Services Are the New Software. What It Means for B2B Marketing.

CEO @ Structured Rebellion

Services Are the New Software. What It Means for B2B Marketing.

The service as a software argument is correct, and it is about to be misread by every marketing team that treats it as a reason to buy more AI. The winners will be the teams that fix the operating model the AI runs on. Everyone else will automate the work they should have redesigned.

Sequoia’s Services: The New Software put a sharp point on a shift that has been building for two years. This is what the thesis actually says, why it is more than venture-capital theater, and what a B2B marketing leader should do about it before the category language shows up in every agency pitch deck.

Key takeaways

Sequoia argues the next trillion-dollar company will be a services firm running on software, because enterprises spend roughly six dollars on services for every dollar on software.
The mechanism matters: sell the tool and you race the model; sell the work and every model improvement makes your service better.
Marketing is a services line. Agencies, retainers, and fractional help are all exposed to the same repricing.
The catch is the one no VC essay leads with: service as a software only compounds on a clean operating model. On a broken one, it automates noise faster.
The marketing teams that win this shift will redesign workflows and buy outcomes, not license more tools.

What Sequoia actually argued

Sequoia’s partner Julien Bek framed it plainly: “The next $1T company will be a software company masquerading as a services firm.” The economic logic is that services dwarf software. In the essay’s words, “for every dollar spent on software, six are spent on services.” Management consulting alone is a $300 to $400 billion market, staffing and recruitment north of $200 billion, IT managed services past $100 billion. That is the pool AI is now reaching, because AI has started doing the work rather than helping a human do it.

The distinction Bek draws is the whole thesis: “A copilot sells the tool. An autopilot sells the work.” And the moat argument: “If you sell the tool, you’re in a race against the model. But if you sell the work, every improvement in the model makes your service faster, cheaper, and harder to compete with.”

Foundation Capital reached the same place from a different door, sizing the services-as-software opportunity at $4.6 trillion, the annual enterprise spend on salaries and outsourced work that AI agents are now positioned to absorb. The label varies. The claim is consistent: the value is moving from software that assists work to systems that deliver it.

Why this is more than a venture narrative

It is easy to file this under VC hype, because venture firms are talking their book. The reason to take it seriously is that the pricing model changes with it, and pricing is where narratives become real.

When a provider sells access to a tool, the customer still owns the outcome. When a provider sells the completed work, the provider owns the outcome and prices against it: per resolved ticket, per booked meeting, per completed audit. That is a different contract, a different risk split, and a different reason to buy. It is already visible in how work is being repriced across services categories, and marketing is not exempt.

The shift also explains a tension marketing leaders already feel. Jasper’s 2026 State of AI in Marketing report found 91% of marketers use AI but only 41% can prove ROI. The service as a software framing reads that gap correctly. The 41% are not the teams with the most tools. They are the teams that redesigned the work the tools were supposed to do. Buying more autopilot does not help if nobody rebuilt the road.

What it means for B2B marketing specifically

Marketing is a services line item. Agencies, retainers, fractional leaders, and freelance execution are all services the buyer purchases because they cannot or do not want to do the work in-house. That is exactly the spend service as a software targets.

So the shift shows up in marketing as three concrete pressures.

The first is on the agency retainer. A retainer prices access to a team’s time. Service as a software prices the work that gets done. As buyers watch AI compress the cost of execution, the hourly and retainer logic gets harder to defend. That is why outcome-based marketing services are moving from a fringe pitch to a buyer expectation.

The second is on what “AI-powered” is allowed to mean. Every agency will soon claim to be a service as a software provider because AI is somewhere in the workflow. Most will be selling faster output on the same broken operating model. Buyers will need a way to tell the difference, which is a positioning opportunity for the providers who can actually show the owned workflow.

The third is on the in-house team’s job. If execution gets cheaper and faster, the scarce skill is no longer producing the work. It is deciding which work matters, defining the outcome, and judging whether the output is right. The human role moves up the stack, not out of the building.

The catch every VC essay buries

Here is the part the thesis pieces mention quickly and then move past: service as a software only compounds on a clean operating model.

An autopilot that delivers the wrong work delivers it faster. AI personalization on a broken ICP scales the wrong message. Outcome pricing on an undefined outcome rewards whoever games the definition. The model does not fix the foundation. It multiplies whatever is already there, which is the same argument behind what service as a software actually means for B2B marketing: the value is in owning the workflow from input to decision, not in the polish of the output.

This is the SR position, and it is deliberately unfashionable. The market is being sold a story about AI replacing the labor. The more useful story is that AI raises the return on a well-designed operating model and raises the cost of a bad one. Faster wrong is not better. It is just more expensive, sooner.

What marketing leaders should do now

Do not respond to the thesis by shopping for autopilots. Respond by finding the workflows worth redesigning.

Pick the recurring, judgment-heavy work where the cost is coordination and interpretation, not creativity: reporting, lead management, campaign QA, competitive research. Ask which of those you would be comfortable paying for by the outcome rather than the hour. That question surfaces two things at once: which workflows are ready for service as a software, and which ones are too broken to price that way yet.

Then hold any provider who uses the language to the ownership test. Which part of the workflow do they own? What does the system do automatically? Where does human judgment enter? What happens when the data is wrong? A provider selling autopilot should be able to answer. A provider selling faster output with AI theater in the deck will not. Our methodology runs that diagnostic before any execution, and the way we engage is built around owning the workflow rather than renting time.

Frequently asked questions

What does “services are the new software” mean?

It means AI has started doing work that used to require human service labor, so the value is shifting from selling software that helps people work to selling the completed work itself. Buyers pay for outcomes, per result, instead of paying for tool access and doing the work themselves.

Is Sequoia right that the next trillion-dollar company will be a services firm?

The economic case is strong: enterprises spend far more on services than on software, and AI is reaching that spend for the first time. Whether a single trillion-dollar company emerges is a prediction. The directional claim, that value moves from tools to delivered outcomes, is already showing up in how work gets priced.

Does service as a software mean the end of marketing agencies?

No. It means the retainer-for-time model gets harder to defend and the outcome-for-work model gets more common. Agencies that own a workflow and can prove it will do well. Agencies that resell faster output on the client’s broken operating model will feel the squeeze.

How is service as a software different from AI marketing tools?

AI marketing tools help your team do the work; you still own the outcome. Service as a software delivers the work and owns the outcome, priced against results. The difference is who is accountable when the work is wrong.

What should a B2B marketing leader do about the shift?

Redesign the highest-value workflows before buying more AI, decide which work you would pay for by outcome rather than by hour, and hold any service as a software provider to a clear ownership test. The advantage goes to teams that fix the operating model first.

So what

The service as a software thesis is not a reason to buy more AI. It is a reason to get honest about which of your marketing workflows are worth owning end to end, and which are too broken to automate safely yet.

The trillion-dollar prediction will take care of itself. The near-term move is smaller and more useful: pick one recurring workflow, define the outcome you would actually pay for, and rebuild it so an autopilot would make it better instead of faster-wrong. Do that a few times and you are not reacting to the shift. You are running it.