A beat the market opened 25% below
Start with what is not in dispute. Net sales rose 10.0% to $6.57B, 9.9% in constant currency, on the same 64 selling days as a year ago. Non-GAAP EPS of $2.91 was up 11.9% and $0.12 over consensus (unverified, consensus aggregator). The chief executive, Christine Leahy, opened the release with "strong second quarter results as customers advanced investments in infrastructure modernization, cloud, and AI-enabled technologies." Every one of those numbers checks against the filing.
The stock opened at $115.45 against a $154.00 prior close, down 25.03%, and finished the day down 9.03% at $140.10 (unverified, market data, not filing figures). I don't read that as the market doubting the orders. It read three lines the release put lower down: gross margin down 70 basis points to 20.1%, operating income up only 2.0% on sales up 10.0%, and six-month operating cash flow of $219.7M against $443.1M. The chief financial officer, Albert Miralles, spoke in the same release of "a disciplined approach to working capital and capital allocation." Standing alone, the second quarter's operating cash flow was negative. Discipline is the word for it only if the June balance sheet turns out to be a timing lump.
Every channel grew, one barely
As an end-market read, this is about as clean as the reseller layer gets, and it is broad. Corporate grew 10.7% to $2.62B. Healthcare grew 9.1%. Government grew 13.6% to $848.0M, which Leahy attributed to "improving federal demand" (unverified pending the official transcript). CDW UK and CDW Canada, reported together as Other, grew 22.9% to $825.7M, and with currency adding almost nothing that is real local demand. Financial Services grew 1.8%. Education grew 0.7%. Nothing shrank. US net sales rose 8.6%, so the international book added a little more than a point to the headline.
One read is missing, and it is the one that used to matter most. On CDW realigned its segments, and Small Business, the most cycle-sensitive customer channel it used to break out, stopped being reported; the 10-Q says its goodwill was "primarily assigned to the Commercial reporting unit." The quarter's most cyclical signal now lives inside a $3.97B Commercial line. I can't tell you what small business did in Q2. Nobody outside the company can.
The rest of this dive is for paid subscribers.
The headline numbers are above. The 8 sections below carry the mechanism, the peer read, the valuation work, and the dated tests that decide it.
- Read the ingredients, not the level
- Where the gross profit came from
- The composite hides the component
- Where the cash went
- The strongest case against me
- What decides it
- The trade (analysis, not advice)
- Bottom line
Already a subscriber? Sign in.
Read the ingredients, not the level
Hardware grew 10.4% to $4.91B, and the growth was concentrated. Data storage and servers rose 31.8% to $844.8M. Netcomm rose 12.1%, notebooks and mobile devices 11.1%, desktops 2.6%, collaboration 1.2%, and other hardware fell 3.0%. The client-device growth came from price: Leahy said "higher average selling prices more than offset lower unit volume" (unverified pending the official transcript). The 10-Q names the cause in its outlook section, "increased demand for memory-intensive products driven by the rapid adoption of artificial intelligence," which "may continue to influence supply chains and drive pricing pressures." Read the ingredients, not the level. A 10.0% headline made of fewer notebooks at higher prices and servers carrying inflated memory is a different business from one made of more seats. The call gave no unit count and no price effect, so the split between the two is the variable the print did not settle.
Where the gross profit came from
The 10-Q explains the margin in one sentence: "Gross profit margin decreased 70 basis points to 20.1%, primarily driven by mix into and lower margin in certain hardware categories, partially offset by a higher contribution of netted down revenue." Read the offset as the main event, because it is. CDW books some software and services as agent, recording only its margin as revenue. That netted-down revenue rose to $474.2M from $408.4M, up 16.1%, and since it is booked at the margin it is close to pure gross profit; on that basis it is 35.9% of the quarter's gross profit, the same share management gave on the call. Gross profit rose $78.6M. Netted-down revenue supplied $65.8M of it, 83.7%.
Everything CDW sold as principal, mostly hardware, rose to $6.10B from $5.57B, $529.8M more. It added $12.8M of gross profit, about 2.4 cents on each extra dollar. The gross margin on principal sales fell to 13.87% from 14.96%. Put the 2.4% next to the distributor that sits one layer up: TD SYNNEX earned a 6.61% gross margin on its whole book last quarter. On the incremental hardware dollar, the value-added reseller earned less than the wholesaler's average dollar.
The segments say the same thing from another angle. Government sales rose 13.6% and its gross profit fell 2.7%, the margin down 320 basis points to 19.3% on "mix into and lower margin in certain hardware categories and services." Education sales barely moved and its gross profit rose 7.9% on "a higher contribution of netted down revenue." The fastest-growing customer earned the least; the flattest earned more. Mix is the softest margin to capitalize, and here it cuts against the quarter: the growth came in the mix that dilutes. Services, the line that would earn the hardware back through deployment, rose 1.1%.
The composite hides the component
Management gave one composite: a three-month cash conversion cycle of 21 days, "within our target of high teens to low 20s" (unverified pending the official transcript). The composite hides the component. Over two years, receivable days rose to 93 from 72 and payable days to 88 from 68, and until this quarter the cycle barely moved, sitting between 15 and 18 days in every quarter but one. The 10-Q gives the reason for that, and it is a fair one: "Netted down revenue results in an increase to both DSO and DPO as the corresponding receivables and payables reflect the gross amounts due from customers and due to vendors while the corresponding sales and cost of sales are reflected on a net basis." The software business inflates both sides of the ledger, and they cancel.
