Oracle highlights structural shift in partner revenue as AI applications surge

Reporting projects evolve into AI solutions; data and LLM expertise become critical.

As enterprise AI adoption accelerates, the partner ecosystem is entering a phase of structural change. Revenue models tied to traditional analytics, reporting and large-scale data warehousing are evolving, while demand for AI-led applications, domain expertise and modern data architectures is rising.

The shift is no longer about adding AI as another service line. It is about redefining where value sits across data, analytics and application delivery.

“When we go and talk to customers, and this is something our partners also see, the biggest area of focus and spend today is AI applications. However, customers also understand that to build these agents and applications, data is at the core,” said Srikant Gokulnatha, senior vice president, AI Data Platform, Analytics, and Analytical Applications Products, Oracle, in an interaction with CRN India.

In B2B use cases, enterprises need to combine private enterprise data with frontier large language models, creating challenges that extend beyond traditional data warehousing, said Gokulnatha.

Unlike earlier data warehouse projects that focused on structured datasets and gold-layer environments, AI-driven use cases require unstructured and transactional data. In many machine learning deployments, data remains in object stores rather than being fully cleansed and moved into centralised warehouses.

Gokulnatha notes that nearly 70 percent of the effort in building AI agents is tied to data engineering, making data the first hurdle organisations must cross.

For partners, the emphasis does not diminish, it shifts as organisations replace traditional data workloads with enterprise lakehouse models to modernise their architecture and support AI applications.

Customers want to implement AI but often struggle to identify specific use cases, creating an opportunity for partners to step in with domain-focused agent portfolios and structured go-to-market approaches, Gokulnatha added.

From IT consultants to forward-deployed engineers

The partner playbook in AI moves beyond a primary focus on data preparation. Traditionally, partners prepared enterprise data first and defined use cases later, but that sequence now shifts. Partners engage at the business requirement stage, start with the problem, and then build AI solutions around it.

“There are two approaches,” said Gokulnatha. “One is the bottom-up approach which is getting the data ready, identifying the use cases, and then building the solution. But increasingly, I see partners starting with the business problem first.”

This shift reflects how fast AI capabilities are evolving. Solutions built around last year’s model constraints may no longer hold.

As frontier models improve, particularly in areas such as coding and natural language processing, capabilities once considered complex, such as NLP-to-SQL translation, are becoming more automated. That means partners cannot afford static solution blueprints but must continuously adapt.

The result is the emergence of a new operating model.

Gokulnatha mentioned that larger firms are no longer positioning themselves as IT or functional consultants.

“Instead, many now describe themselves as ‘forward deployed engineers’ - teams embedded closely with customers to define AI-driven business outcomes and build, test, and iterate domain-specific agents.”

This structural shift requires reskilling.

“While nearly 70 percent of AI project work remains rooted in data engineering, an area where most partners already have depth, the real investment is moving toward agentic AI competence and advanced data science capabilities,” Gokulnatha said.

Partners are building internal training programmes and working closely with platform vendors to strengthen these skills.

“Where partners need help, and where they are investing in building internal competence, is around how to use agentic AI, and how to deepen and broaden their knowledge of data science and related areas. That is a key area of investment for them,” said Gokulnatha.

He mentioned that partners are increasingly setting up Centres of Excellence (CoE) as part of their AI strategy.

These CoEs function both as labs and as demonstration environments. Partners use them to build, test, and showcase agents developed for specific business use cases.

Gokulnatha added that this is changing how partners go to market. Instead of leading with a number of trained professionals or certifications, partners are presenting ready-built agents designed for specific domains or vertical industries.

Customers can visit these centres, see the agents in action, and assess the use cases directly.

He said this model is increasingly becoming the preferred go-to-market approach for AI-led engagements.

Where partners need to invest next

Traditional analytics and reporting projects are not slowing down as much as they are changing forms. Instead of existing standalone data warehousing or dashboard initiatives, these capabilities are increasingly being embedded within broader AI applications.

Gokulnatha explained that in large enterprises, typically around 50 core reports are widely used across the organisation. However, over time, the total number of reports often expands to 1,500 or more, driven by a long tail of reports used by only a handful of individuals. Maintaining that long tail creates operational overhead.

With AI-driven conversational interfaces, organisations can retain the standardised core reports while replacing the long tail with chat-based access to insights. Users can query data directly instead of relying on static dashboards.

Gokulnatha said, “The underlying demand for analytics does not disappear. Instead, reporting is being translated into AI-powered experiences. As a result, standalone dashboard projects may decline in importance, but analytics remain central, now embedded inside AI agents and enterprise applications.”

This shift has direct implications for partner investment priorities. Gokulnatha said partners should focus on two areas over the next 18 months.

The first is modern data architecture.

He pointed to the growing adoption of open standards, noting that managing data through an open lakehouse approach is becoming a foundational requirement for AI agents and applications.

The second is a deep and continuous understanding of how large language models are evolving.

The assumptions about model capabilities can change within months, said Gokulnatha.

Partners must understand not only model features but also architectural trade-offs, including context window limits, latency implications, and accuracy considerations.

Gokulnatha said, “Patterns such as multi-agent orchestration are becoming more relevant as organisations attempt to move AI into mission-critical workflows. Deploying AI in these environments requires architectural depth, not just experimentation.”

As AI adoption deepens, partner revenue will increasingly depend on reusable domain IP and architectural depth rather than standalone reporting projects or headcount-driven delivery.

Those that combine modern lakehouse foundations with strong agent design capabilities will be better positioned to participate in mission-critical AI deployments, where long-term value and margin will accrue.