Research: Early Gen AI Adopters See 41% ROI

Generative AI is here to stay, and it’s creating tangible value for enterprises right now. As organizations worldwide implement gen AI solutions, they're discovering both remarkable returns and significant challenges. The key questions now focus on how much value gen AI is actually delivering and how each enterprise can devise an optimal gen AI strategy.

To understand the real-world impact of gen AI, we surveyed more than 3,300 organizations worldwide and identified 1,900 early adopters making progress with gen AI. Their results, detailed in our new report "The Radical ROI of Gen AI," reveal compelling evidence: 92% of these early adopters report positive returns. The majority who quantified their ROI see an average 41% return — a figure that's leading them to increase investment across data infrastructure (81% of early adopters), LLMs (78%), supporting software (83%) and talent (76%).

These are global averages. The report provides key highlights for eight specific global regions and six major industries. Here I’ll preview more global numbers, but for regional and vertical insights, you should check out the full report.

From code to customer care 

Generative AI is delivering impressive results across the enterprise, with early adopters reporting "game-changing" or "significant" impacts in percentages consistently above 75%. In technical domains, the technology is revolutionizing how teams work: 54% of development teams are using gen AI to improve code quality and bug detection, while 70% of IT ops teams leverage it for infrastructure optimization and cost analysis. Security teams aren't far behind, with 65% of them focused on improving security posture and reducing incident response times.

The technology is equally transformative in customer-facing functions. While adoption in sales remains relatively low at 38%, those using gen AI report substantial gains in revenue growth and forecast accuracy. Marketing teams (44% adoption) are seeing higher engagement rates through personalized content generation, while customer service units (56% adoption) report improved satisfaction scores through AI-powered chatbots and knowledge management.

Even traditionally less tech-centric departments are seeing strong returns. HR teams are using gen AI to streamline everything from recruitment to performance management, with 60% reporting higher-quality hires. In procurement, 76% of users report game-changing or significant impact, particularly in analytics and contract management. Manufacturing operations are seeing improved demand forecasting and maintenance scheduling, with 79% of users reporting substantial benefits.

These consistently strong results across functions suggest we're just scratching the surface of gen AI's enterprise potential.

Navigating the implementation landscape

While most surveyed organizations (69%) pursue their highest-priority gen AI initiatives, many face difficult strategic choices. Eighteen percent believe customer-facing projects would deliver the strongest impact but focus on employee-facing initiatives due to infrastructure limitations, security concerns and accuracy issues. Another 13% prioritize customer-facing projects despite seeing greater potential in employee applications, often because they've identified ready-to-deploy solutions with more predictable returns.

The challenge of unstructured data is particularly pressing. While comprising 80-90% of enterprise information, only 11% of early adopters have more than half their unstructured data ready for LLM applications. Organizations struggle with time-consuming data management (55%), quality issues (52%) and data sensitivity concerns (50%). The CDOs I’ve been talking to tell me that just in the past year they’ve moved from managing structured data to experiencing a Wild West of unstructured information. The power of generative AI has unlocked their reservoirs of unstructured data, but they sometimes feel like they’re drowning in it.

With practice and established success, we can expect organizations to become more adept at implementing gen AI and working with the vast quantities of data that underlie it. Even so, the technology will continue to evolve in sophistication. Most organizations are pursuing multimodel strategies, whether commercial, open source or both. Ninety-three percent of early adopters plan to deploy at least two LLMs in the next year, with 59% planning for three or more.

Model customization has become standard practice: 96% of early adopters are training, tuning or augmenting their LLMs. This includes fine-tuning with proprietary data (80%) and implementing retrieval-augmented generation (RAG) (71%) to enhance contextual awareness and accuracy. These efforts often require processing multiterabyte data sets, adding another layer of complexity to implementation.

All told, the challenges are dwarfed by the benefits of gen AI, as seen in both the ROI numbers and the near-universal commitment to continued, increased investment in the technology.

Data is the foundational imperative 

Initial proofs of concept can demonstrate gen AI's potential, but scaling to production requires a robust data infrastructure. Early adopters clearly recognize this: 81% plan to increase investment in cloud-based data warehousing over the next year, projecting an average 24% boost in spending.

Security leads organizations' priority list (84% rating it important or critical), matched by demand for advanced AI functionality and integrated analytics capabilities. However, success requires more than technical infrastructure — organizations need well-designed use cases and comprehensive measurement systems to track and optimize performance.

At Snowflake, we’ve been providing customers the infrastructure and insight to meet the current AI moment and prepare for the rapidly growing opportunities around agentic AI. We’re seeing customers approach AI from a systemic, end-to-end perspective that allows them a fuller understanding of their investments and ROI and that lets them create a single, polished user experience.

We’re barely past square one 

While there’s still a lot of discussion in the public space about how generative AI will change the world, and what policies and approaches are best for tech companies, adopters and society at large to pursue, it’s clear that in the enterprise, gen AI is already making a strong mark.

And I say that knowing we haven’t seen an AI-native experience yet. So far we have been taking what we already do and using gen AI to make it faster, better and cheaper. But soon we’ll see AI in the consumer space, with capabilities we haven’t thought of yet. A common analogy from the mobile era is that the first wave of smartphone apps didn’t anticipate anything like Uber, and now there are many variations on ride-hailing and delivery apps that are part of our daily lives.

To learn more about how enterprises are achieving remarkable returns with gen AI, download our full report, "The Radical ROI of Gen AI."