Hyperscaler contracts in 2026 are no longer a price-list exercise. AWS Enterprise Discount Programs, Azure Consumption Commitments, and GCP Committed Use Discounts each carry distinct commercial mechanics, distinct underspend risks, and distinct marketplace levers. This pillar covers the negotiation architecture across all three.
AWS, Microsoft Azure, and Google Cloud Platform each operate distinct commitment-discount frameworks. AWS uses the Enterprise Discount Program (EDP). Azure uses the Microsoft Azure Consumption Commitment (MACC), typically negotiated alongside the EA. GCP uses Committed Use Discounts (CUD) and the broader Google Cloud commitment frameworks. The mechanics differ enough that a buyer with material spend across two or three of these vendors needs three distinct negotiation strategies, not one.
In our experience across 340+ engagements, the cloud commitments that produce the best long-term economics share three properties: the commitment is sized at 70–80% of mid-case forecast (not 100%), the marketplace offset opportunity is mapped before signature, and the multi-year shape (flat, declining, or ramping) is deliberately chosen rather than accepted as default.
The AWS EDP is a multi-year commitment to a defined annual AWS spend in exchange for a flat percentage discount applied across most AWS services. Commitments typically run three years with annual spend tiers. Discount levels range from approximately 5% (at $1M commitment) to 25%+ (at $50M+ commitment), with tiered steps between. The discount applies to most AWS services, with explicit exclusions for AWS Marketplace third-party purchases (which have their own offset mechanism), Free Tier, and certain enterprise support fees.
Underspend on an AWS EDP is generally billed at the committed value — the buyer pays the difference between actual consumption and commitment. AWS provides a six-month "cure period" in some agreements but the cure is bounded. The underspend risk falls fully on the buyer; the discipline is to size the commitment conservatively. The pattern across enterprises we work with: AWS account teams will push commitment levels at the high end of forecast; buyer-side negotiation should anchor 20–30% below.
AWS Marketplace eligible spend can be applied against the EDP commitment under the AWS Private Pricing for SaaS framework. Eligible products include Databricks, Snowflake, Confluent, MongoDB Atlas and a growing list of enterprise SaaS. Routing eligible third-party spend through AWS Marketplace can offset EDP commitment and often capture additional negotiated discounts on the third-party product itself. The marketplace offset should be mapped before the EDP is signed.
The Microsoft Azure Consumption Commitment is typically negotiated alongside the EA. A MACC commits the buyer to a minimum aggregate Azure consumption over the term in exchange for committed-use discounts, Azure credits, and (most importantly) preferential commercial terms across the broader EA. The MACC has become the lever Microsoft most often uses to extract additional value from the EA — Microsoft commercial teams will move on M365 and Copilot pricing to land a larger MACC.
MACC commitments are typically three-year, expressed as total commitment value with annual schedules. The commitment includes Azure first-party services, eligible Marketplace purchases, and certain other Microsoft cloud services. Underspend is at buyer risk; overspend is at list price unless pre-negotiated. Marketplace consumption (Databricks, Snowflake, Confluent, MongoDB Atlas, Elastic, and many others) counts toward MACC, providing the same offset opportunity as the AWS EDP framework.
The MACC should be negotiated as part of the EA renewal sequence, not as a standalone Azure conversation. The lever exchange between MACC commitment, Copilot pricing, and core M365 pricing only exists at the EA renewal moment. Customers who negotiate MACC in isolation routinely accept commitment levels that are 20–30% higher than they would have committed to in the integrated EA negotiation.
Independent benchmarking across AWS EDP, Azure MACC and GCP CUD. Five working days.
Google Cloud Platform uses two commitment frameworks: Resource-based CUDs (committed to specific machine families and regions, deepest discounts at 25–55%) and Flexible CUDs (committed to spend at 17–28% discount, applicable across most Compute Engine and select services). GCP is generally the most aggressive of the three hyperscalers on commitment-discount depth, particularly for buyers willing to commit to specific machine families.
For larger commitments, GCP offers an Enterprise Agreement equivalent that combines CUD with broader commercial terms — credits, marketplace offsets, and bundled discount on additional GCP services (BigQuery, Vertex AI, Anthos). The GCP EA mechanics are less standardised than AWS EDP or Azure MACC; the negotiation is typically deeper and more flexible, but requires more analytical preparation by the buyer.
Full negotiation playbook covering AWS EDP, Azure MACC, GCP CUD, marketplace offset and egress strategy.
In response to EU Data Act and competition pressure, AWS, Azure, and GCP all changed their data transfer policies in 2024. AWS introduced free egress for customers leaving the platform; Azure and GCP followed with similar policies. These changes did not eliminate egress as a cost — they bounded the worst-case exit scenario. Day-to-day egress between cloud and on-premises, and between cloud regions, remains a material cost line for data-intensive workloads.
Egress is rarely negotiated to zero but is increasingly capped or credited at significant scale. The lever is typically a per-TB egress price cap or a fixed egress credit pool included in the commercial agreement. The right structure depends on the workload pattern — predictable egress patterns benefit from credit pools; spiky patterns benefit from per-TB caps.
Many enterprises operate material workloads across two or three hyperscalers. The temptation is to negotiate a unified cross-cloud master agreement; the practice rarely works. The hyperscalers compete on price specifically because they each want full-platform commitments. Forcing them to negotiate against each other in a single contract conversation typically produces worse terms than separate negotiations executed with deliberate sequencing.
The defensible multi-cloud commercial discipline is sequencing: renew the smallest cloud commitment first, use the benchmark to inform the medium-sized commitment, and bring both to bear in the largest commitment negotiation. The sequence creates compounding leverage and ensures the largest negotiation benefits from real benchmarks rather than vendor-supplied benchmarks.
Independent benchmarking across the full hyperscaler set. We work for buyers only.
If a hyperscaler commitment is in the next 12 months, three preparation moves pay back. First, project consumption at 70–80% of mid-case forecast and commit to that level, not the vendor-suggested level. Second, map marketplace offset opportunities before signing — eligible third-party SaaS that you already buy independently can be routed through marketplace to offset commitment. Third, negotiate egress caps and credits explicitly; left to the standard terms, egress becomes the single largest commercial surprise of the contract. Each of these moves rewards a disciplined software contract negotiation run on the buyer's timeline rather than the account team's quarter-end.
Our cloud advisory practice benchmarks AWS, Azure and GCP commitments across enterprises. Former hyperscaler commercial executives on the team.
Weekly compliance intelligence for IT leaders.