Selling More, Earning Less

Computacenter PLC — a large-scale technology reseller with an attached IT services arm, selling to large corporations and the public sector. All figures in pounds sterling.

Computacenter is a technology and services provider, marketing primarily to large corporations and the public sector. At heart it is a large-scale technology reseller — it buys hardware and software from manufacturers and sells it on to big organisations — with an attached IT services arm (professional services and managed services) bolted on top. The reselling generates enormous volume at thin margins; the services side is smaller but more profitable.

Revenues have oscillated a bit over recent times but, in general, have not really been heading downwards. FY 2021 saw a negative period of −7.5%, which we can largely forgive given the lockdown impacts of the COVID pandemic. Things bounced back dramatically in FY 2022, with revenue growth of 28.5% to £6.471bn. FY 2023 saw sales rise 7.0% to £6.923bn, FY 2024 was nearly flat at +0.6% to £6.965bn, and then FY 2025 saw a sudden jump again — revenue up 32% to £9.194bn. Interestingly, I saw something similar recently with Softcat, which operates in a very similar space.

Their gross margins are reasonably poor — which is what I would anticipate in a sector like this — but they were especially low this year, hitting 12.4% in FY 2025, the lowest in over five years. Generally they have oscillated around 14–15%. Profitability has remained reasonably flat year on year, although FY 2025 did see post-tax profit reduce from £172m to £157m. That meant earnings per share went down for the second year in a row.

Six-year financial history

Dividends rose again last year despite profits slipping — baffling on the face of it, but in line with what a lot of companies seem to prioritise currently (dividend per share up to 74.6p). Even so, the yield is low at 1.8%, and given slipping profitability I would wonder whether they can really offer anything greater. Borrowing, though, is at a good level: net funds (negative net borrowing) of £426m positive. So the balance sheet is genuinely strong — that's the clear positive in this story.

The business splits into two reporting segments. The FY 2025 picture shows exactly where the volume — and the margin pressure — sits:

Segment revenue (FY 2025)

The revenue surge is almost entirely Technology Sourcing — high-volume, low-margin hardware reselling — while the higher-margin Services arm grew only 3.2%. A shift in mix toward the low-margin segment is why group gross margin fell to 12.4% even as revenue jumped by a third. They are selling far more, but of the least profitable thing.

Sources: Computacenter FY2025 Final Results, 12 Mar 2026 (investors.computacenter.com); Sharecast; Computer Weekly/Microscope. "cc" = constant currency. Group invoiced income (GII — the total value of goods invoiced, including items where the company only books the commission) grew 31% to £13bn.

Computacenter operates across several countries. The half-year picture shows North America as the clear engine:

Revenue by region (H1 2025, six months to 30 June)

This is an AI-infrastructure story, and specifically a North American one. The 32% surge was driven overwhelmingly by North America Technology Sourcing, where the company grew networking and data-centre volumes with both enterprise and hyperscale (cloud) customers on the back of buoyant demand for AI-related infrastructure. The UK contribution also leaned on AI data-centre projects. It is not primarily a government/public-sector upgrade cycle — if anything public-sector demand in Germany and France was soft for much of the year. It is the AI build-out, sold as hardware.

Two corroborating signals. First, the product order backlog ended 2025 at a record £7.1bn, up 200.3% year on year in constant currency, driven by Technology Sourcing order intake in North America and the UK — so the AI-hardware demand is sitting in the pipeline for 2026 too. Second, the company even ran a temporary North American facility holding £137.7m of inventory at year-end to support a single major data-centre programme.

Source: Computacenter FY2025 results and CEO commentary (Mike Norris); joshthompson.co.uk analysis.

I don't think I'm going to investigate this company further. Their margins have always been a bit weak, but they seem especially low right now.

Since 11 February they've seen a 35% rise in share value, which I read as a market reaction to expectations of companies dramatically increasing infrastructure spend on AI requirements. But in any case, I don't really like where the company is heading. There is a lot not to like about business-to-business AI infrastructure: it is a squeezed area, and while this company can clearly achieve high total sales and do so profitably, its ability to make profit from those sales is slowly decreasing. I don't really see where this company — like other software and computer-service companies — is heading.

I'm being a little generous here. I'm using the last reported earnings per share of 145.5p, a valuation multiple of 8x, and a growth rate of 2%. That is kinder than the company's recent record arguably deserves — earnings per share shrank 4.8% last year, and the two-year average is around −8.3% — but I suspect they will be able to achieve a 2% long-term growth rate, helped by the dramatic revenue rise. I then apply a 10% valuation reduction, because I don't like this whole infrastructure-reselling section: it's a squeezed area, and profitability is slowly eroding. I set a relatively low 15% risk factor, and — because I haven't researched the share heavily — a probability of 26%.

VALUATION SNAPSHOT

This gives an actual valuation of 1,215.22p against a current share price of 4,248.00p — a value opportunity of −71.4%. On these conservative inputs the shares look dramatically overvalued.

I should stress the obvious counter: the market is paying a trailing price-to-earnings ratio of around 24x (on adjusted earnings) precisely because it is capitalising the AI-infrastructure growth story, whereas my valuation deliberately refuses to extrapolate a low-margin hardware surge into durable earnings. If the AI-hardware demand proves both large and lasting — the £7.1bn backlog is the bull's exhibit A — my 2% growth assumption will look far too cautious. But on the quality of the earnings as they stand — thinning margins, falling reported earnings per share, growth concentrated in the least profitable line — I am comfortable being cautious.

• The AI-hardware demand is large and lasting: The £7.1bn record backlog (+200% cc) is real and points to a strong 2026. If this is a multi-year build-out rather than a one-off, my 2% growth assumption is far too cautious — the single biggest risk to a Sell.
• North America keeps compounding: US profit nearly doubled and now drives ~39% of group operating profit. Continued momentum there could carry earnings well past my estimate.
• Strong balance sheet: £426m of net funds gives real resilience and optionality (buybacks, acquisitions, weathering a downturn).
• Adjusted earnings tell a better story: On the adjusted measure, earnings per share rose 9.5% to 175.1p — the underlying trading is healthier than the reported figure I've used.
• Mix could re-balance: If the higher-margin Services arm re-accelerates, group margins could recover from the 12.4% trough.