Enterprise AI Reality Check, Services-Centric ERP, Infosys 2026

ConstellationTV Episode 125 features three conversations about where enterprise AI and automation are really headed: the economics and architecture of AI at scale, the evolution of services-centric ERP, and how global systems integrators like Infosys are turning AI strategy into execution.

Below is a recap of the key themes and takeaways.

Enterprise AI Reality Check: GTC, Tokenomics, and Agent Orchestration

Martin and Larry open the episode with a look ahead to Nvidia GTC, which Martin describes as the “mecca” or even the “Woodstock” of AI. Beyond the spectacle of new chips, they’re focused on a more pressing question: is the current AI spend model sustainable for most enterprises?

From Hype to Economics: Tokenomics and Cost Models

The discussion centers on tokenomics and the underlying AI cost model:

• Today’s AI leaders—Nvidia, Anthropic, OpenAI, the hyperscalers—benefit from an infrastructure-heavy, scale-driven model where more infrastructure can translate into more revenue.
• For “the average bear” enterprise, that dynamic doesn’t hold. The current wave of experimentation, endless pilots, and loosely defined use cases is driving spend without clearly demonstrating value.

Martin and Larry argue that we’re heading toward a reckoning in AI investment. As budgets tighten, organizations will need:

• Clear, well-defined use cases with measurable outcomes
• A shift away from “we’re doing AI because we have to” toward value-anchored initiatives
• A more sane spend profile, especially around inference costs and token usage

Less Agent Sprawl, More Orchestration

Another key theme is agent orchestration. Many organizations have rushed to build agents for extremely narrow use cases, leading to:

• Agent proliferation: anyone can spin up an agent; almost everyone has
• Complexity in IT, governance, security, and maintenance
• Rising costs without corresponding gains in business outcomes

The future, they suggest, is fewer, smarter agents that:

• Support multiple use cases per agent
• Produce outputs that can feed multiple workflows and applications
• Are managed via robust wrangling and orchestration capabilities

In other words, enterprises must move from “one agent per micro-use-case” to strategically designed agent ecosystems that optimize spend and value.

Anthropic vs. OpenAI and the Rise of the Context Layer

Martin also calls out the ongoing Anthropic vs. OpenAI narrative. While headlines focus on model releases and benchmark gains, the real shift is:

• Both vendors are building vertically integrated stacks around their LLMs
• The true battleground is not just the model layer, but the context and business layer that sits between models and applications

Referencing Satya Nadella, they note that context will decouple from the model. This points toward:

• Multi-model futures where enterprises mix and match commercial, open-source, and custom models
• The need for business context fabrics that understand the enterprise’s data, processes, and vertical nuances
• Vendors like Zoho and Salesforce are already building these context layers inside their platforms and vertical offerings

The implication: the big “foundation models” we see today may be global proof-of-concepts. Long-term advantage will come from domain-specific, industry-specific, and context-rich AI built on top of (and independent from) any given base model.

Services-Centric Cloud ERP: Automation, Decision Velocity, and a Winner-Takes-Most Market

In the second segment, R "Ray" Wang introduces Constellation’s services-centric ERP ShortList, focused on finance and professional services–driven organizations.

Market Outlook: AI, Automation, and Zero-FTE Ambitions

Ray outlines a forecast for the services-centric ERP market through 2031, with strong growth driven by:

• AI and automation embedded across finance, operations, and services
• A push toward intelligent process automation that moves beyond insight to actual decisioning and execution
• Ambitious goals like moving parts of the back office toward “zero FTEs”—not by eliminating humans entirely, but by automating repetitive processes and elevating human work to higher-value decision making

He frames the future of ERP as being less about transactional systems of record and more about:

• Decision velocity: how quickly an organization can sense, decide, and act
• Augmenting humans with machines (and vice versa)
• Embedding AI and automation into core workflows, rather than bolting it on

What Separates ERP Winners from Laggards

According to Ray, successful services-centric ERP platforms today typically provide:

• Strong enterprise financials
• Grant management and elements of HCM
• Indirect procurement capabilities, with professional services automation, project planning, and travel and expense as important, sometimes optional, pieces

What really differentiates the leaders:

• Native AI and automation capabilities
• Ability to enable intelligent process automation across end-to-end workflows
• Support for fast, high-quality decision making (decision velocity) across finance and services

Intuit Enterprise Suite and the Mid-Market Opportunity

Ray highlights Intuit Enterprise Suite as an important solution on the shortlist, especially for:

• Small-to-midsized enterprises
• Divisions of large companies that need services-centric ERP capabilities without the overhead of heavyweight enterprise systems

The key insight: many still see Intuit as purely small-business focused, but the enterprise suite is increasingly moving up-market and competing in a space that demands serious automation, compliance, and scalability.

Live from Infosys 2026: Data to Decisions, Governance, and the Speed vs. Velocity of AI

The final segment switches to a live cut from an Infosys analyst day in New York City, featuring Liz Miller, Mike Ni, and R "Ray" Wang.

From Data Foundations to Decisioning

Mike emphasizes a shift many enterprises are trying to make:

• Moving beyond data foundations alone
• Focusing on processes where AI and automation directly drive decisions
• Collapsing decision loops and decision trees so that insights lead to automated or semi-automated actions, not just dashboards

This is where platforms like Infosys’ Topaz come into play: enabling domain-specific, jointly developed solutions with customers that speed up time to value.

Responsible AI and Governance as Accelerators

A recurring theme at Infosys: governance and responsible AI.

Liz notes that:

• Governance was front and center in customer conversations
• Infosys’ long-standing reputation for solving complex, high-stakes, long-term transformation problems makes it a natural leader on this front
• For regulated industries, clear governance and responsible AI frameworks are not blockers—they are enablers that allow organizations to move faster with confidence

Infosys’ customers aren’t chasing hypothetical innovation projects; they are solving real, pressing problems, repeatedly bringing those problems back to Infosys over many years.

Expertise vs. Experience, and the Role of Platforms

Ray draws a sharp line between expertise and experience:

• Expertise—knowing tools, methods, and patterns—is increasingly commoditized.
• Experience—deep knowledge of industry processes, regulations, and operational nuance—is what enables organizations to build AI solutions quickly, safely, and at scale.

He argues the winners in this market share three traits:

1. Deep industry and process experience
2. A strong platform foundation for data, governance, automation, and AI
3. A clear strategy and order of operations for transformation

This leads to a critical distinction: speed vs. velocity.

• Speed is movement for its own sake—spinning up pilots and proofs of concept.
• Velocity is speed with direction, aligned to strategy, sequencing, and measurable impact.

In an AI‑driven, winner‑takes‑most market, those that first reach a viable solution in the right domain can capture outsized share—Ray points to patterns like 60% for the first mover, 20% for the next, and a scramble for the remaining 20%.

To win, enterprises need not just to move fast, but to move fast in the right direction, with governance, platforms, and industry experience underpinning their efforts.

Why This Episode Matters

• AI hype is giving way to economics: tokenomics, cost models, and ROI are taking center stage.
• The context and business fabric layer is becoming the true differentiator, not just LLM performance.
• Agent orchestration and governance—of agents, data, and decisions—are now core enterprise capabilities.
• Platforms with deeper industry experience are what turn AI pilots into market‑shaping capabilities.
• Strategy and order of operations determine whether you’re merely fast—or whether you have velocity in a winner‑takes‑most AI landscape.

For CIOs, CTOs, CFOs, COOs, and business leaders navigating AI, ERP modernization, and large‑scale transformation, Episode 125 serves as both a reality check and a pragmatic roadmap for where to focus next.