Are CRMs Destined to Become a Utility?

In 1926, the Pennsylvania Railroad was the largest corporation in the world. A $210 billion enterprise in today's terms. It employed 279,000 people — more than the federal government. It operated 6,700 trains daily across 28,000 miles of track. Every ton of oil, coal, steel, and agricultural product moving through the American economy traveled on its rails.

The railroad had perfect visibility into commerce. It knew what was moving, where, in what volume, and at what frequency. It sat on top of the most valuable commercial intelligence asset in the world.

And it charged per ton-mile.

The intelligence value of its own traffic — which customer needed which combination of products, at what price, at what time, against which competitor — was dispersed across the heads of thousands of salesmen and hundreds of companies. When they left, the intelligence left with them. The railroad never captured it. Never systematized it. Never monetized it.

If the Pennsylvania Railroad had understood what it was sitting on — if it had captured the intelligence of its own traffic instead of just moving the cargo — it would have controlled American commerce without owning a single barrel of oil or ton of steel. It didn't need to own the goods. It needed to understand them.

Instead, it owned the tracks. And when trucks, cars, and planes arrived, the tracks were all it had left. By 1970, the Pennsylvania Railroad was bankrupt — forty-four years after its peak, brought down not by a single catastrophe but by the slow, steady migration of value from infrastructure to intelligence.

CRM is on the same trajectory. The only question is whether anyone will act in time.

The Evidence Is Already Here

Two weeks ago, Salesforce co-founder Parker Harris stood on stage and asked a question that would have been heresy five years ago: "Why should you ever log into Salesforce again?"

He wasn't being rhetorical. Salesforce's Headless 360 initiative exposes every capability in the platform as an API, MCP tool, or CLI command — so that AI agents can operate the entire system without a human ever opening a browser. Over 100 new tools shipped immediately. The message was explicit: the CRM interface is becoming optional. The platform is becoming infrastructure.

ServiceNow reported 22% year-over-year revenue growth in Q1 2026 and watched its stock drop 17.7% in a single day. Twenty-two percent growth. Negative eighteen percent market reaction. The market wasn't punishing performance. It was punishing relevance.

The stock is down 44% year-to-date. Wall Street has decided that AI agents will bypass structured workflow platforms entirely. ServiceNow's earnings call could have been the Pennsylvania Railroad's in 1926 — record revenue, record throughput, and a market that no longer believes the tracks matter.

HubSpot launched over 100 AI updates in its Spring 2026 Spotlight, including Smart Deal Progression and expanded Breeze agents. Their CPTO said the line that every CRM executive is thinking but few are saying publicly: "Most AI tools have access to data. What they don't have is context."

Every major CRM is racing to become invisible. API-first. Agent-operated. Headless. The user interface — the thing that defined CRM for twenty-five years — is being deliberately dismantled by the companies that built it.

The question is: when the interface disappears, what's left?

What a Utility Looks Like

A utility is essential infrastructure that everyone uses and nobody values. Electricity. Water. Broadband. The defining characteristics are universal adoption, commoditized delivery, and compressed margins. Nobody chooses their utility provider based on innovation. They choose based on reliability and price.

CRM is already crossing this threshold. Three forces are driving it there.

The first is interface commoditization. When Salesforce tells you to stop logging in, they're acknowledging that the screens, dashboards, and workflows that differentiated them for two decades are no longer the value. The value is in the data and the pipes — and pipes are, by definition, utility infrastructure.

The second is AI agent interchangeability. If an AI agent can operate Salesforce through APIs, it can operate any CRM through APIs. The agent doesn't care which platform it's connected to. It cares about the data it can access and the actions it can take. That makes the CRM interchangeable at the agent layer — which is exactly how utilities work. You don't care which power plant generates your electricity. You care that the lights turn on.

The third is margin pressure from platform proliferation. As every CRM adds the same AI capabilities — agent frameworks, workflow automation, intelligent routing, predictive analytics — the features that once commanded premium pricing become table stakes. The race to add AI to CRM is paradoxically accelerating the commoditization of CRM, because every platform ends up offering the same capabilities through the same architecture. When everyone has the same rails, the rails become a commodity.

What the Railroad Teaches Us

The Pennsylvania Railroad's failure is taught in every business school as a cautionary tale about defining your business too narrowly — they thought they were in the railroad business when they were in the transportation business. That's the Theodore Levitt version. It's correct, but it's incomplete.

The deeper failure wasn't about business definition. It was about value capture.

The railroad had something no one else had: perfect visibility into the flow of commerce. Every transaction, every shipment, every customer relationship in the industrial economy passed through its system. The railroad could have used that visibility to become the intelligence layer for American commerce — understanding not just what moved, but why, for whom, and against what alternatives.

