Construction AI: the Building Safety Regulator's backlog finally shifts, and a central bank flags the bill behind the data centre boom

Here's a story that doesn't come with a launch video or a funding round, and matters more than most that do. The Building Safety Regulator published its latest Gateway 2 performance data in the second week of July, covering the 12 weeks to 28 June 2026, and for once the direction of travel is up. It made 368 decisions in the period, up from 358 the previous 12 weeks, with a 77% approval rate, 284 cases, against 75% before. New-build decisions ran at an 89% approval rate. External remediation, the bit that's been the sorest point since the regime started, reached 85% over the rolling period, well clear of the 65% minimum the BSR set itself for year-end.

The backlog number is the one I'd put in front of a board. Following the External Remediation Improvement Plan the regulator brought in this April, the pile of legacy 2024 applications is down to 14, from 42 at the start of 2026. That's not a rounding error, that's a queue clearing. Process changes, more people, updated guidance and, notably, closer engagement with applicants before submission are all feeding into it. The BSR also published a Remediation Toolkit to walk dutyholders through the control process, which tells you the regulator has worked out that a chunk of the delay was never the rules, it was applications turning up half-built.

But hold the champagne. Higher-risk internal works still sit at a 73% approval rate with a median determination time of 28 weeks. So if your job is a refurbishment inside an occupied higher-risk building, the regime is still asking you to plan around the better part of seven months for a decision, and to expect better than one in four to come back for more work. There were 1,505 live applications in the system at 28 June, and 57% of all decisions related to London. What that means for the AI compliance tools that have been piling into this space, your BuildwellAIs, your Newformas, your golden-thread platforms, is that the regulatory spine they're selling into is finally load-bearing. The regime moving is what makes the record-keeping worth paying for. A golden thread nobody inspects is just an expensive filing cabinet.

For your board pack: Pull your live Gateway 2 submissions and re-baseline the programme against a 28-week determination for internal higher-risk works. If you've been quoting eight to twelve weeks to a client, that conversation is better had now than at week twenty.

For weeks this brief has carried the good-news side of data centres. Savills says vacancy's down to 8%, Turner and Townsend has them as the UK's second most in-demand sector, Clearstone's putting £3bn into Ebbsfleet. So here's the counterweight, and it comes from an unlikely place: the Bank for International Settlements, the institution central banks themselves bank with.

In its flagship annual report, given a fresh run by Bloomberg on 14 July, the BIS lays out the shape of the risk. The five largest hyperscalers are on course to spend more than a trillion dollars on AI-related capex across 2025 and 2026. That spending is running ahead of what those firms earn and ahead of their free cash flow, so some are issuing debt to cover it. Worse, in the BIS's reading, is how it's all wired together. Hyperscalers take equity stakes in AI labs, the labs commit to multi-year purchases of chips and compute from those same hyperscalers, and the terms, in the BIS's own words, are typically poorly disclosed. It flags the risk of the same asset being pledged more than once. The report reaches for canal mania in 1830s America and the dotcom bust for its comparisons, both of which, as it dryly notes, ended in sharp corrections.

Now, I'm not sure the doom scenario survives contact with how much real demand there is for compute, and the BIS's job is to worry for a living. But the direction matters for anyone in construction, because the thing being financed here is buildings. Sheds full of racks, on our land, built by our contractors. A lot of firms have quietly leaned their two-year forward book on the assumption the data centre pipeline holds. The comparison only goes so far, but think of it like a single anchor tenant carrying a whole retail scheme. When the covenant's strong, everyone's happy not to ask questions. The moment it wobbles, you wish you'd diversified. The people this lands on first aren't the hyperscalers, they're the commercial director who committed labour and the subbie who geared up for one client.

The procurement filter: If a single sector is now more than a fifth of your pipeline, treat it like client concentration risk, because that's what it is. Ask where the funding for each scheme actually sits before you commit long-lead resource to it.

The number doing the rounds this month comes from ServiceTitan's 2026 Commercial Specialty Contractor report, a survey of more than 1,000 construction leaders. It found 38% of contractors now report a measurable business impact from AI, up from 17% a year ago. The work it's actually showing up in is unglamorous and telling: cost estimation and budgeting at 24%, bid management at 22%. That's the shift worth noticing. This isn't AI writing site newsletters, it's AI turning up where the money is decided.

Two honest caveats before anyone puts this on a slide. First, ServiceTitan sells software to contractors, so it has an interest in the line going up, and its sample leans towards US commercial specialty trades rather than UK main contractors. Read it as a direction, not gospel. Second, "measurable business impact" is doing a lot of quiet work in that sentence, and every firm will measure it differently. What I'd take from it, though, holds up against what we're seeing elsewhere: the pilots are over, the gap between the firms getting something out of this and the ones still watching is real, and it's widening fastest on estimating and bidding. That's the same place a labour market short of roughly half a million people is squeezing hardest.

So tie the week together. The regime is finally rewarding good records, a central bank is telling you not to bet the whole book on one boom, and the evidence says your competitors are quietly getting faster at the bid. None of those is a reason to buy the shiniest tool in the room. All three are a reason to know where your own numbers are made and to get honest about how long they take. That's what it's about. Not the technology, the estimator who now turns round three bids in the time one used to take, and the PM who isn't the last to know a decision was changed.

A practical step: Time your next three tender returns end to end, today, before you change anything. If you can't say how long a bid takes you now, you'll never know whether any of this helped.