BCG and AWS surveyed 1,100+ orgs and found that 75% view partners as major contributors to GenAI ROI.

The report uses a broad buyer-side definition of partners that includes Systems integrators (SIs), consulting, MSPs, data platforms, AI vendors/ISVs.

So here is my partner breakdown of each AI stage:

1. Experimenting: 71% companies engage partners

Key partners involved: consulting partners

What partners do:

identify use cases, build the business case, prioritize opportunities, model value, align and enable teams, and assess dependencies (data readiness, platform foundations, governance, integration).

Partner involvement was 90% in 2024 here - companies moved some GenAI exploration in-house.

2. POC and piloting: 87% / 86% engage partners

Key partners involved:

• Consulting and advisory partners involvement reach 39% during POC.

• SIs rise to 41% during piloting

• 30% of companies cite access to models/tools as the top value partners-add, hinting at ISVs.

What partners do:

stand up POC environments, select the tech stack, manage cloud costs, provide model/tool access, test cost, latency, security, and data readiness, then prove value for budget approval.

3. Deploying: 84% engage partners

Key partners involved:

• SIs are still at 36%

• Data platform partners reach peak importance here.

What partners do:

integrate GenAI into existing systems, support teams and use cases, manage cloud cost, accelerate model training, and provide governed data access at scale.

4. Fully operationalizing: 86% engage partners

Key partners involved:

• AI tooling partners - 33%

• SIs - 30%

• MSPs - 27%

I’d read “AI tooling partners” as AI infra software vendors: observability, model/cost monitoring, data quality, retrieval/search, governance. BCG examples: Dynatrace, Elastic.

5. Agentic transformation: 90% engage partners

The report highlights more reliance on partners broadly at the Agentic stage.

What partners do:

orchestration, governance, workflow design, architecture, and translating operational KPIs into financial impact.

Agentic AI touches the model, data, permissions, workflows, systems, monitoring, and compliance. Companies need help to put this together.

How companies choose and find partners:

• “Experience with the tech stack” moved to #1 criteria

• Speed to execute jumped from #7 to #4

• Cost dropped to #9

For tech partners, cloud provider recommendations rank highly due to “their perspective on ecosystem integration and capabilities necessary”.

For service partners companies rely on peer recommendations.

Lessons for Cloud GTM leaders:

1. Partners are key for companies to build Agentic AI use cases

2. Build proof around technical depth, speed, and business impact - cost is no longer the top 5 selection signal

3. Outbound and marketing are hard - it’s literally at the bottom of selection criteria

Source: BCG