blog.demetriusnew.com

Category: Uncategorized

  • AI Revenue Is Growing Rapidly. So Why Isn’t the Entire Business Growing as Fast?

    Fast growth in an AI category can coexist with modest company-wide growth. The difference between those two measures matters.

    Infosys’s October 23 results offer a useful case study in how AI growth can appear inside a larger business. The company said AI revenue represented 10.5% of quarterly revenue and grew 29% quarter over quarter. Total revenue, measured in constant currency, grew 1.2% sequentially. Infosys also lowered its FY27 revenue growth guidance to 1%–2%, from 1.5%–3%.

    Those figures do not show that AI caused the slower overall growth. They do show why executives need to look beyond growth within a new product or service category and ask whether it is adding to the company’s total economic value.

    Category growth and company growth measure different things

    A business can build a fast-growing AI practice while its overall revenue grows slowly. That can happen when the new activity is only part of the business, when other areas are flat or declining, or when AI work replaces some existing services.

    The key distinction is whether AI revenue is incremental.

    If a client hires a provider for a new AI project that would not otherwise have existed, that may represent new demand. If the client shifts budget from a traditional service to an AI service, the provider may be changing the mix of its revenue without increasing the total. If AI helps the provider deliver existing work with fewer hours, productivity may rise even as labor-based billings come under pressure.

    These situations can look similar in an AI revenue headline. Their implications for growth and profitability are different.

    Productivity gains can change who captures the value

    Automation can help a services business complete work faster, reduce rework, or deliver more within the same timeframe. That may improve the provider’s cost structure. But if the contract is priced around hours or staffing levels, the client may also expect a lower price when fewer hours are needed.

    Reuters reported that Infosys cited pricing pressure and client requests to pass along AI-related productivity gains among the factors reflected in its revised guidance. That is useful context for understanding the business-model challenge. It does not mean AI alone explains the company’s overall results.

    When a provider saves time, the commercial question becomes: Who captures that productivity value?

    The provider may retain more of the savings through improved margins. The client may capture part of it through lower prices. Or the parties may agree on outcome-based pricing, where payment is tied more closely to results than to labor hours.

    Each approach changes the economics.

    What executives should measure

    AI revenue is worth tracking, but it should sit alongside measures that help explain its contribution to the whole business:

    • How much AI work comes from new customer demand versus existing budgets being redirected?
    • What revenue is won, retained, or replaced because of AI?
    • Does delivery productivity improve margins, reduce prices, or produce a mix of both?
    • Are customers buying additional outcomes, or receiving familiar outcomes through a new delivery method?
    • How are contracts and pricing adapting as the labor required for delivery changes?

    These questions help separate growth in an AI category from net-new growth for the enterprise.

    The executive question

    Infosys’s results are a reminder that a growing AI line item and modest company-wide growth can exist at the same time. That relationship deserves attention, but it is not proof of a single cause.

    The more useful question is whether AI is creating genuinely incremental value—or changing how existing value is delivered, priced, and shared.

    For an entrepreneur, that distinction is practical: Can AI make the business more productive while reducing revenue from the services that built it?

    The answer depends on what customers are paying for, how the work is priced, and whether the business can turn its productivity gains into new demand or stronger economics.

  • The Curtailment Question Data Centers Need to Ask About Air Permits

    Grid flexibility can shift the emissions question

    As electricity demand grows, grid operators are looking for ways to manage periods when supply is tight. Data centers may be asked to reduce demand, and some emergency reliability measures can direct large customers to use onsite backup generation as a last resort.

    That creates an important distinction. A facility can reduce the electricity it draws from the grid without reducing its total energy use. If onsite engines take over some or all of the load, the grid sees a reduction in demand—but the facility has shifted where its electricity is produced.

    That shift raises an air-permitting question alongside the electrical one: What operating rules and permit conditions apply to the generators during the event?

    “Emergency generator” is a regulatory status with conditions

    Stationary engines are subject to federal requirements that can include the New Source Performance Standards for compression-ignition engines under 40 CFR Part 60, Subpart IIII, and the RICE NESHAP under 40 CFR Part 63, Subpart ZZZZ.

    For engines regulated under Subpart IIII, the rules distinguish emergency operation from other use. They generally allow up to 100 hours per calendar year for maintenance and testing, with certain non-emergency use limited to as much as 50 hours within that total. The rules restrict using that allowance for ordinary peak shaving or non-emergency demand response, while providing a narrow exception for specified local reliability dispatches.

    The details of a particular grid event matter. EPA has also stated that operation under certain 2026 Department of Energy emergency orders is considered emergency operation for covered stationary engines. Those orders apply in defined circumstances; they do not establish that every curtailment request qualifies as an emergency.

    So “the grid asked us to curtail” is not, by itself, a complete answer. The operator needs to know who issued the dispatch, under what authority or protocol, and how the facility’s permit addresses the proposed operation.

