Construction & Infrastructure Executive Leadership Insights

Everywhere you look, LinkedIn, ‘almost every post’, the web, everywhere! You see the hallmarks of AI generated information. Text, visuals & the highly polished ‘Oscar’ worthy advert or post from business of all industries promoting their organisations, their products & their projects.

Digital technologies use in the construction industry is far from a new phenomenon, it’s been utilised for decades. My first understanding of wide adoption of digital use was Computer Aided Design (CAD). The system wasn’t adopted by everyone early stage (much like the current uptake of AI) but gradually CAD advanced through architect practices, filtering down to the Tier ones & soon became the standard.

The CAD standard then advanced to Revit in the early 2000’s. The introduction of Revit software was hugely successful & Revit was one of the first software packages that brought Building Information Management, or (BIM) to life. BIM has advanced tenfold since & is advancing further with AI.

In the construction, infrastructure & manufacturing sectors, I have seen both sides of the AI uptake.

I’ve seen organisations adopt as many AI interfaces as possible throughout their businesses, whilst knowing others who still don’t really understand what it is, how they could adopt it or what impacts it could or can make.

The best adoption witnessed was AI fully integrated within the business. Following a couple of weeks uploading schedules, pricing documents, detailed preliminary books including pricing schedules, schedules of rates for labour & a schedule of rates for materials. The AI system was utilised ‘end to end’ & used throughout the company on projects from concept through to completion.

As an example,

A tender is received & all the tender documents are uploaded to ChatGPT / Claude AI. The system scans every document & produces a ‘high level summary’ of the tender. Typical time to complete, 2 or 3 minutes.

Now let me explain what was reviewed in those few minutes,

• Invitation to tender
• Instructions for bidders
• Form of tender
• Contract conditions
• Technical specifications
• Design drawings
• Employer’s requirements
• Pricing schedules
• Required delivery programme
• Preliminaries allowed – schedule of attendances
• Pre-construction information (if applicable).

All the documents have been reviewed, catalogued, discrepancies noted & a 3-page high level briefing note produced explaining to the reader, what the tender is, who it’s for, where the project is, the projects timeline, & the desired deliverables.

A few minutes to produce what would take an experienced professional 24 / 48hrs to produce dependant on the amount of reading material in the above noted documents.

The process moves a step further.

The AI has identified when the desired ‘tender opportunity’ is to be delivered on site via the tender docs & reviewed the companies work Calendar. The 3-page summary produced identifies, 1 – we have capacity to deliver this project or 2 – we don’t have capacity as we have 2 or 3 projects running in the same delivery window. The AI already knows what forms of contract & contract conditions the company are comfortable working with. The AI already has a fully comprehensive preliminaries book with all schedules & rates costed out. The AI already has the companies pricing schedules for materials & labour that are reviewed & updated every quarter.

In a few minutes, the AI has produced a 3-page summary document detailing a high-level overview of the company’s availability to undertake with the company’s current Calendar, a yes or no, on the form of contract or contract conditions, suggested preliminaries schedule for the project (taken from the proposed tender programme) & a very basic pricing schedule for materials if the tender information is thorough enough. 99% of the time it isn’t & this is where company managers experience comes in to ‘fill the gaps’ to produce a coherent tender response.

At this point, said experienced manager will also review the work Calendar & ascertain if they can squeeze the project in. (In the knowledge that in construction, things rarely run to programme or Calendar) Said experienced manager will review & can generally have a simplistic tender return completed ‘filling the gaps’ in a couple of days. The completed information will then be re-uploaded to ChatGPT or Claude AI for a sense check before issue to client. Again, with the chosen AI completing the task in a couple of minutes.

I’ve personally witnessed Claude AI review a full JCT 2016 Design & Build contract (with amendments) & produce a suggested ‘change schedule’ to the highlighted onerous elements of the contract & replace with detailed amendments schedule to the onerous conditions. These changes comply with UK law & the contract. This exercise was completed in under 10 minutes!

At the other end of the process, I have witnessed ChatGPT produce a fully comprehensive operations & maintenance manual, for a temporary laboratory, drawing all the information & schedules from the companies’ servers for production. Again, all in a couple of minutes.

This is happening now, companies embracing AI & utilising it adoption widely, are already ahead of the curve. These businesses can produce comprehensive documents in minutes, review extensive documents & tenders in minutes, return detailed tender responses in a quarter of the time previously taken (ultimately translating to more opportunities for the business as more tenders returned equals more opportunities) review complicated contracts in minutes & produce highlighted ‘onerous’ ‘elements to avoid’ schedules with suggested amendments listed, in minutes! & here’s the kicker, most AI systems available in the market today, have a ‘business subscription’ for under £100 per month.

An assistant, that takes no breaks, no holidays, is never off sick, is on call literally 24/7, can read, produce & schedule 100’s of pages of text in seconds, build websites, produce marketing materials & a myriad of other tasks for under £100 per month. AI is a powerful, disruptive technology that’s reshaping workflows in every sector. All that said, AI is not infallible, it still requires human oversight & is reliably unreliable on some tasks.

