68 VIEWPOINT From fiscal cliff to fiscal cushion Negotiating Microsoft renewals is vital to optimise value, but you’ll need the right data to address both per-seat and token-based spend Ant Attfield is spend optimisation services solutions architect at SHI Wastage across any product or platform in a business typically runs at 30 per cent, so it’s imperative to optimise purchases, renewals and ongoing management. When it comes to licensing, big discounts lead to over-provisioning, only for the business to get caught out at renewal and teeter on the brink of a fiscal cliff. AI, too, introduces usage-based, token-metered consumption with no ceiling – a second, less visible cliff that no seat count can predict. The dilemma is no longer just how to rationalise per-seat provisioning and renegotiate licensing terms by pinpointing where value isn’t being achieved. It’s also how to rein in token consumption – and that requires visibility into which teams, agents and models are burning tokens. Microsoft Copilot Cowork requires an M365 Copilot seat, but the actual work is billed on usage in Copilot Credits, sorted into light, medium and heavy tasks. Cowork is turned off by default and, once it’s accessed, heavy users run costs up fast. Similarly, GitHub Copilot moved to usage-based billing in June 2026, replacing Premium Request Units with token-metered AI Credits. Base subscription prices didn’t change, but there’s no ceiling unless you actively set one, and the old fallback-to-acheaper-model safety net was removed. Data-driven insights provide the negotiating power to address both per-seat and token-based spend when it’s time to renew. Microsoft provides detailed invoice and entitlement data, but dissecting it and accurately mapping usage to users requires specialised tools and expertise. That data should be analysed as part of an optimisation strategy. Beginning the process three to six months ahead of renewal allows the analysis to confirm the required bill of materials (BOM) and negotiation strategy. Large corporates usually bring in a third party to assist with this process and deliver insights to inform the approach the business will take – but their level of engagement can vary. If the partner doesn’t have Microsoft ecosystem expertise and understand the complexity of cloud, on-premises and hybrid licensing – not to mention AI consumption-based pricing which demands advanced modelling – their input will have limited long-term value. That’s because the complexity and interplay between these different solutions demands FinOps and information technology asset management (ITAM) to be combined to offer crossfunctional expertise spanning financial, technical and contractual elements. A truly comprehensive approach to service spend optimisation seeks to explore and understand how the business operates. It should map AI consumption to owners and cost centres, forecast it before it scales, and put guardrails and showback in place. From a per-seat perspective, it should analyse historic licensing rates to see where prices “AI, too, introduces usagebased, token-metered consumption with no ceiling”
RkJQdWJsaXNoZXIy NzQ1NTk=