This is investment research, not personal financial advice.

Computacenter (LSE:CCC) closed 7.77% lower at 5,165p on 8 September after reporting an 87% rise in first-half adjusted profit before tax and lifting its full-year floor to £380 million. The result looked like an upgrade. The shares traded as if it were a warning.

That reaction becomes less strange once revenue is separated from value retained. First-half revenue rose 68.4% to £6.85 billion, yet gross profit increased only 8.7% to £657.9 million. Operating cash flow was negative £82.6 million. The order backlog reached £9.3 billion, more than four times the £2.2 billion reported at December, but much of the increase came from low-margin North American data-centre hardware (Computacenter H1 2026; Computacenter 2025).

The market's direction was rational. The full 7.8% fall looks more like a reset from a record price than a judgement that demand has broken. At £51.65, the shares still capitalise the company at £5.49 billion and value a plausible £290 million of normal owner cash at about 19 times. That is neither a distressed valuation nor an impossible one. It leaves one question: can a fourfold backlog recover cash and gross profit quickly enough to deserve that multiple?

The market sold the mix, not the profit

The headline numbers were unusually strong. Adjusted profit before tax rose from £95.1 million to £177.9 million, adjusted operating profit rose 86%, and management said FY2026 adjusted profit before tax would be at least £380 million. North America supplied most of the acceleration. Revenue there rose from £299.4 million to £1.70 billion, and regional gross profit advanced from £18.2 million to £125.3 million (Computacenter H1 2026).

Yet the stock opened at 5,875p, touched a record 6,045p and finished at 5,165p. Volume of 556,334 shares was about 2.2 times the ten-day average shown by Google Finance. The session therefore contained its own repricing. Early trading rewarded the raised profit floor; later trading asked what those profits would look like after the first wave of infrastructure orders.

An independent market report captured the conflict: higher guidance and a £9.3 billion order book sat beside a first-half profit result that was slightly below one analyst's timing assumptions (Proactive Investors 2026). The reaction was not about a missing order. It was about how much economic value sits inside each order, when the cash arrives, and whether the current surge is repeatable after data-centre customers finish their build phase.

There was an expectations problem too. Even after the fall, Google Finance showed the shares up 51.7% over twelve months. Computacenter entered the result with a record valuation and a market already aware that profit would be materially ahead of the previous year. A company can raise the floor and still disappoint when its price has moved faster than the evidence about cash quality.

£2.8 billion of extra sales produced £53 million of extra gross profit

Start with the two reported revenue lines. H1 revenue increased by £2.78 billion, from £4.06 billion to £6.85 billion. Gross profit increased by £52.6 million. Each additional pound of reported revenue therefore carried less than two pence of incremental gross profit.

That does not mean the new work is loss-making. It means the mix changed violently. Group gross margin fell from 14.9% to 9.6%, a 528-basis-point contraction. Technology Sourcing, where Computacenter procures and integrates third-party hardware and software, grew far faster than Services. Large North American deployments added enormous invoice values, but the cost of equipment passed almost straight through the income statement (Computacenter H1 2026).

North America also shows why a simple group-margin verdict is incomplete. Its gross margin improved from 6.1% to 7.4% even while the region pulled the group average down. The company is earning a better spread on a much larger North American base. The problem is mix at group level: a region with a 7.4% margin has become much more important than established European operations with a higher service component.

The backlog carries the same ambiguity. Management's £9.3 billion figure is a forward gross-income measure, not contracted gross profit or future operating cash. It includes orders that may be delivered over different periods, with different hardware content, customer payment terms and service attachments. The H1 presentation shows orders by route to market, but it does not disclose the gross profit, working-capital requirement or cancellation terms embedded in that £9.3 billion (Computacenter H1 deck 2026).

The International Energy Agency gives the demand case substance. Global data-centre electricity use has grown about 12% a year since 2017, and its base case has consumption more than doubling by 2030. It also estimates that grid constraints could delay about 20% of planned data-centre projects (IEA 2025). Computacenter is exposed to both facts. More data-centre construction creates hardware demand; power and grid delays can shift delivery dates after equipment has entered the order book.

A reseller can grow faster than the value it retains

Computacenter has three connected activities. Technology Sourcing supplies hardware, software and commercial structures. Professional Services designs and installs systems. Managed Services operates customer infrastructure under longer contracts. The sourcing order often opens the door, while integration and management provide the stickier relationship and better gross-profit density.

This combination matters more than revenue. Computacenter describes itself as one of the world's six largest value-added resellers and says it works with more than 1,800 partners across more than 70 countries. A multinational customer can use one counterparty for procurement, configuration, rollout and support. Vendors gain a route into large accounts. Computacenter gains purchasing scale and the chance to attach service work (Computacenter 2025).