However, business utilising the technology correctly & leveraging it to do the heavy lifting on data driven tasks are reaping the rewards & are infinitely more efficient than other not using it in their sector.

So, what does it mean?

In the construction sector, the AI boom is real, not the adoption, but the construction & infrastructure required to facilitate its growth. McKinsey predicts that data centre infrastructure investment will be between £6 & £7 Trillion by 2030! The major hyper-scalers being, Amazon, Google, Microsoft & Meta are spending hundreds of billions separately in CapEx to facilitate AI’s growth.

Who remembers the ‘Penetration Pricing’ strategy deployed by Adobe?

Adobe originally sold its software (like Photoshop & Illustrator) as expensive, one time purchase discs. In 2013, they pivoted entirely to a cheap monthly subscription model called ‘Creative Cloud’.

• Adoption, the low monthly entry fee made it incredibly accessible for students, freelancers & agencies to adopt the software. It quickly became the undisputed industry standard taught in design schools globally.
• Dependence, because Adobe created proprietary file formats (like .psd and .ai), entire global industries became entirely dependent on their ecosystem to share files & collaborate.
• Monetarisation, once users accumulated years of archives, & the competition was effectively starved out, Adobe steadily increased subscription prices & cracked down on legacy software users. Because switching to an alternative means losing access to your project ecosystem, businesses & professionals are forced to pay the higher rates.

Herein lies a truth, no organisation will invest this heavily without a thorough & robust ROI strategy. Translating to the investments currently being made, will in time require reimbursement, with interest. So what’s the plan? As previously noted, business can have a ‘business subscription’ for under £100 per month, it’s feasible this type of subscription, won’t stay this price in the future. I believe the current strategy is ‘penetration pricing’ on AI products in the global market.

Prediction 1, AI service costs will increase

The costs for AI services will start ramping up in the next 12 to 18 months. Business & professional subscriptions will be the first to increase, then the lower subscriptions will begin to offer less or limit input or output by either timing or data uploaded or downloaded. This will be the beginning of the ROI strategy.

Prediction 2, The AI infrastructure bubble,

AI advancements are exponential; AI is now actively developing its own future chips & infrastructure! Hardware limitations aren’t the only bottleneck to advancing AI. Developers are relying on machine learning & generative tools to optimise & build next generation hardware. AI is being utilised in data centre architecture to design ultra-low latency networking fabrics, in an effort to manage the immense gigawatt class power & cooling systems required to operate the server clusters.

Here’s the problem, the construction industry can’t keep up with the technology advancements. Data centre design, procurement & construction takes years to come to fruition on any given project.

In boardrooms across the world an uncomfortable truth is being discussed.

Data centres completed or nearing completion are either outdated or obsolete due to the speed in which AI technology is advancing. Real estate, building permits, power infrastructure, cooling technology & the talent needed to construct what’s required is already behind the demand & falling further every given month.

Further pressures are added by New York banning data centre permits for a year. In a video posted to X, Governor Kathy Hochul has stated,

“The scale and speed of development have put unprecedented demand on energy and water resources and threatens to drive up utility costs. Before it goes any further, I need safeguards in place to protect New Yorkers.”

How long before NIMBY’s here in the UK follow suit & petition against data centre developments.

https://www.ft.com/content/1c390476-9d93-4df5-9b41-7381db43b4ea?syn-25a6b1a6=1

A solution to such is already being pursued by SpaceX. Rather than continuing to build ever larger terrestrial data centres, SpaceX have already filed plans for an orbital AI computing network, arguing that space removes many of the constraints currently limiting AI's growth on Earth.

Whist this may sound like science fiction to some, SpaceX launched the very first batch of 60 operational Starlink satellites on May 24, 2019. In July 2026, SpaceX reached a milestone with its latest Starlink launch, sending the satellites into orbit on its 600th flight of a flight-proven booster. The launch came on the second of two Falcon 9 missions that lifted off less than eight hours apart.

This isn't simply about technology. It's about infrastructure. Every AI model, every large language model and every autonomous system ultimately rely on physical assets. Land, power, fibre, cooling, concrete, steel and highly skilled people capable of delivering it, because which out such, AI can’t exist.

There is a belief or line of thought that believes terrestrial infrastructure has become AI's biggest constraint. SpaceX has therefore proposed an orbital AI data-centre network "Starmind" comprising up to one million AI computing satellites powered by continuous solar energy in low Earth orbit. The highly ambitious project faces significant engineering & regulatory challenges which is nothing new to Elon & SpaceX, but its mere existence highlights something important. The world's largest technology companies are already looking beyond Earth for AI infrastructure.

Whether orbital AI computing ultimately succeeds isn’t the point. The important point is that some of the world's largest technology companies already see terrestrial infrastructure as one of the limiting factors to AI's future growth.

Witness the speed in which SpaceX have deployed their satellites into space below.

Could the AI infrastructure bubble be about to burst here on Earth,

https://www.youtube.com/shorts/2oKdgSxWuRs