    Permitting consequences require a facility-specific analysis

    A generator’s operating classification is only one part of the permitting picture. Title V and New Source Review requirements depend on factors such as the facility’s potential to emit, pollutant totals, source aggregation, location, and applicable state rules.

    A major-source determination does not automatically mean that every facility will need the same controls. PSD applicability and its requirements—including any BACT determination—depend on the source and the proposed construction or modification. SCR, diesel particulate filters, continuous emissions monitoring, or dispersion modeling may be relevant in particular cases, but they are not universal consequences of a generator being used during a grid event.

    The same care applies to temporary generation. A portable engine used or intended to remain at one location for 12 months or longer can cease to qualify as a nonroad engine under federal rules. That fact alone does not answer every permitting question: source status, emissions aggregation, permit obligations, and enforcement depend on the circumstances and applicable requirements.

    Treat the dispatch plan as an air-permitting input

    Before relying on onsite generation as part of a demand-response or emergency plan, engineering and environmental teams should review:

    • The facility’s permit conditions and engine classifications.
    • The dispatch authority, purpose, and applicable reliability protocol.
    • Runtime limits and how testing or other operation counts toward them.
    • Potential emissions from the engines operating together.
    • Whether planned changes require agency review or a permit revision.

    There may be other ways to manage a grid event, including reducing flexible workloads, shifting computing tasks, using stored energy, or combining approaches. The right mix depends on the facility’s design, reliability obligations, grid program, and permits.

    Curtailment can help manage grid reliability. But the effect depends on what the site actually does. If the response transfers load from the grid to onsite combustion, the plan should account for the emissions and permit implications as carefully as it accounts for electrical continuity.

    The practical question is not only whether the data center can stay online. It is what powers it during the event—and whether that operating plan fits the facility’s air permit.

    Regulatory references

  • The AI Agent Isn’t the Risk. Unmanaged Autonomy Is.

    A Shift Most Organizations Haven’t Fully Recognized

    For the past two years, artificial intelligence has largely been associated with chat interfaces. Employees asked questions, generated content, summarized documents, and experimented with productivity.

    That phase is ending.

    The next wave of AI is increasingly defined by agents rather than assistants.

    Unlike traditional chatbots, AI agents perform tasks. They don’t simply recommend actions—they increasingly execute them.

    This represents a meaningful shift in enterprise risk.

    From Information to Action

    A chatbot produces information.

    An AI agent may:

    • Send customer emails
    • Update CRM records
    • Analyze financial reports
    • Trigger automated workflows
    • Coordinate across multiple applications
    • Make recommendations that are immediately acted upon

    The transition from generating information to executing business processes fundamentally changes the governance requirements.

    Execution introduces accountability.

    The HR Analogy

    Imagine hiring a new employee.

    Before their first day, HR and management establish:

    • Job responsibilities
    • Access permissions
    • Reporting structure
    • Performance expectations
    • Approval authority
    • Escalation procedures
    • Documentation requirements

    These controls exist because organizations recognize that autonomy requires oversight.

    Ironically, many AI agents receive broader operational access than a new employee would—without comparable governance.

    Governance Must Scale With Autonomy

    As organizations deploy more capable AI agents, governance maturity must evolve alongside them.

    Key questions include:

    Who approves the agent?

    Who authorized deployment?

    Who owns ongoing oversight?

    What decisions can it make independently?

    Every autonomous action should have clearly defined boundaries.

    Not every task deserves full automation.

    What systems can it access?

    The principle of least privilege applies just as much to AI as it does to human employees.

    Who reviews its work?

    Human oversight remains critical for high-impact decisions.

    The objective is not removing humans.

    The objective is placing humans at the correct control points.

    How are mistakes investigated?

    Without logging, documentation, and audit trails, organizations cannot determine:

    • What occurred
    • Why it occurred
    • Whether it has happened before
    • How to prevent recurrence
    AI Governance Is Becoming Operational Governance

    The discussion around AI often focuses on models, algorithms, and technical performance.

    Those remain important.

    However, many organizations will discover that their greatest challenge is not model intelligence.

    It is operational accountability.

    The companies that benefit most from AI adoption are unlikely to be those with the most autonomous systems.

    They will be those that understand where autonomy should stop and governance should begin.

    Final Thought

    Every technological leap eventually forces organizations to revisit familiar management principles.

    AI agents are no exception.

    The question isn’t whether AI can perform work.

    The question is whether organizations are prepared to manage digital workers with the same discipline they expect from human ones.

    Because as AI gains autonomy, governance can no longer remain optional.

    Demetrius New

    Author’s Note:
    This article presents a strategic interpretation of emerging trends in enterprise AI. While individual organizations will vary, the observations reflect broader shifts in compute demand, infrastructure investment, energy planning, and AI governance that are increasingly influencing enterprise adoption.