Goldman Sachs tracked what S&P 500 enterprises said to investors in the last two quarters:

• Discussed AI productivity: 54% to 54%

• Quantified a specific use case: 10% to 11%

• Connected productivity gains to earnings: 1% to 2%

​Separate research by Neil Thompson from MIT’s Computer Science and AI Lab added another reality check:

“Enterprise adoption remains limited. We find that only around 10-20% of S&P 500 firms are currently using AI in ways that are generating revenue and meaningfully benefitting them.”

Enterprise value is trailing AI capability and infrastructure investment.

What is causing the gap?

SVB’s recent survey of VC-backed enterprise software companies found that the biggest barriers to deeper AI adoption were:

• Lack of expertise: 44%

• Tool fragmentation: 41%

• Data privacy: 34%

• Cost: 18%

​This is a large implementation market: selecting valuable workflows, connecting private data, integrating existing systems, controlling cloud costs, redesigning processes and proving financial impact.

AI is not automatically economical either

Goldman simulated a coding agent at $13.39 per day versus a $300 human benchmark - clearly high ROI use case. But a call-center AI agent still costs $92.90 versus $90 for a human.

But cost isn't the main bottleneck.

Hyperscalers are responding by putting their own engineers inside customer teams

Microsoft committed $2.5B and 6,000 industry and engineering experts to work inside customer teams. AWS committed $1B to FDEs, with thousands of engineers building alongside customers.

These teams turn AI infrastructure and models into production systems and measurable outcomes. Their scale tells us that faster enterprise adoption has become a strategic priority for hyperscalers.

Hyperscaler teams form the high-touch front end, but enterprises still need a broader partner delivery network that would extend deployments across industries, workflows and geographies.

BCG + AWS surveyed 1,100+ orgs - 94% use multiple partner types across their Gen AI journey

75% view partners as major contributors to ROI.

Yet 42% are not satisfied with their partner mix.

Why don't we see the forming of a partner ecosystem that helps companies to adopt AI faster?

Curious about your take - will partners pick up the AI deployment momentum, or lag behind the AI adoption curve themselves?

Three lessons for alliance leaders:

1. Combine product, implementation and measurement in one partner motion

2. Build repeatable plays around specific workflows, with clear cost and outcome benchmarks

3. Package validated deployments for cloud co-sell and marketplace distribution