CDW is the dominant North American IT VAR with sticky relationships across SMB, enterprise, education and government. The scorecard's $234-$516 IV range looks generous for a low-margin reseller; the real question is whether 7-8% reinvestment returns and -1.7% implied growth deserve more than a single-digit multiple of operating income.

Plain English

CDW is the middleman that helps companies, schools and government offices buy computers, networking gear, software and cloud services from hundreds of brands like Cisco, Dell, Microsoft and Apple. Instead of every customer calling fifty vendors, they call CDW, which handles paperwork, financing and setup. CDW makes a thin profit on each sale but does enormous volume. It earns money for shareholders by being big enough to get vendor discounts, sticky enough to keep customers, and disciplined enough to buy back shares.

Thesis

CDW Corp is the largest US-based IT value-added reseller (VAR), aggregating roughly 100,000+ hardware, software and cloud SKUs from vendors like Cisco, Dell, HPE, Microsoft, Apple, Lenovo and AWS, and selling them to ~250,000 small business, mid-market, enterprise, education, healthcare and federal/state/local government customers. Its job is procurement glue: configuration, financing, logistics, integration and a small but growing services layer (managed, professional, cloud advisory via Sirius and Mission Cloud).

Why it might compound: CDW posts a 10-year average ROIC of 15.4%, has shrunk shares ~2.25% over a decade, converts >115% of earnings to free cash flow, and sells at a P/E of 17.07 versus its own 10-year average of 28.82. The scorecard composite is 79/100 and the price/IV ratio is 0.32, with a base IV of $426.70 against $136.03.

Why I am skeptical of that math: ROIIC over 5 years is only 8.21%, the implied reverse-DCF growth is -1.74%, and net debt/EBITDA sits at 2.6x with no interest coverage figure available. The scorer flagged maintenance capex uncertainty twice. A reseller with a 4-5% operating margin, mid-single-digit organic growth, and structural exposure to AI workloads bypassing traditional channel reselling does not deserve a 28x average multiple.

Price/IV math: at $136 vs base IV $426.70, the implied margin of safety is 68%. But if true owner earnings durability is half of what the scorer assumes (a reasonable haircut for cyclical hardware mix and channel disintermediation risk), the realistic IV is closer to $200-$240, leaving a more modest ~30-50% margin of safety. That still rates a Buy, but with medium conviction and a small initial position.

Moat

CDW's moat is real but narrow. I find evidence in three of the five categories.

1. Cost advantages from scale (NARROW). CDW does roughly $21B in annual revenue and is the largest pure-play IT VAR in North America. Per Damodaran [3], scale-driven cost advantages — like Home Depot's vs local hardware stores — are durable when scale produces lower input costs and distribution leverage. CDW's scale gives it the deepest vendor rebate ladders (back-end incentives from Cisco/HPE/Microsoft tied to volume and certifications), the lowest cost-to-serve per transaction, and the broadest in-stock SKU breadth. Smaller VARs (SHI, Insight, Connection, ePlus, Computacenter) operate at meaningful but smaller scale; only Synnex/TD-SYNNEX rivals it, and that is a distributor (one tier upstream), not a reseller. Stress test: a $10B competitor with five years could not replicate CDW's 7,000+ field/inside sales reps with vendor co-funded compensation. But this is not a Coca-Cola moat — it is more like McLane [4][5], a thin-margin distribution machine that earns 'one cent per dollar' on huge volume. Verdict: real but capped.

2. Switching costs (NARROW). Damodaran's switching-cost framework [2][3] notes that the strongest examples (Microsoft Office) come from data lock-in and ecosystem integration. CDW's switching costs are softer: account team familiarity, embedded procurement workflows in customer ERPs, contractual GPO/state-purchasing-cooperative relationships (E&I, OMNIA, NASPO, Sourcewell), and configuration histories. A federal customer with a CDW-G contract vehicle and an embedded sales rep will not casually re-bid every PO. But these are relationships, not databases; on any given large refresh, three VARs will be invited to bid. Annual revenue retention is high but gross margins on each renewal are competitive. Verdict: present but porous.

3. Intangibles — vendor authorizations (NARROW). CDW holds top-tier partner status (Cisco Gold, Microsoft Solutions Partner, AWS Premier, Dell Titanium, etc.) across hundreds of vendors. These authorizations are the legal-ish 'license to sell' — accumulated certifications, trained engineers, demo lab investments. Per Damodaran [2], legal protections that come from a third party (the vendor here, rather than a government) are conditional and revocable. Cisco can change rebate structures at any annual partner conference; Microsoft has done so repeatedly with the New Commerce Experience and partner program redesigns. Verdict: useful but rentable from the principal.

