Verdant’s tally: UK datacentres will deliver far fewer permanent roles than industry claims
Verdant, an environmental thinktank, calculates that the datacentres currently planned for the UK will directly create about 10, 400 permanent jobs, roughly one-quarter of the headline figure industry lobby TechUK has promoted. TechUK says the sector “supports 43, 500 jobs, and could support 40, 000 more by 2035, ” while the government rejects Verdant’s jobs per megawatt approach as misleading.
Methods at a glance
- Verdant: based its estimate on publicly available staffing data for a sample of 20 projects (planning applications, company accounts, press releases and international comparisons). Verdant focuses on direct, permanent operational staff and divides those headcounts by the sites’ nameplate electrical capacity (megawatts, MW). It explicitly excludes temporary construction roles.
- TechUK: defends a much larger employment total and says its estimates are “built on a published methodology using industry data and standard economic impact assessment techniques.” That approach typically includes indirect and induced jobs, supply‑chain effects and construction employment as part of the sector’s broader economic contribution.
- Government: has criticised Verdant’s use of jobs per MW and warns that electrical capacity (MW) is not the same as actual energy consumed over time (megawatt‑hours, MWh), a distinction that changes the denominator in any jobs intensity calculation.
Why the headlines diverge
The dispute is about method, not just arithmetic. Verdant counts ongoing, direct operational roles and links them to a site’s nameplate capacity in MW. TechUK applies a broader lens and includes construction peaks, supply‑chain jobs and local services supported by datacentre activity using standard economic multipliers.
Both sides can be correct within their own definitions. The problem comes when those definitions get mixed in a single headline. A “jobs” number that folds in construction and indirect effects answers a different policy question from a count of permanent on‑site staff months or years after commissioning.
MW vs MWh, the single most important technical clarification
MW measures the maximum power capacity a site is built to handle at an instant (nameplate power). MWh measures energy consumed over time (power × hours). A jobs per MW ratio divides headcount by maximum capacity and implicitly assumes high, sustained utilisation. Jobs per MWh divides headcount by actual annual energy use and often better reflects operating intensity.
Which you use matters. If a 100 MW facility typically runs at 20 percent average load, its MWh consumption is one fifth of a naive MW‑based expectation. Jobs per unit of actual energy used would therefore look five times higher than jobs per MW suggests.
The numbers Verdant presents (and what they mean)
- Verdant’s headline: ~10, 400 permanent jobs tied to currently planned UK datacentres (based on its 20‑project sample and public staffing data).
- Verdant’s snapshot of existing sites: fewer than 4, 400 permanent roles currently employed in UK datacentres.
- Verdant’s jobs per MW: an average of about 8.6 jobs/MW for existing sites; by contrast, TechUK’s figures as cited give ~43.7 jobs/MW (reflecting a different counting method).
- For some proposed hyperscale projects, Verdant warns jobs intensity could fall further, possibly as low as 1.6 permanent roles/MW, driven by automation, remote operations and standardised designs.
Verdant contrasts datacentre jobs per MW with other intensive services to show the scale difference, but those comparisons only work if the same denominator and scope are used consistently. A hospital’s staffing per MW of installed power will look enormous compared with a datacentre, but hospitals and datacentres operate differently and have different load profiles. Treat cross‑sector jobs per MW comparisons with caution.
“We’re being rushed into supporting a frantic pace of datacentre buildout on very flimsy economic grounds. Clearly artificial intelligence and related technologies have great potential. But vast new datacentres arrive at significant real costs to local communities and even our climate ambitions, and their use case has to be made by the government and industry, rather than assuming we can build first and then work out why we did later.”
– James Meadway, co‑director, Verdant
Resource and political stakes
Verdant flags energy and water intensity and local community impacts. Opponents worry about water allocation in drought‑prone catchments and about developers deploying on‑site fossil fuel back‑up (or even primary) generation if grid connections lag. Campaign groups such as Foxglove have specifically warned developers could use on‑site gas turbines to bridge grid delays, locking in emissions.
Politics is sharply divided. Labour has designated AI datacentres as “critical national infrastructure” to speed planning and emphasise sovereign compute capacity; Kanishka Narayan has framed domestic datacentre capacity as a matter of national defence and sovereignty (quote reported to The Times and cited in coverage on 28 August 2026). The government has warned that pausing construction risks moving investment, and related jobs, overseas and leaving sensitive data processing dependent on foreign facilities (government comments reported in The Guardian, 28 August 2026).
