Computacenter just punched into the FTSE 100. The market cheered. AI boom narrative pulled the stock up like a liquidity injection into a thin order book.
I looked at the filings. I looked at the business model. I see a fragile structure dressed in server racks.
Let me be blunt: this is not a tech company. It is a glorified hardware reseller with a consulting arm. The AI gold rush is real, but the pick-and-shovel sellers that own no mines and hold no ore get crushed when the hype cycle turns.
Context: The IT Services Playbook
Computacenter does exactly what it says: sets up and maintains IT infrastructure for large enterprises. Hardware procurement, deployment, managed services, a thin layer of consulting. Revenue mix? Heavy on low-margin hardware resale — think 5-10% gross margins — and a smaller slice of higher-margin services.
The AI narrative is simple: enterprises need GPU clusters, liquid cooling, and private cloud setups to run their models. Computacenter can design, procure, and manage that infrastructure. So the market priced in a growth premium.
But dig deeper. The company’s technology is not proprietary. Its certifications from Cisco, HPE, VMware are vendor relationships, not moats. The real asset is the number of certified engineers on payroll. You can hire those. Competitors can poach them.
Core: The Numbers Don’t Lie — They Just Hide
Let me quantify the fragility using the same frameworks I apply to DeFi yield strategies.
Revenue Quality
Hardware resale: ~60% of revenue. Gross margin: ~8%. Service revenue: ~40% of revenue. Gross margin: ~25%. Blended gross margin hovers around 15%. That is not a tech company margin. That is a logistics company margin.
For comparison, a SaaS business does 70-80% gross margins. A DeFi protocol with smart contract fees often sees 50%+ take rates. Computacenter’s model is capital-intensive, low-return, and hard to scale without hiring more bodies.
Customer Lock-in vs. Cloud Cannibalization
Yes, switching costs are high. Migrating an enterprise data center is painful. But the cloud — AWS, Azure, GCP — is offering managed AI services that bypass the integrator entirely. Why pay Computacenter to rack and stack GPUs when you can spin up a GPU cluster on AWS in minutes?
The company’s own filings acknowledge this risk. The market ignores it because AI euphoria overrides paranoia.
Growth Sustainability
The AI infrastructure boom is real, but it is a capex spike, not a recurring revenue stream. Once the GPU clusters are deployed, the ongoing service revenue is maintenance — lower margin, commoditized. The real money in this cycle goes to NVIDIA (chip monopoly) and the hyperscalers (scale efficiency). The middleman gets squeezed.
Based on my experience arbitraging GPU rental markets during the 2021 mining craze, I learned that hardware-heavy plays always revert to mean when supply normalizes. The same will happen here.
Contrarian: Retail Sees AI Golden Goose; Smart Money Sees Commodity Race
Retail narrative: “AI infrastructure demand will grow for years — Computacenter is a great proxy.”
Reality check: The infrastructure layer is becoming a commodity. Any certified integrator can deploy NVIDIA DGX systems. The differentiation is zero. The only moat is the existing relationship, and that erodes as procurement departments chase lower costs.
Look at what happened to traditional IT services firms during the cloud transition. Accenture, IBM, Wipro survived by pivoting to high-value consulting and software. Computacenter’s pivot is still mostly hardware. That is a dangerous place to be.
Smart money — institutional investors who read past the press release — will notice the service revenue percentage is not growing fast enough. If next quarter’s earnings show hardware resale outpacing services, that is a sell signal.
I have seen this pattern before. In DeFi, every yield farming protocol that hyped its TVL without showing sustainable fee revenue eventually collapsed. The market is efficient at pricing narratives, but inefficient at pricing structural weakness. That is where the edge lives.
Takeaway: The Real AI Infrastructure Play Isn’t Public Yet
The winners in this cycle will be the projects that own the data, the models, or the proprietary algorithms — not the ones that install the servers. Whether you look at traditional IT or crypto infrastructure, the lesson is the same: the middleman has a target on his back.
Gas is the toll for chaos. Computacenter is collecting tolls, but the highway is about to be bypassed.
Liquidity dries up when fear sets in. Watch for the first earnings miss. That will be the exit liquidity event.
Code is law, but bugs are fatal. In this case, the bug is a business model that depends on low-margin hardware resale during a boom that could end any quarter.