The model is capital-light in fixed assets but heavy in trust and working capital. A large customer order can require Computacenter to pay a vendor before collecting from the customer. Inventory can rise while equipment is configured or held for a deployment window. Receivables expand with gross invoiced income. The balance sheet therefore acts as part of the product.

Revenue accounting adds another complication. The 2022 annual report restated 2021 revenue from £6.73 billion to £5.03 billion after determining that the group acted as agent, rather than principal, for some standalone software and third-party service contracts. Only the net amount is now reported as revenue for those transactions. Gross invoiced income remains an alternative measure because it better tracks invoice and working-capital movements (Computacenter 2022).

The opposite accounting judgement applies to many hardware transactions, where Computacenter records the gross sales value and cost. Two contracts with similar gross profit can therefore create very different revenue. That is why the fall in gross margin is not, by itself, evidence of weaker pricing. It is evidence that the revenue denominator has become less informative.

There is still an accounting risk to monitor. The FY2025 auditor identified revenue recognition on unshipped customer-designated sales as a key audit matter. Large equipment orders can qualify for revenue before physical delivery only when strict control conditions are met. The auditor also focused on the value of non-current assets in France, where weaker performance has already produced impairment charges (Computacenter 2025). Neither matter invalidates the accounts. Both show where judgement is concentrated as volumes rise.

Cash has already separated the good years from the large ones

The history below uses filed figures in pounds. Capital expenditure is cash spent on property, equipment and intangible assets. Owner cash is an author calculation of operating cash flow less that capital expenditure. The ROIC series is also author-computed: adjusted operating profit after the reported effective tax burden divided by average invested capital, where invested capital is equity plus borrowings and lease liabilities less cash and current investments. It is approximate because acquisition adjustments, pensions and surplus cash can change the denominator.

Period Revenue (£m) NPAT (£m) OCF (£m) Capex (£m) Owner cash (£m, computed) ROIC (computed) Adjusted net funds (£m)
FY2021, restated 5,034.5 186.5 220.4 28.9 191.5 23.4% 120.1
FY2022 6,470.5 182.8 342.1 50.8 291.3 30.2% 244.3
FY2023 6,892.5 200.8 410.6 42.6 368.0 29.0% 459.0
FY2024 6,964.8 170.8 341.4 72.0 269.4 28.1% 459.1
FY2025 9,193.9 153.7 239.2 56.0 183.2 24.7% 489.1
H1 2026 6,845.2 109.0 -82.6 11.8 -94.4 not meaningful 83.2

Sources: filed annual and interim accounts (Computacenter 2021; Computacenter 2022; Computacenter 2023; Computacenter 2024; Computacenter 2025; Computacenter H1 2026). The frontmatter records adjusted net funds as negative net debt. H1 is a six-month period and is not comparable with a full year.

The business produced £1.30 billion of computed owner cash over FY2021-FY2025, an average £260.7 million and median £269.4 million. That is evidence of a cash-generative model. It is also a warning against using revenue growth as the main compounding measure. Revenue rose 83% from the restated FY2021 base to FY2025, but adjusted operating profit rose only 1.2%, from £255.6 million to £258.7 million. Computed owner cash in FY2025 was half its FY2023 peak.

H1 2026 widened that gap. Operating cash flow of negative £82.6 million less £11.8 million of capital expenditure produced negative £94.4 million of computed owner cash. Management attributes the movement to working capital tied to the record Technology Sourcing book and expects a substantial second-half reversal. Seasonality supports that possibility. Computacenter commonly carries more working capital at June than at December.

But a seasonal explanation is a forecast, not cash. The adjusted net-funds position fell from £489.1 million at December to £83.2 million at June. The company also drew a short-term financing facility to support the order surge. Liquidity remained ample, with committed facilities, and the group still reported net funds rather than net debt. Survivability is not the issue. The issue is whether the extra balance-sheet use earns more gross profit or merely finances customer deployment schedules.

North America fixed one weakness and created a new dependency

Computacenter spent years building North America through acquisitions. FusionStorm, Pivot and Business IT Source added customer relationships, staff and integration capacity. The H1 result suggests that strategy has finally reached a different scale: £1.70 billion of regional revenue in six months, nearly six times the prior period, with gross profit up almost sevenfold (Computacenter H1 2026).

The margin improved as well. North American gross margin rose from 6.1% to 7.4%. That is evidence that purchasing terms and execution can improve with volume. It supports the widening-moat case in the region, especially when hyperscale and newer cloud customers need large batches of equipment across sites.