“This report relies on a deeply misleading jobs‑per‑megawatt metric, makes false comparisons with hospitals and schools, and appears to confuse electricity capacity with actual consumption. More seriously, it discounts the wider economic value of compute [computing capacity] and the national security and sovereignty case for building critical digital infrastructure in Britain.”
– Government spokesperson (as reported in The Guardian, 28 August 2026)
The Green party and local campaigners have called for moratoria and staged protests in places where community consent is weak; Green leader Zack Polanski has warned of water pressure after a record hot summer (comments reported in The Guardian, 28 August 2026).
What leaders should demand before accepting any headline jobs claim
Headline numbers should be dissected. Here’s a practical checklist for C‑suite executives, local planners and policymakers who must weigh a datacentre proposal.
- Ask developers for a breakdown of employment by category and timeframe: direct operational FTEs (by role and shift), contractor staff, and peak construction FTE‑months.
- Require clarity on the denominator: are jobs being reported per MW (nameplate capacity) or per MWh (expected annual consumption)? Ask for average utilisation assumptions and the basis for them.
- Insist on local resource modelling: projected grid load as a percent of the local transformer/substation capacity, planned dates for grid reinforcement, and water use in m3/year plus the percent of local non‑household or catchment supply.
- Get backup‑generation plans in writing: type of generation, fuel source, emissions profile and triggers for use. If gas turbines are proposed, require a timeline and commitment for grid connection and fuel switching or retirement.
- Request a transparent economic‑impact statement: separate direct permanent roles from construction, supply‑chain, and induced jobs and show the multipliers and assumptions used to calculate “supported” jobs.
Reconciling the argument, a simple mental model
Use three clearly labelled figures when someone quotes “jobs from datacentres”: (1) direct permanent operational FTEs, (2) temporary construction FTEs (or FTE‑months), and (3) indirect/induced jobs (supply chain, local services). Match each to the denominator you prefer, per site, per MW, or per MWh, and you’ll stop comparing apples to oranges.
If your policy goal is local employment, emphasise direct, permanent jobs and construction‑period opportunities. If your goal is national digital sovereignty or defence resilience, the right metric is available compute capacity and the geography of that capacity; jobs then become a secondary metric.
Key questions, and short, actionable answers
- How many permanent jobs will planned datacentres actually create?
Verdant estimates about 10, 400 permanent jobs across currently planned UK datacentres based on public staffing data for 20 projects; TechUK produces a larger figure (cited as roughly 43, 500 supported jobs) using a broader economic‑impact methodology. Treat each number as answering a different question, direct on‑site jobs versus total economic footprint.
- Why do Verdant and TechUK give such different numbers?
They use different scopes and denominators. Verdant focuses on direct, ongoing operational staff and relates headcount to nameplate MW capacity; TechUK’s published methodology (industry standard impact assessment techniques, per TechUK) includes construction, supply‑chain and induced jobs and uses economic multipliers.
- Is jobs‑per‑megawatt a reliable metric?
It can be useful if everyone agrees on the numerator (headcount as FTEs or raw roles) and the denominator (MW vs MWh). Alone, jobs per MW is misleading when capacity is rarely run at peak; prefer jobs per MWh or direct FTEs per site for operational intensity.
- Should the UK pause datacentre approvals because of Verdant’s findings?
That’s a political judgement. Verdant raises valid questions about local impacts and long‑term job intensity; the government and industry argue pausing risks losing investment and sovereign compute. A practical middle path is to tighten transparency and local impact requirements rather than an outright moratorium.
- What non‑job impacts should be weighted most heavily?
Energy and grid implications, water use in stressed catchments, the risk of on‑site fossil backup generation, and cumulative local impacts (traffic, night shifts, security). Require project‑level modelling for these before consenting large new sites.
Verdant’s analysis forces a re‑examination of headline job claims. The sensible policy response is not to pick one side without scrutiny, but to demand transparent methodologies, insist on local resource impact assessments, and separate national sovereignty arguments from local employment promises so each can be evaluated on its merits.