What Is Actually Happening Right Now
The IT channel is going through something structural, not cyclical. That distinction matters because a cyclical problem gets better when the economy improves. A structural problem requires a different response — a different service mix, a different value proposition, or both.
Three of the channel's biggest revenue lines have moved simultaneously. Hardware margins have compressed to 3–8% as Amazon Business and direct manufacturer portals captured procurement. Software licensing economics changed fundamentally when Microsoft retired Gold and Silver partner designations. And AI-driven self-service tools are reducing the L1 and L2 ticket volume that justified many support contracts.
Microsoft retired Gold and Silver partner competency benefits on January 22, 2025. From October 1, 2025, direct-bill CSP partners must demonstrate $1,000,000 in trailing-12-month CSP revenue to maintain direct status — a threshold that excludes a significant proportion of the mid-sized MSP market. Source: Microsoft Partner Center FAQ — Microsoft Learn ↗None of this reflects poor management. The businesses that built the IT channel were built to serve clients a certain way at certain margins. The platform changed under them.
Why the AI boom has not reached the channel
Global IT spending is at record levels. The AI wave is real. But the majority of AI infrastructure spending — GPU credits, foundation model API access, cloud-native AI services — flows directly to AWS, Azure, and Google Cloud. It bypasses the channel entirely.
Despite widespread confidence in AI as a growth category, many leading partners — including the largest distributors, global SIs, and resellers — are experiencing flat or declining revenues. The AI money is real. It is not reaching the channel. Source: ChannelE2E, April 2026 ↗Meanwhile, clients are deploying AI tools without their MSPs. They built chatbots, automated reporting, ERP integrations, and workflow agents using Zapier, Make, Claude, ChatGPT, and no-code builders — without involving their IT partners. It was faster. It felt cheaper. And most of those deployments are now fragile.
The Margin Reality — Where the Problem Lives
Not all services are created equal. The gross margin on different service lines varies enormously. The following table shows why the current service mix produces unsustainable economics for most MSPs — and where the growth opportunity sits.
| Service line | Gross margin | Trend in 2026 |
|---|---|---|
| Hardware and device resale | 5–8% | Declining sharply. Direct buying has made this nearly unviable. |
| Software licensing (CSP) | 8–15% | Declining. Microsoft CSP changes compressed this materially. |
| Break-fix / hourly support | 12–18% | Declining as clients use AI self-service for simpler issues. |
| Help desk / L1–L2 support | 18–25% | Under pressure from AI-assisted client self-service. |
| Cloud managed services | 38–45% | Stable to growing, but increasingly competitive. |
| Managed security services | 40–55% | Growing. Still high margin. |
| Integration managed services | 55–65% | Growing. Most MSPs do not yet offer it. Near-zero ongoing labor with the right platform. |
| AI Operations Management | 65–80% | Emerging. Clients have built fragile AI deployments. Nobody owns maintaining them yet. |
The two highlighted rows are the opportunity. Both are growing. Both are underserved. Neither requires hiring an integration engineer — if the underlying platform self-heals when APIs change.
What Your Clients Built and Can't Maintain
Over the last 18 months, your clients have been busy. They built AI chatbots, automated email flows, ERP integrations, invoice processing tools, and reporting agents using Zapier, Make, no-code builders, and AI coding assistants. Most of those deployments are now in production. Most of them are fragile in ways their owners do not fully understand.
The vibe coding problem
Andrej Karpathy coined the term "vibe coding" in early 2025 to describe a workflow that was already widespread: describe what you want to an AI, accept whatever it generates, check whether it looks right, and ship it. For prototypes, this is fine. For production systems serving real business operations, the consequences compound.