4. Brand (NONE). CDW has procurement-officer recognition but zero pricing power with end users. Customers do not pay a premium for the CDW logo on a Cisco switch. Compare to Buffett's See's Candies [6] where brand creates real per-unit pricing power; CDW gets none. Brand value is in the sales force and back-office, not the name plate.

5. Network effects (NONE). No two-sided marketplace dynamics. Adding a customer does not make the vendor side more valuable to other customers in any meaningful way.

Cross-cutting threats:

• Vendor disintermediation. Cisco, HPE, Dell, Microsoft and AWS all have direct sales motions, partner-direct programs, and marketplace transactability that compress the channel. Microsoft's CSP model and AWS Marketplace already let large customers buy software/cloud direct.
• Cloud-shift mix. Hyperscaler IaaS/PaaS spend, where CDW's economics are referral fees and managed services, is structurally lower-margin than reselling a Cisco Catalyst switch. The mix has moved against the highest-margin product categories for a decade and continues.
• Distributor compression. TD SYNNEX and Ingram Micro are happy to skip the VAR entirely with white-label managed services for end customers.

Buffett would notice [4][5] that thin-margin distributors like McLane can still compound when run brilliantly by jockeys like Grady Rosier, but the inherent moat is operational excellence, not a structural advantage. CDW's logistical machine is impressive; the moat behind it is not deep.

Moat verdict: NARROW.

Management

CEO Christine Leahy took over in January 2019 from Tom Richards and has continued the post-IPO playbook of disciplined organic growth, tuck-in services acquisitions, and aggressive capital return. The team communicates clearly through the customer-end-market segmentation (Corporate, Small Business, Public — federal, state/local, education, healthcare, plus UK and Canada) and uses 'customer end-market' growth versus US IT market growth as their durable benchmark. That kind of self-imposed scoreboard is a positive signal.

Let me grade the five capital allocation choices:

1. Reinvest in the business. Capex runs ~$110-160M on $20B+ in revenue, so reinvestment is naturally low — this is an asset-light reseller, not a builder of plants. The hidden reinvestment is in coworker count (technical pre-sales engineers, services delivery) and IT systems. ROIIC over 5 years is 8.2%, well below the 15.4% 10-year ROIC average — this is a meaningful warning sign. Incremental capital is going in at returns close to the cost of capital. Grade on this dimension: C+.

2. Acquisitions. CDW has been a serial tuck-in buyer: Sirius Computer Solutions (2021, ~$2.5B, services-heavy), Locus Recovery, AMPLIFIED IT, Berbee, Kelway (UK platform), Mission Cloud Services (2024, AWS premier services), Sandy (federal cyber). The pattern is to buy services capability and bolt on capacity. Sirius arguably overpaid given the post-deal hardware slowdown but added durable services revenue. Mission Cloud was small ($100-200M range, undisclosed). Discipline here looks reasonable: no transformative bets, no debt-fueled lunacy. Grade: B.

3. Debt. Net debt/EBITDA at 2.60x is on the higher end of comfort for a cyclical reseller. The scorer notes interest coverage is null — likely a data quirk, but worth flagging. CDW carries ~$5-6B in long-term debt, refinanced repeatedly at attractive rates pre-2022; some of that wall now sits at 5%+ versus prior 3% notes. Management has stated a 2.0-3.0x leverage target which is being honored. Not aggressive, not conservative. Grade: B-.

4. Buybacks. Share count is down only 2.25% over 10 years — less than I'd expect for a company that's been free-cash-flow generative for the entire decade. Why? Because CDW also pays a meaningful dividend (~1.8% yield) and made the Sirius acquisition. The buyback record is uneven — purchases continued through periods when the stock traded above the long-run average multiple. Average P/IV at point of repurchase is hard to reconstruct, but my read is they have not been counter-cyclically aggressive. Compare to a Buffett-style operator who would have hammered buybacks in 2022's IT-spending recession; CDW kept it steady. Grade: B-.

5. Dividends. Steadily growing per-share dividend since 2014 IPO, with explicit annual review. This is a mature posture appropriate to the business. Grade: A-.

Communication quality. 10-Ks and earnings calls are direct: they distinguish between transactional product revenue and netted-down software-as-a-service/Software/cloud revenue, they disclose end-market growth versus US IT market growth, and they give clear color on the 'large transaction' lumpiness that distorts quarters. They do not over-promise on AI-driven growth, which I respect.

