (please see disclaimer at the end).
Contents
Robotics
Agentic AI solutions
Generative Design
Design-Build / Preconstruction Solutions
The Data Center Rebellion
Robotics
In 2025 there was a growing interest in robotics and autonomous equipment solutions.
From conversations with investors, it appears that the underlying technology has matured to a stage where it is commoditized making it easier for teams to develop new solutions. This is due to two factors:
Software
AI and new Vision-Learning-Action models make it faster and cheaper to develop software for robotics.Hardware
Components such as sensors have become commoditized and cheaper to purchase.
Together, these factors make it easier for founders to build solutions with leaner teams and less funding, drastically lowering the cost of a MVP.
One particular area that we are interested in is robotics development in China. They rank first on robot density (the number of robots per 10,000 employees) and there are over 200 companies developing humanoids compared to 16 prominent firms in the US.
Additionally the ConTech sector in China has a deep robotics talent pool. This is due to a major real estate player called Country Garden which had a subsidiary called Bright Dream Robotics with ~3,000 employees that developed intelligent construction solutions and robots. Due to the property downturn, a number were laid off and a portion either founded robotics companies or joined startups. One notable example in 2025 was Partner Robotics who raised a Series A round with a co-founder from Bright Dream Robotics.
While geopolitical considerations make adoption of Chinese robots in Western markets unlikely, given the talent density and adoption behaviour, keeping a close eye on developments in China and how they can be translated to other markets presents an opportunity.
Agentic AI solutions
In 2025 there was a shift away from chatbot based messaging to workflow based messaging. This was highlighted in June by analyzing how AI construction startups had updated their taglines over the last 12 months. People want AI to be seamlessly embedded into their workflows and to handle tasks proactively rather than waiting for instruction or prompts.
To us, this is a major shift and was reflected in our highest performing Linkedin post of 2025: Systems of Record vs Systems of Work.
To explain this further, traditionally the most valuable software companies have grown by becoming a System of Record (SOR). They replaced paper files and spreadsheets with centralized cloud native software that became the source of truth. The SOR now becomes the backbone for core workflows, compliance and collaboration across teams.
This creates a strong data gravity.
As the SOR has the data, new tools are forced to integrate to it (e.g Procore Marketplace). This allows the SOR to capture even more data, and even when better point solutions exist, data still has to flow back to it, reinforcing its value and making it incredibly difficult to displace.
Borrowing from David Talpalar’s framework, SOR’s have grown because they assume a human-centric world:
Humans do the work
Humans enter the data
Software stores and organizes it
But with agentic systems, AI now:
Participates directly in a core workflow
Generates or captures the output of that work
Stores the resulting data for future use
This has provided a wedge for new AI startups to begin to disrupt the SOR by focusing on the workflow first and the record second. The pattern they are seeing for GTM is:
Enter through a high-frequency workflow
Do meaningful work, not just assist
Capture data as a byproduct
Store it in an AI-native way
Expand until the legacy system becomes redundant
This is interesting as I believe we are in the midst of a shift of where the data gravity is.
By entering through the workflow, AI is able to capture the chain of reasoning that is native to how users actually work. For example, when using a copilot or chat based interface, startups can record the steps, decisions and feedback that make up a real workflow, alongside the surrounding context. This reasoning data allows workflows to be tailored to a company and even to individual users, so systems adapt to how people work rather than forcing users into rigid interfaces.
Startups that capture the most high quality reasoning data could emerge as winners, because they can fine tune models to deliver marginally better outcomes. While a one percent improvement may seem small, in large, multi step workflows these gains compound across each step, making agents meaningfully more accurate and useful over time. For example, for a 10 step workflow at 95% accuracy per step, the probability of a fully correct outcome is about 77% (0.955). At 90 percent accuracy per step, this drops to 59% (0.95).
This reasoning data may become a moat. While it is possible to switch underlying AI providers, a new platform would still need time to relearn the reasoning patterns embedded in workflows which are specific to a company. That knowledge may not be transferable, creating a switching cost around fine tuning and performance.
Whether this holds in practice remains to be seen.
In conversations with CTOs, the consensus seems to be that we are capturing new forms of data but it is still unclear how defensible or durable this advantage will be or how effectively it can be harnessed today to fine tune a model (if anyone knows, would love to chat!).
Generative Design
Could this 2026 be the year the billable model finally gets broken?
AECOM’s acquisition of Consigli for $390m was our highest performing post of the year (and ranking #2 on Google!). The CEO’s earnings call was illuminating as they highlighted that the way design is to be undertaken will fundamentally change:
“using AI with our teams that we can deliver things at an incredibly fast pace. So in the design process, what might take months, AI can do almost instantaneously…And then for us, it profoundly changed the ways that our team will work.”
