What evidence does Wells Fargo cite that Microsoft and ServiceNow will benefit from cheaper AI?

Microsoft 365 Copilot surpassed 30 million paid seats during fiscal Q4, while Azure revenue exceeded $100 billion for the fiscal year. ServiceNow's AI products crossed $1 billion in annual contract value during Q2, with subscription revenue climbing 24.5% to $3.88 billion. Wells Fargo sees these embedded AI features as examples of how lower-cost intelligence can be layered into entrenched enterprise ecosystems.

Which companies are driving the decline in AI inference costs?

DeepSeek and other open-model providers are accelerating development, alongside renewed U.S. efforts from Meta and Nvidia. Wells Fargo analysts note that more market participants are arguing for adoption of open models in response to accelerating Chinese model distillation and development.

Large Language ModelsOpen-Source AIAI Business & IndustryYahoo Finance AI2h agoAug 13, 2026, 06:00 JST4 min read

Wells Fargo: Cheaper AI Models Boost Microsoft, Not Disrupt It

Send on LINE

Key takeaway

Wells Fargo argues that cheaper AI models from open-source competitors like DeepSeek will actually benefit incumbent software giants Microsoft and ServiceNow rather than disrupt them.
As inference costs fall, these vendors can embed lower-cost AI deeper into their existing enterprise platforms—Microsoft 365 Copilot has already surpassed 30 million paid seats, and ServiceNow's AI products crossed $1 billion in annual contract value—allowing them to expand usage and customer switching costs without needing to compete on model quality alone.

3 Key Points

What happened
Wells Fargo analysts, led by Michael Turrin, argue that falling inference costs from rapid open-model development by DeepSeek, Meta, Nvidia, and others will strengthen established software vendors rather than undermine them. The bank points to Microsoft's Copilot (which surpassed 30 million paid seats in fiscal Q4) and ServiceNow's AI products (which crossed $1 billion in annual contract value during Q2) as examples of how incumbents can embed cheaper intelligence into platforms customers already rely on.
Why it matters
Rather than displacing dominant software vendors, lower-cost AI could expand total AI usage across enterprises—allowing companies like Microsoft and ServiceNow to deepen customer lock-in by integrating AI directly into workflows and products. Wells Fargo sees this as favorable to established players with large installed bases and distribution networks, while smaller, narrower AI vendors face pressure. For investors, the question is whether falling costs translate into higher usage without eroding the pricing power these vendors maintain.
What to watch
Monitor Copilot seat growth, Azure AI consumption, and ServiceNow's AI contract value for signs that usage is accelerating faster than AI economics deteriorate. A sustained rise would validate Wells Fargo's thesis; if customers view AI intelligence as interchangeable, software companies will need to prove that workflow integration, proprietary data, and distribution are strong enough to preserve margins.

Summaries like this, in your inbox every morning.

Get news like this every morning — free

In Depth

Read the full story

Wells Fargo analysts, led by Michael Turrin, have published a research note arguing that the rapid improvement and falling costs of open AI models will benefit established software vendors like Microsoft and ServiceNow rather than disrupt them. The thesis centers on a specific mechanism: as inference costs decline due to accelerating development from DeepSeek and other open-model providers, alongside renewed U.S. efforts from companies including Meta and Nvidia, incumbent software vendors can embed cheaper AI directly into their existing products. This approach is preferable to selling AI as a standalone offering because it leverages an already-established customer base and distribution network.

Microsoft exemplifies this strategy through Copilot, which has been layered into the Microsoft 365 ecosystem. The product surpassed 30 million paid seats during fiscal Q4, demonstrating substantial adoption within the company's enterprise customer base. Azure, Microsoft's cloud platform, generated revenue exceeding $100 billion for the fiscal year, indicating the scale of infrastructure and lock-in already in place. ServiceNow shows a similar pattern: its AI products crossed $1 billion in annual contract value during Q2, while its subscription revenue climbed 24.5% to $3.88 billion, underscoring how AI is increasingly embedded directly into enterprise workflows that customers depend on daily.

Wells Fargo also identifies Datadog, MongoDB, Atlassian, and GitLab as beneficiaries of lower AI costs, arguing that cheaper intelligence will encourage more software development and usage. Conversely, smaller, narrower AI vendors focused on a single capability or model are viewed as facing pressure in this environment. For investors in Microsoft, the critical question is whether falling inference costs will translate into higher AI usage without eroding the company's pricing power. The bank suggests monitoring Copilot seat growth, Azure AI consumption, and ServiceNow's AI contract value as key metrics. If usage accelerates faster than AI economics deteriorate, Wells Fargo's thesis holds. The primary risk is commoditization—if customers increasingly view AI intelligence itself as interchangeable across vendors, software companies will need to prove that workflow integration, proprietary data, and distribution channels remain defensible sources of margin.

Context & Analysis

Wells Fargo's thesis hinges on a structural advantage that incumbents in enterprise software enjoy: a large installed base of customers already paying for their platforms. As open-model development accelerates—driven by DeepSeek's progress and renewed U.S. competition from Meta and Nvidia—the cost of inference (the computational step where an AI produces an answer) is falling. Rather than seeing this as a threat, Wells Fargo argues it is a tailwind for vendors like Microsoft and ServiceNow because they can afford to embed cheaper AI directly into products customers already use and depend on for critical workflows. This contrasts with a scenario in which a single frontier model dominates the market and customers switch to specialized AI-native vendors.

The bank's confidence rests on observable momentum: Microsoft 365 Copilot has already grown to 30 million paid seats, and ServiceNow's AI products generated $1 billion in annual contract value by Q2—both signs that enterprises are adopting AI not as a standalone tool but as a feature of their existing software stack. However, the critical risk Wells Fargo identifies is margin erosion through commoditization. If AI intelligence itself becomes a fungible input—something customers view as interchangeable across vendors—then the pricing power and switching costs that protect these incumbents will erode. The outcome depends on whether usage growth outpaces the decline in AI economics.

FAQ

What evidence does Wells Fargo cite that Microsoft and ServiceNow will benefit from cheaper AI?: Microsoft 365 Copilot surpassed 30 million paid seats during fiscal Q4, while Azure revenue exceeded $100 billion for the fiscal year. ServiceNow's AI products crossed $1 billion in annual contract value during Q2, with subscription revenue climbing 24.5% to $3.88 billion. Wells Fargo sees these embedded AI features as examples of how lower-cost intelligence can be layered into entrenched enterprise ecosystems.
Which companies are driving the decline in AI inference costs?: DeepSeek and other open-model providers are accelerating development, alongside renewed U.S. efforts from Meta and Nvidia. Wells Fargo analysts note that more market participants are arguing for adoption of open models in response to accelerating Chinese model distillation and development.
What is the main risk Wells Fargo identifies for these software vendors?: Commoditization. If customers increasingly view AI intelligence as interchangeable, software companies will need to prove that workflow integration, proprietary data, and distribution remain strong enough to preserve margins.

Yahoo Finance AIRead Original Article