Overall blended grade: B. This is a competent, disciplined, shareholder-aware management team running a structurally average business. They are not Henry Singleton, but they are not destroying capital either. The 8.2% ROIIC is the single most important number on this scorecard and the one to watch quarter by quarter.

Capital allocator: B.

Industry

Porter's Five Forces applied to the IT VAR/reseller industry:

1. Threat of new entrants — LOW to MODERATE. Starting a true competitor to CDW from scratch is essentially impossible: you need vendor authorizations across 1,000+ OEMs, accumulated technical certifications, GPO contract vehicles, a national logistics footprint, and a few thousand sellers. However, niche entrants targeting a single end-market or product category (e.g., a security-only reseller, a pure cloud MSP) appear constantly and chip away at edges. The barrier protects the top tier; the floor is leaky.

2. Bargaining power of suppliers — HIGH and rising. This is the most important force. Cisco, Microsoft, Dell, HPE, Apple, Lenovo, Palo Alto Networks, AWS — these vendors set rebate structures, change incentive programs at will, can launch direct sales motions, and increasingly use marketplaces (AWS Marketplace, Azure Marketplace) and embedded transactability. When Microsoft moves Office 365 from CSP-incumbency-protected to NCE with annual lock-ins, partner economics shift overnight. CDW captures a slice of vendor margin only as long as vendors find the channel cheaper than direct fulfillment. Damodaran [3] points out that distribution-cost-advantage moats can erode fast when the supplier finds a cheaper route.

3. Bargaining power of buyers — HIGH for large customers, LOW for SMB. The 250,000-customer base is bimodal. SMB customers buy on relationship and convenience and accept catalog pricing — favorable. Enterprise, federal and large education customers run formal RFPs against three VARs every refresh cycle and squeeze gross margin to the bone — unfavorable. Public-sector contract vehicles (NASPO, OMNIA, GSA) standardize pricing and reduce CDW's ability to differentiate on price.

4. Threat of substitutes — HIGH and structural. This is the bear case. Cloud (AWS/Azure/GCP) substitutes server hardware. SaaS substitutes packaged software with annual licenses. Vendor-direct programs substitute the channel. Hyperscaler marketplaces substitute software resellers. AI workloads gravitate to GPU cloud, not on-prem servers (and even where they don't, the GPU supply is allocated by NVIDIA directly to hyperscalers and large customers, often bypassing channel). The substitution is gradual but relentless. Per the Damodaran caution about Yahoo/Excite [3], when there is little switching cost and the substitute is meaningfully cheaper or better, the channel is at risk.

5. Competitive rivalry — HIGH. SHI, Insight Enterprises, Connection, ePlus, Presidio, World Wide Technology, Computacenter, Softchoice, plus regional players. Gross margins (excluding netted-down software) sit in the 21-23% range; net product margins after partner program incentives are thin. Rivalry is held in check by the difficulty of lateral customer poaching but not by friendly behavior — every refresh cycle is contested.

Value-pool location and trajectory. The IT spend pool itself is growing — Gartner projects mid-single-digit IT spend growth — but the channel-accessible portion is shrinking as a share. Within CDW's slice, the highest-value pockets are services (cybersecurity, cloud advisory, managed services, professional services) where CDW competes with Accenture, Deloitte, Capgemini and pure-play MSPs. Hardware reselling is being commoditized. Software resale is being marketplaced. CDW management knows this and is moving toward services and netted-down 'solutions' revenue, but it is a 10-year migration on a knife edge.

Industry Verdict: Average. Defensible scale at the top, but five-force pressure is structurally negative on margin trajectory.

Inversion

I am now playing a short-seller. I want CDW to go to $80 within three years. Here is the case, without softening.

1. The single event that kills this. AWS, Microsoft, and Google jointly announce simplified, standardized partner-direct procurement — a 'one-PO, multi-vendor, marketplace-native' workflow that lets a CIO buy Cisco, Dell, Microsoft, AWS and Palo Alto on a single line of credit through Azure or AWS Marketplace, with vendor-direct integration support. This already exists in pieces — AWS Marketplace, Microsoft's NCE, Google Cloud Marketplace — and is gaining transactability and integration tooling every quarter. The day a CFO can run a competitive integrated-IT-stack RFP through three hyperscaler marketplaces and get net-30 terms with embedded financing, CDW's logistical and consolidation moat collapses. This does not require a single dramatic event; it is happening incrementally already. The reseller's right to exist is the customer's unwillingness to manage multi-vendor procurement complexity. Hyperscalers are buying that complexity away.