Traditionally it has been challenging to adopt design solutions due to the billable hour model. When you are paid by the hour or on a cost plus model, there is a perverse incentive to work to the budgeted hours rather than trialling a system which costs less as it makes you less billable.
The mindset is built into design engineers as every week you need to put in a timesheet which starkly shows your billability. As soon as you drop under 80% billable hours (projects vs overheads), you’re on the cutting block and a month of this leads to conversations and the potential for layoffs.
Now that AECOM, the ENR #1 Design Firm is saying design will change due to AI and they are testing new processes, we could see widespread change. Already Consigli’s competitor Endra was able to raise $20m in Seed funding, the largest ever for a Swedish startup as the acquisition is a signal to investors on the potential of this market.
Design-Build / Preconstruction Solutions
Design-Build (DB) is one of the fastest growing and most widely used project delivered methods in the US. By 2028 it is projected to account for more than 47% of construction spending.
In this method the owner signs a single contract which covers both design and construction services. It contrasts with Design-Bid-Build (DBB) where the owner first contracts a designer to produce drawings, then puts the project out to tender and finally signs a separate contract with a contractor to deliver the build.
Because the designer and contractor share one contract in DB, preconstruction is much more important. If the design contains errors or gaps, it is more challenging to submit variations. Tighter coordination is required on constructability with design changes requiring tighter feedback loops and immediate notification to the contractor.
For example there may be a long lead time for a product and a design substitution must be implemented to avoid delays later. Enabling and supporting this shift and the surrounding workflows is an opportunity.
The Data Center Rebellion
The US GDP growth is incredibly dependent on AI capex.
As Harvard economist Jason Furman put:
Investment in information processing equipment & software is 4% of GDP. But it was responsible for 92% of GDP growth in the first half of this year. GDP excluding these categories grew at a 0.1% annual rate in H1.
This has impacts on the industry as construction spending on data centers has hit an all-time high of $40 billion, up 30% from last year. Data center spend is now on track to surpass total office construction, a gap that was almost $60 billion just three years ago.
The buildout has had national support with Executive Orders however we are starting to see the first signs of pushback.
As the Energy Secretary noted:
“In rural America right now, where data centers are being built, everyone’s already angry because their electricity prices have risen a lot…I don’t want them in my state’ is a common viewpoint.”
This is reflected in the numbers.
Data center watch reported Q2 2025 as a turning point in development risk. 25 projects were blocked in 2025, with 21 of those cancellations occurring in the second half of the year. It’s a sharp increase from 6 cancellations in 2024 and 2 in 2023.
Even utilities are pushing back as power officials have been raising concerns that data centers are pushing parts of the US power grid to failure. The problem is acute within the PJM Interconnection, largest power grid operator, proposing that data centers must bring their own power source or otherwise agree to have service cut off when supplies get too tight.
Prices have risen due to the data center development in the PJM region and now a group of Senators have launched an investigation on the relationship between the construction and utility costs. Big Tech providers like Microsoft are attempting to mitigate concerns by creating community first plans, promising to pay premium utility rates to ensure costs are not passed to local residents and to replenish more water than it uses.
The Trump administration is also stepping in, asking PJM to hold an auction for 15-year contracts for new generating capacity and asking tech companies to bid on it, even if they don’t need the power for their data centers.
Despite these promises, as 2026 progresses local opposition could coalesce nationally as independent groups, frustrated by rising costs and new data center development, begin to actively learn from each other and push officials. The issue is bipartisan, occurring in red and blue states and could have repercussions for the construction industry and the wider economy.
2026 has the potential to be an impactful year for the ConTech ecosystem as investment in physical AI solutions is accelerating, with the industry interested in piloting and adopting. At the same time, digital AI capabilities continue to advance rapidly, raising open questions around where value ultimately accrues and how legacy software providers defend their positions.
Moves such as Procore’s acquisition of Datagrid, AECOM’s purchase of Consigli and Turner Constructions adoption of OpenAI shows that the industry is actively experimenting. AI is increasingly viewed as both a competitive threat and a strategic opportunity for firms willing to move early.
For startups, this marks a meaningful shift. Buyers are no longer just open to new solutions, they are actively seeking them, creating a window where market pull is strong and clear winners have yet to emerge. In 2026, we could see an increase in acquisitions as incumbents look to build differentiation by acquihiring teams for their execution speed and AI capability, while preserving optionality in a market where long-term moats remain uncertain and fast-following may prove strategically valuable.
The SOR vs System of Work framework is incredibly clarifying for whats happening in construction tech right now. The compounding accuracy gains across multi-step workflows is the kind of structural advantage that could actually shift where value accrues. Watched a similar pattern in other enterpris software where workflow capture became more valuable than data storag alone.