Instead, it treated itself as a pipe. It moved cargo from point A to point B, collected its fee, and let the real value — the commercial intelligence — remain dispersed across thousands of salesmen, traders, and merchants who used the railroad but never shared what they knew with it.

The intelligence walked off the platform every night. The railroad was left with tracks.

The CRM Parallel Is Almost Exact

Today's CRM platforms have the same asset the railroad had: perfect visibility into the flow of enterprise commerce. Every deal signal, every meeting transcript, every email thread, every stakeholder interaction, every competitive mention, every pricing conversation, every qualification decision — it all flows through the CRM.

And what does the CRM do with it?

It stores it. It lets sellers log it. It builds dashboards that display it. It charges per seat per month.

The intelligence — the understanding of which customer needs which combination of products, at what price, through which stakeholder, against which competitor, at what moment in the decision cycle — remains dispersed across the heads of individual sales reps. When they leave, the intelligence leaves with them. When they change territories, the new rep starts from scratch. When they go on vacation, the deal stalls.

The CRM sees everything. It understands nothing.

This is not a technology problem. The data is there. The compute is there. The AI models are there. The problem is architectural. CRM platforms were built to be systems of record — databases that store what happened. They were never built to be systems of intelligence — engines that understand what the data means and act on it.

The shift from record to intelligence is not an incremental feature upgrade. It's a different category of system, requiring different infrastructure: persistent memory that survives across meetings and sellers, commercial ontology that maps products to buyer needs, delta reasoning that recalculates strategy when signals change, stakeholder verification that tracks who's involved and how their influence shifts, governed autonomy that runs workflows to real decision boundaries.

None of these exist in any CRM platform today. They're not on anyone's roadmap. They require years of domain engineering in enterprise sales — and CRM companies don't do domain engineering. They build platforms. Platforms are not intelligence.

The Answer Is Not "Better Rails"

The railroad's instinct, when it began to lose relevance, was to build better trains. Faster locomotives. More efficient switching yards. Electrified lines. Each investment improved the infrastructure. None of it addressed the fundamental problem: the value was migrating from the tracks to the intelligence that directed what moved on them.

CRM companies are making the same move today. Salesforce's response to the AI threat is Headless 360 — better rails. ServiceNow's response is AI Agent Fabric and AI Control Tower — a more sophisticated switching yard. HubSpot's response is Breeze agents and Smart Deal Progression — faster trains.

Each of these investments improves the platform. None of them builds the intelligence layer that captures the value of what flows through it.

The answer to "will CRM become a utility?" is not better CRM. It's an intelligence layer on top of CRM — one that captures, reasons over, and acts on everything the platform sees. Not a dashboard. Not a copilot. A persistent reasoning engine that understands deals, not data. That knows which customer needs what, through whom, at what price, against which competitor. That compounds with every interaction and never forgets.

The CRM that hosts the brain becomes the system sellers can't work without. The CRM that remains just a pipe becomes the system sellers are required to use and resent every minute of it. That's the difference between a platform and a utility.

The Lesson for Enterprise Technology

The Pennsylvania Railroad went bankrupt in 1970 — forty-four years after its peak. The decline wasn't sudden. Revenue grew for decades after the value started migrating. The trains kept running. The employees kept working. The earnings reports kept coming. The stock paid dividends for over a hundred consecutive years.

But the value was leaving. Slowly, invisibly, irreversibly. By the time the railroad understood what was happening, the intelligence — the understanding of how to move goods efficiently to the customers who needed them — had been rebuilt by new players in new modalities. The railroad was left with infrastructure that cost a fortune to maintain and no longer commanded a premium.

CRM platforms aren't going bankrupt tomorrow. The Pennsylvania Railroad didn't go bankrupt for forty-four years after its peak. But the value migration has started. The market is telling them — through stock price collapses, analyst downgrades, and the explicit acknowledgment by their own founders that the interface is dying — that infrastructure alone is no longer enough.

The platforms that capture the intelligence will remain relevant. The platforms that remain pipes will become utilities — essential but undifferentiated, used by everyone, valued by no one.

The railroad had the visibility. It missed the intelligence. Every CRM platform today is making the same choice — whether to capture the intelligence flowing through their system, or to keep charging per seat while the real value walks out the door every night, dispersed across the heads of thousands of salespeople who will eventually leave.

The tracks are beautiful. But tracks don't close deals. Intelligence does.

Kevin Gregson is CEO and Co-Founder of SAIQ, the deal intelligence layer for enterprise CRM. SAIQ captures the intelligence value of everything flowing through CRM — persistent deal memory, commercial ontology, and governed autonomy that gives every rep the institutional knowledge of the organization's best performer.