Concentration works both ways. One large customer or deployment schedule can now change group revenue, inventory and receivables. Computacenter does not disclose enough customer-level data to measure the backlog's concentration. It also does not quantify how much of the North American hardware will lead to Professional or Managed Services. Without that bridge, a larger backlog may improve vendor relevance without improving the durability of earnings.

Softcat offers a useful peer check, with limits. Softcat's H1 FY2026 gross invoiced income rose 33.3%, gross profit rose 22.6%, underlying operating profit rose 27.3%, and underlying cash conversion reached 147.6% (Softcat H1 2026). Its UK-focused, reseller-heavy model and reporting period differ from Computacenter's multinational delivery mix. The comparison does not establish that one margin is superior. It does show that large technology demand can translate into gross profit and cash at very different rates.

Computacenter's stronger defence is its international reach. A customer deploying across the United States, Germany and the UK can use the same group operating model, vendor relationships and integration process. That creates switching friction and improves the chance of service attachment. The counter-evidence is France, where weaker performance and asset impairment show that a global footprint does not guarantee local returns (Computacenter 2024; Computacenter 2025).

Capital allocation now includes financing the customer

Management has used cash across acquisitions, dividends, internal systems and working capital. The acquisition record created the North American platform now driving growth, but it also introduced goodwill, customer-relationship intangibles and integration risk. The FY2025 accounts record £14.9 million of restructuring and one-off charges connected with French operations. The H1 2026 result includes further acquisition-related costs and amortisation adjustments between statutory and adjusted profit (Computacenter 2025; Computacenter H1 2026).

Those adjustments matter at today's valuation. Statutory H1 profit before tax was £156.4 million, below adjusted profit before tax of £177.9 million. The gap was £21.5 million before tax. A valuation built on adjusted earnings assumes that acquisition amortisation and other excluded costs do not recur as an economic burden. That is easier to defend when acquisitions are infrequent. It is less comfortable when bought assets are a recurring part of geographic expansion.

The dividend is covered by the historical cash record, but the first call on cash in 2026 is working capital. This is a change in capital allocation even if no board resolution describes it that way. Hundreds of millions of pounds moved from December net funds into inventories, receivables and customer deployments by June. The return on that temporary investment will be visible in the second-half cash unwind and in gross profit from the same orders.

Founder and chief executive Mike Norris has led the business for decades, a continuity advantage in vendor and customer relationships. It also concentrates strategic judgement. The current North American surge tests whether long management tenure produces discipline when demand is unusually strong. Volume is easy to accept; profitable volume with enforceable payment terms is the harder decision.

£51.65 assumes cash conversion returns before margin does

Computacenter's £5.49 billion market value equals £51.65 multiplied by 106.24 million shares. At the H1 effective tax rate of 31.9%, management's £380 million adjusted profit-before-tax floor converts to about £259 million after tax, or £2.44 a share. The current price is 21.2 times that simplified after-tax floor. This is an author calculation, not company guidance for earnings per share.

Owner cash gives a second frame. The FY2021-FY2025 median is £269.4 million. A normalised £290 million, close to the five-year centre after allowing for some growth, is £2.73 a share. At £51.65, the stock trades at 18.9 times that amount. The price therefore assumes that second-half working capital reverses and that normal cash generation lands above FY2025's £183.2 million.

The sensitivity below changes only normal owner cash and the equity multiple. It does not add the £83.2 million of June net funds separately because owner cash is being valued as an equity cash flow and the cash balance is needed to support a larger order book.

Normal owner cash (£m) 17x 19x 21x
250 £40.0 £44.7 £49.4
290 £46.4 £51.9 £57.3
330 £52.8 £59.0 £65.2

The post-result price sits almost exactly on the middle cell. Reverse the calculation and £51.65 implies £288.8 million of owner cash at 19 times, £322.8 million at 17 times, or £261.3 million at 21 times. The market is not pricing a collapse. It is choosing between a lower cash number with a premium multiple and a stronger cash recovery with a less demanding multiple.

A conventional enterprise-value multiple tells a similar story but is less clean. June net funds reduce enterprise value, while lease liabilities and working-capital facilities move it back up. Adjusted operating profit also excludes acquisition and restructuring charges. The owner-cash frame keeps the decisive uncertainty visible instead of burying it in an adjustment stack.

From £19.50 to £80, the gap is cash rather than orders

The severe-downside range of £19.50-£29.50 assumes normal owner cash of £160 million-£210 million and a 13-15 times multiple. In that path, data-centre demand fades, grid or customer delays slow backlog conversion, and working capital remains above its historical level. The balance sheet can survive it, but the valuation loses both cash and quality.