2. Why the moat is narrower than bulls think. Bulls cite scale and relationships. Both are weaker than they appear. Scale: CDW is large but TD SYNNEX/Ingram Micro are larger one tier upstream and have been quietly building VAR-style services. Relationships: the average tenure of a customer's primary procurement contact is far shorter than the average tenure of a CDW account rep, but the new procurement contact does not inherit the relationship — they re-evaluate. Moreover, CDW's supposed switching costs are nothing like Microsoft Office's [2][3] — there is no data trapped, no workflow embedded, no file format lock. A six-week transition migrates a customer from CDW to SHI. Per Damodaran's warning about Yahoo and Excite [3], when switching cost is low and substitutes exist, growth assumptions are dangerous.

3. Why management is worse than it appears. The headline ROIC of 15.4% is a 10-year average flattered by the COVID-era IT spending boom (2020-2022) when corporate refresh cycles compressed and government stimulus fueled education hardware purchases. The 5-year ROIIC of 8.21% is the cleaner number, and it tells you incremental capital is going in at returns barely above cost of capital. The Sirius acquisition (~$2.5B in 2021) was paid for at cycle-peak hardware demand and has not delivered the services growth promised. Net debt/EBITDA at 2.60x with rising refinancing costs (the 5%+ rate environment is biting) and an interest coverage figure that the scorer could not even compute is concerning. Management's modest 2.25% share count reduction over a decade tells you they were not opportunistic — they were running a pre-set program. A really shareholder-focused jockey would have leaned in hard during the 2022 sell-off.

4. What bulls are extrapolating that won't hold. Bulls extrapolate three things: (a) a return to the 28.82x 10-year average P/E from the current 17.07x; (b) mid-single-digit revenue growth resuming after the 2023-2024 reset; (c) the services/'solutions' mix shift solving the margin compression problem. All three are suspect. The 28.82x average was set during a zero-rate era and a hardware refresh super-cycle; it should not return. Mid-single-digit growth requires the channel to track US IT spend, which itself is increasingly cloud (where CDW captures less). The services mix shift is real but takes a decade and competes head-to-head with Accenture, who plays for keeps. Importantly, the scorer's reverse-DCF implied growth of -1.74% is the market already telling you it does not believe the bull narrative.

5. Valuation trap (multiple compression / regime change). Here is the trap. The scorecard's IV base of $426.70 against a price of $136.03 looks like a screaming buy: 0.32 P/IV. But the IV is built on owner earnings of $1.25B trailing TTM, multiplied by an implicit terminal multiple, and CRITICALLY assumes the maintenance capex assumption holds — and the scorer flagged that uncertainty TWICE. If true maintenance capex is double the assumed figure (plausible for a services-heavy mix shift requiring continuous reinvestment in talent and tooling), owner earnings could be $700-900M, not $1.25B, dropping IV by 30-40%. Combine with multiple compression to a fair 12-14x for a structurally-pressured distributor and you get a fair value of $110-130. The current price is then no margin of safety at all. The 'value' is mirage.

Additional structural concerns reinforcing the bear case:

Concentration in a softening end-market. Roughly a third of CDW's revenue comes from the Public segment (federal, state/local, education, healthcare). Federal IT spend faces continuing budgetary pressure and procurement reform that favors marketplace-style buying (the GSA's evolving e-commerce platforms, OASIS+, etc.). K-12 was massively over-bought during the 2020-2022 ESSER stimulus cycle and is now in digestion mode that is lasting longer than management originally projected.

Working capital as hidden risk. CDW carries large vendor receivables and inventory that can swing meaningfully with the cycle. In a hardware down-cycle, days-payable benefits invert as vendors tighten credit and DIO rises. The 1.15x FCF conversion the scorer credits is partially a working-capital release that does not repeat.

The AI narrative does not save this story. Bulls hope that on-prem AI infrastructure (NVIDIA H100/H200/B100 servers, AI-PCs, edge inference boxes) will drive a refresh cycle through the channel. The reality so far: hyperscalers are getting GPU allocation directly from NVIDIA, large enterprise GPU purchases are routed through specialty integrators (Supermicro, partners with deep AI engineering), and the AI-PC refresh has been a slow build. CDW captures a fraction of this and at competed margins.

If I am right, the stock could be worth $80 within three years.

Lollapalooza Bias Check

Biases active in me as the analyst right now:

Anchoring (high). The IV of $426.70 in the scorecard is a powerful anchor. Even after I argue the maintenance capex assumption may be aggressive, my mental 'fair value' keeps drifting back toward $300+. I should consciously discount the anchor and ask: if I had no scorecard IV, what would I pay? My instinctive number is closer to $180-220 — much lower than the scorecard's range. That gap is the anchor working on me.