85% of professional developers now use AI coding tools at least weekly. The majority follow the vibe coding pattern: prompt, generate, glance, commit. By 2026, 85% of developers use AI tools at least weekly. Forrester predicts 75% of technology decision-makers will face moderate to severe technical debt by end of 2026. Source: Kyros, March 2026 ↗ 45% of AI-generated code samples introduce OWASP Top 10 security vulnerabilities — including SQL injection (CWE-89), cross-site scripting (CWE-80), and insecure cryptographic algorithms. This pass rate has not improved across multiple testing cycles from 2025 through early 2026, despite vendor claims to the contrary. Source: Cloud Security Alliance / Veracode, April 2026 ↗ Vibe-coded cloud infrastructure can inflate costs by up to 400% at production scale due to unoptimised database schemas and inefficient query patterns generated by AI tools without architectural oversight. Source: Ravoid, May 2026 ↗The most common failure mode is not dramatic — it is silent. An upstream API changes. A field is renamed. An endpoint moves. The vibe-coded integration keeps running. It just starts producing wrong data. The business discovers the problem weeks later when a report doesn't balance or an order is missing. By that point, the damage has compounded.
The question that opens the conversation with every client: "Have you built any AI tools, automations, or agents in the last 18 months?" Almost every client will say yes. Then ask: "How do you know they're still working correctly?" The pause that follows is the opportunity.
Two New Service Lines — What You Can Now Offer
1. Integration Management as a Recurring Service
Connecting the systems your clients run — their ERP, CRM, finance tools, HR systems — and keeping those connections working automatically. Not as a one-off project. As a managed service you charge for monthly.
What makes this different from what the channel has attempted before: the integration platform must self-heal. When an API changes — and every SaaS platform changes its API regularly — the platform should detect the change, re-infer the data mapping, rebuild the connector, and reactivate it. No engineer required. No support ticket. The client does not know it happened.
At an average billing of £900–£1,100 per client per month, 20 integration clients adds £216,000–£264,000 per year in new service revenue. At 55–65% gross margin and near-zero ongoing labor, this adds roughly £130,000–£170,000 to gross profit — from clients already in the book.
2. AI Operations Management — the emerging category
Your clients have deployed AI tools. Those tools are fragile. They need someone to own keeping them working as the underlying systems change. That someone should be you.
100% of surveyed enterprises plan to expand their use of agentic AI in 2026 — and nearly three-quarters consider it a critical priority or strategic imperative. One-third of all business software will contain agent functions by 2028, a 33× increase in four years. Source: Pixelmojo / Gartner, January 2026 ↗AI Operations Management covers four distinct services:
- The AI Audit — a one-week engagement mapping every automated tool a client has deployed, identifying what is fragile and why. Priced at £3,000–£6,000. Produces its own follow-on pipeline.
- Integration Hardening — rebuilding a fragile vibe-coded connection on a self-healing, monitored, governed platform. Priced at £8,000–£25,000 depending on complexity.
- AI Operations Monitoring — ongoing monthly monitoring of all live agent deployments and integrations. Alert response. Self-healing for issues the platform can resolve. Priced at £600–£1,500 per client per month.
- AI Governance Setup — defining what AI agents are permitted to do, implementing audit trails, producing compliance reporting. Multi-quarter engagement for enterprise clients.
What This Does to Your P&L
Using a £3 million revenue MSP as the example. Conservative assumptions — modest uptake from existing clients, no new client wins, no changes to the rest of the business.
| Line item | Amount |
|---|---|
| Current annual revenue | £3,000,000 |
| Current EBITDA (9% — industry median) | £270,000 |
| New: Integration managed service (20 clients × £11K/yr avg) | +£220,000 |
| New: AI Audit projects (12 audits × £4,500 avg) | +£54,000 |
| New: AI Operations Monitoring (30 clients × £900/mo) | +£324,000 |
| New: Hardening projects (8 projects × £15K avg) | +£120,000 |
| Total new revenue | +£718,000 |
| New gross profit (blended 63% on new services) | +£452,000 |
| Additional operating cost (minimal — platform maintains itself) | −£40,000 |
| New total EBITDA | £682,000 |
That is EBITDA moving from 9% to approximately 21% — from a business that is statistically below the industry break-even threshold to one operating well above it. From clients you already have. Without hiring an integration engineer.