What did not cancel this quarter is the hardware. Against Q1, receivable days fell 3, inventory days rose 2 to 16, the most in nine quarters, and payable days fell 6. Inventory is $981.8M, up $418.4M or 74.3% from December, which the filing puts down to "customer-driven stocking positions as a result of higher demand and increased hardware cost." In the same quarter CDW set up a product financing arrangement, $62M outstanding, in which an intermediary holds inventory for "a customer's hardware procurement program" over "an extended delivery period" beyond 90 days. A reseller that ships and bills does not need that. A reseller holding stock for a customer's rollout at rising component prices does.
Where the cash went
Over six months, receivables absorbed $1.05B and inventory $418.3M; suppliers funded $896.2M through payables. Take the $274.8M of Q1 operating cash flow out of the six-month $219.7M and the second quarter alone was negative $55.1M; after $27.5M of capital spending, free cash flow was negative $82.6M. Adjusted free cash flow for the half was $278.4M, 41.8% of non-GAAP net income, against management's rule of thumb of 80% to 90%, which the chief financial officer said this year will fall short of (unverified pending the official transcript).
Capital return did not slow down to match. CDW bought back $343.7M of stock in the second quarter and $544.7M in the half, 3.3x six-month free cash flow, and paid $160.9M in dividends. The gap came from $175.0M of net revolver draws and a $256.9M fall in cash to $361.8M. Then, in September, CDW sold $1.5B of senior notes at 5.700% to 6.350% to retire $1.0B of 2026 notes and fund most of the roughly $525M purchase of Lovelytics, a Databricks data and AI services partner. That deal is the right strategic answer to the margin problem, buying more of the part of the stack that earns. It is also being paid for with debt at a moment when the operating business is not producing much cash.
Set CDW beside TD SYNNEX and the balance sheets look nothing alike, but they are moving the same way. TD SYNNEX carries inventory about equal to its receivables, $15.29B against $14.95B, and burned $976M of free cash flow last quarter, 4.5% of revenue. CDW carries receivables, with inventory at 13.4% of current receivables and another $1.39B of noncurrent unbilled receivables from multi-year software deals, and burned $82.6M, 1.3% of net sales. Both cash cycles ended their quarters at 21 or 22 days, each up five or six days on the year (TD SYNNEX's figure unverified, from its call). The distributor funds the AI server build; the reseller funds the enterprise refresh at inflated component prices. On , the day TD SYNNEX reported, CDW fell 4.58% (unverified, market data). The market already trades them as one balance-sheet question.
The strongest case against me
The best argument against me is seasonal, and CDW's own history supports it. The third quarter holds the US federal fiscal year end, and last year the cycle dropped to 11 days in September on payable days of 93; the fourth quarter then brought $433.8M of operating cash flow as receivables were collected. The inventory is tied to customer programs, not speculative stock, and the product financing is sized to one rollout. Management's guide already concedes the margin, second-half gross margin below the second half of 2025 (unverified), so the market is not being surprised. And the netted-down growth the bridge credits is the business CDW has been trying to build for years; at 35.9% of gross profit it is the reason the reseller model still earns a 20.1% margin at all. On that reading June was a delivery lump in a strong hardware quarter, the 25.03% opening drop (unverified) overshot, and I'm confusing a lump with a new level.
I take it seriously, and I still come down the other way, for one reason. The test already allows for the season. I am not asking the September cycle to look like June's 21; I am asking whether it holds the same five-day gap to its own year-ago quarter. If the inventory is simply program stock awaiting delivery, the gap closes by September and I'm wrong. If the memory-driven price level keeps pushing hardware cost into both inventory and receivables, it doesn't, and "customer-driven stocking positions" becomes the operating level of a business growing hardware at 2.4 cents of gross profit on the incremental dollar.
What decides it
One number, on one date: the three-month cash conversion cycle for the quarter ending , published with Q3 results in early November. At 16 days or more, against 11 a year earlier, the reseller is now carrying hardware working capital as an operating cost, and the 41.8% cash conversion is a rate rather than a dip. At 12 days or fewer, the June build converted, the seasonal steelman was right and I'm wrong. Anything in between is a partial result, and I'd call it that. Two secondary checks: whether netted-down revenue keeps outgrowing principal sales, since that is where the gross profit comes from, and whether nine-month adjusted free cash flow gets back to the $668.0M of last year's first nine months.
The trade (analysis, not advice)
At $135.43 on (unverified, market data) the stock is below its $140.10 close on print day and 12.06% under the $154.00 it held before the release. The margin is priced; management guided it and the tape took it. I don't think the cash is priced yet, because the buyback runs at about $343.7M a quarter and the half's free cash flow was $165.8M. The asymmetric setup is to wait for the September cycle. A number at or below 12 would re-rate CDW on cash quality with the demand already confirmed. A number at 16 or above would mean the buyback is increasingly a debt decision, and the next leg of the de-rating is still ahead. Anyone who wants the server and storage demand, 31.8% here, without the thin incremental margin can own it one layer up, from the component and system vendors, through picks-and-shovels capture. Anyone buying CDW is buying the netted-down software franchise with a hardware balance sheet attached.
Bottom line
CDW's second quarter confirms that enterprise, government and international IT demand is growing, in every customer channel, if barely in education, and it shows what that demand is worth to the reseller. The hardware half of it arrived at higher prices and fewer units, and it earned 2.4 cents of gross profit on each extra dollar; the netted-down software half earned the quarter. My view is that the working capital behind the hardware is the new price of the mix, not a June lump. The number that settles it is the cash conversion cycle at 16 days or more, or 12 or fewer, on .