Authority bias (medium). The scorecard is described as 'deterministic Python' and 'ground truth.' That framing makes me reluctant to push back on the inputs even where I have a real qualitative reason to (the maintenance capex flag is the obvious one). I am told not to redo the math, but the math is only as good as the maintenance capex assumption inside it. I should treat the scorer as a smart colleague's first draft, not as scripture.

Recency bias (medium-high). CDW had a brutal 2023 and a soft 2024 with hardware demand resetting after the COVID super-cycle. The current 17.07x P/E reflects that reset. I am tempted to either (a) extrapolate the bad period (too bearish) or (b) assume a clean snap-back (too bullish). The honest read is somewhere in between — a slow normalization with structural margin pressure overlaid.

Confirmation bias (medium). I came into this with a pre-existing skepticism of channel/VAR business models from prior reading on Microsoft, AWS marketplace dynamics, and the broader history of distribution disintermediation. I will tend to find evidence for that thesis and discount evidence against it. The strongest counter-evidence I can muster is the durability of the SMB segment and the federal/SLED contract vehicles — both of which I should weight more.

Availability bias (medium). Stories of failed distributors (Tech Data pre-merger, Ingram Micro's stock-market history) are vivid in my memory. Stories of distributors that compounded for decades (Fastenal, Pool Corp, Watsco) are less salient because they are in different end-markets. The base rate for scaled distributors is actually not bad — better than my gut says.

Deprival super-reaction (low). I do not currently own this; nothing to lose. This bias is dormant.

Social proof (low). CDW is not a hot consensus name. No FOMO pressure.

Incentive bias (low). I have no compensation tied to this call.

Net effect: I am probably tilted slightly too bearish, but not dramatically. The anchoring on the scorer IV is the strongest counter-pull. My calibrated 'true' fair value range is $200-260, which still leaves a meaningful margin of safety at $136 but nothing like the 68% the scorer suggests.

10-Year Outlook

Same fundamental business model in 10 years? Mostly yes, but materially different mix. CDW will still aggregate vendors, still field thousands of sellers, still hold GPO contracts. But the revenue mix shifts: hardware reselling shrinks as a share, services and managed/professional offerings grow, software/cloud transactability gets increasingly netted-down (commission economics rather than gross product reseller economics). The business in 2035 looks more like a hybrid Accenture/Computacenter than a 2015-style box-pusher.

Customer base larger? Probably modestly larger. SMB and SLED segments grow with US economy. Federal and education spend is policy-dependent but baseline-stable. Enterprise customer count holds steady. Net: customer count up 10-25% over a decade.

Profit per customer higher? This is the critical question. Two opposing forces: services mix shift raises gross margin (services run 35%+ vs hardware 14-18%), but the absolute revenue per customer in the highest-margin product categories falls as cloud substitutes on-prem. My honest estimate: blended profit per customer roughly flat to up 10% in real terms. Not a compounder's profile.

Moat wider? No. The vendor disintermediation pressure is structural and one-way. Best case: CDW migrates fast enough to services that the moat in services (talent, certifications, customer relationships) replaces the eroding moat in hardware reselling. Worst case: services migration is too slow and the hardware moat erodes faster than the services moat builds. Probability-weighted: moat about the same as today, possibly narrower.

Single biggest threat? Hyperscaler marketplaces becoming the default IT procurement workflow for large enterprises and SLED customers. The threat is not a sudden cliff but a steady annual 1-2 point share migration that compounds. Secondary threat: AI capex concentration in NVIDIA/hyperscaler hands, leaving CDW with small-share GPU resale economics.

Confidence: The business is understandable and management is competent. But the structural trajectory of the value pool is mildly negative and the maintenance-capex uncertainty (flagged twice by the scorer) means the IV could be 30-40% lower than the base case. I am fundamentally uncertain whether 10-year owner earnings per share end up higher or flat. That uncertainty pulls me toward medium rather than high confidence.

CONFIDENCE: medium

Position Guidance

• Recommendation: Buy
• Conviction: medium
• Target buy price: $130 (current price already qualifies; add aggressively under $115)
• Target trim price: $260 (above this, even my haircut IV is exceeded; full trim above scorer base IV $426)
• Position sizing: Initial position 1.5-2% of portfolio. Scale to 3-4% only if price falls below $115 OR ROIIC trends back above 12%. Cap at 5%. Reassess if net debt/EBITDA exceeds 3.0x or if services revenue mix stalls below 25% of gross profit for two consecutive years.