The bear range of £31-£41.50 uses £220 million-£260 million of owner cash and 15-17 times. Adjusted profit reaches roughly the new floor, yet hardware remains the main growth source and the cash released in H2 does not restore the five-year norm. This range treats the H1 margin mix as persistent.

The base range of £43-£58.50 uses £270 million-£310 million of owner cash and 17-20 times. Backlog converts on schedule, a large part of the first-half cash absorption unwinds, and Services growth offsets some hardware dilution. The current £51.65 sits inside this range. That is why the 7.8% reaction looks proportionate rather than extreme.

The bull range of £62-£80 uses £330 million-£370 million and 20-23 times. North American scale improves vendor economics, large deployments produce service attachments, and the group turns its international network into recurring gross profit. This path needs more than the £380 million profit floor. It needs evidence that the backlog is a customer-acquisition engine rather than a hardware queue.

These are author scenarios, not forecasts from Computacenter. The valuation is most sensitive to two assumptions that H1 does not resolve: sustainable owner cash after working capital normalises, and the multiple justified by the resulting mix. A one-turn improvement in either can move value by roughly £5-£8 a share across the central matrix.

The anti-thesis is that low-margin hardware buys the next service contract

The cautious reading can be wrong. Technology Sourcing is not merely pass-through turnover. Hardware reaches the customer before engineers can integrate it and before support teams can manage it. A wave of data-centre equipment can therefore depress current gross margin while creating future Professional and Managed Services revenue.

North America's margin improvement gives this argument some evidence. Gross profit grew faster than revenue there, even from a small comparison base, and management says the new customers broaden the addressable market. Vendor rebates and purchasing terms may improve as volumes rise. If the £9.3 billion backlog brings service work, today's diluted group margin understates the economic value of the customer relationship.

The IEA demand outlook also argues against treating the order surge as a single-quarter accident. Data-centre power needs are rising much faster than total electricity use, and the largest planned facilities consume the equivalent power of entire cities. That supports years of infrastructure procurement, even if project timing is uneven (IEA 2025).

The anti-thesis fails if service gross profit does not follow. Computacenter's FY2025 adjusted operating profit was only 1.2% above FY2021 despite much higher revenue. That history says scale alone has not yet compounded operating profit. The burden of proof is therefore not another record backlog. It is a higher amount of cash and gross profit per customer after deployment.

March will test whether the backlog is an annuity or a queue

The first crux is mechanical. FY2026 operating cash flow less capital expenditure needs to recover above £250 million for the second half to resemble an unwind. A lower figure would leave cash below the five-year median and show that the larger order book requires a permanently larger balance sheet.

The second is mix. A full-year group gross margin below 10.5% would indicate that hardware remains dominant even after the busiest deployment months. North American gross margin below 7.0% would weaken the purchasing-scale argument. The FY2026 result, expected around March 2027, will report both the year-end balance sheet and the second-half income statement.

The third is service attachment. Services gross profit growth below mid-single digits at FY2026 or H1 2027 would show that infrastructure orders are not yet feeding the higher-retention part of the model. Management's backlog disclosures would also be more useful if they separated hardware from service value, duration and customer concentration. None of those details was available in the H1 packet.

Through FY2027, the evidence should arrive in that order: cash at the full-year result, margin as the delivery mix matures, then Services growth after installations. The share-price reaction has already moved the debate away from demand. At £51.65, the market is pricing a backlog that converts and a cash cycle that repairs, but not yet the full service annuity imagined in the bull case.

Source notes: partial verification and missing facts

The triggering H1 release, presentation and five annual reports were retrieved and read. Computacenter's current investor-document host blocked direct automated access to some static PDFs, so archived copies of the same official documents were used for extraction while the source ledger retains the official URLs. The Companies House record confirms the legal name as COMPUTACENTER PLC (Companies House 2026).

The point-in-time Finance API health and authentication checks passed, but its exact LSE resolver returned no match for CCC. Identity was therefore reconciled through Companies House and the official Computacenter filings. Market price, previous close, volume, shares and capitalisation were checked against the dated Google Finance page. That unresolved API identity and the archived-PDF retrieval route are why verification is marked partial rather than full.

ROIC, owner cash, reverse valuation and scenario ranges are author calculations. The annual report sources directly support their inputs, not the computed outputs. The backlog lacks disclosed gross profit, payment terms, duration and concentration. Those omissions prevent a precise discounted cash-flow model and remain the largest evidence gap.

References