The TD Cowen Insight
In our survey of 215 North American IT decision-makers, we see that GenAI workloads are migrating quickly to public cloud hyperscalers and Software as a Service (SaaS) players. While just 12% of cloud spend is on GenAI workloads for 2025, that number is expected to rise to 28% by 2028. This shift implies that GenAI spend should increase by four times in coming years, driven by rising adoption across virtually all company functions.
Our Thesis
Our proprietary survey work, captured in this report across seven TD Cowen analyst sectors, suggests GenAI’s tectonic shift is well underway in cloud as adoption of GenAI tools, workloads and budgets continues to rise. Within three years, 42% of surveyed firms expect to spend more than 30% of their cloud platform budget on GenAI and 90% expect it to be at least 10% of their budget.
Alongside our survey work, we include detailed profit and losses of major public cloud platforms, demonstrating the upward trajectory of those businesses amid GenAI’s secular shift.
Over the next several years, we believe revenue growth at major public cloud platforms will be driven by:
ramping GenAI workloads migrating to the cloud and
further non-AI workload migration, with 19% of workloads supported by public cloud platforms in 2025 (24% in five years).
Although we expect healthy topline growth for leading cloud platforms in 2026, the modest deceleration in our out-year forecasts may prove conservative based on our survey data and ongoing historic AI infrastructure capital expenditure cycle from the hyperscalers.
What Is Proprietary?
Since 2012, we’ve published a multidisciplinary annual survey on public cloud computing that has been predictive of critical industry trends. Our latest report largely focuses on GenAI trends and details including:
GenAI spend expectations with public cloud providers over the next three years and
GenAI adoption trends and most impactful use cases across companies:
leading foundation models used and how companies access those models,
agentic AI adoption and
key risks to GenAI adoption.
Our respondents are split evenly among small to medium-sized businesses (SMB), mid-size and enterprises.
Financial and Industry Model Implications
GenAI cloud spend could rise four-fold in 2025 to 2028. Respondents expect cloud-related GenAI spend to rise from 12% of total cloud spend to 28% in three years.
Per our survey data, roughly 60% of companies spend less than 10% of their cloud budget on GenAI workloads. Within the next three years, that workload is expected to be at least 10% of cloud budget for 90% of respondents. Respondents also expect cloud spend to grow 22% year over year in 2026.
What To Watch
GenAI cloud adoption is rising: Three out of four (75%) respondents are using or implementing GenAI Tools compared to 66% in last year’s survey, with adoption rising across nearly all company functions. The use of multiple GenAI tools across organizations continues to rise with 63% using two or more tools, and over 20% using 4+ tools.
Ramping GenAI workloads should drive huge cloud spending increase: Currently, most respondents are spending less than 10% of their cloud spend on GenAI. Looking ahead three years, approximately 90% of surveyed firms expect to spend more than 10% of their cloud budget on GenAI, and 42% of firms expect to be spending 30% or more.
GenAI tools expected to be used across more company functions: Our survey results reveal a large year over year increase in use cases across various company functions including legal (+51%), human resources (+43%), marketing (+23%), customer service (+21%) and finance (+21%). More than nine out of ten respondents who have deployed GenAI workloads have seen productivity and efficiency gains, though fewer than half have seen a commercial impact.
Most impactful GenAI use cases across company functions: Software development (55% identified it as most impactful); customized copy generation for marketing (34%); external facing chatbots (26%); agentic support for customer support (29%); legal drafting for legal (46%); and summarizing historical company performance, simulating scenarios for risk assessment and agentic AI for finance (all at 19%), all within the finance function.
Public Cloud and SaaS are leading platforms for GenAI deployment: The majority of respondents are considering public cloud and SaaS for deployment.
Agentic AI: One-quarter of respondents use agentic AI solutions, led by enterprise adoption, with that number expected rise to 37% in the next three years.
Cloud and GenAI provider relationship: It is somewhat important to select the same GenAI infrastructure provider as their cloud provider according to 63% of those surveyed.
GenAI usage risks: Data security (71%), lack of clarity on the value of deploying GenAI workloads (38%) and fears of model accuracy (34%) are top concerns among our survey respondents.