‘AI Roll’ Investors Finify Services firms, can trade more like software companies. That’s what they do wrong

Investors among the investment world, investors are scored for a seductive dissertation: Generative AI, lower margin service enterprises to high margin software companies. Several recognized platforms have signed billions of companies in this strategy and began to bet. This is how the thesis goes:

First, get the traditional business process (BPO) as call centers and accounting firms in the modest assessment of 1x income. These enterprises usually exhale the largest structural resistance to 10-15% EBITDA (interest, depreciation and depreciation) and automation.

Second, to automate core workflows, cut headlines and expand and expand EBITDA margins 40% or more. Once required, hundreds of accountants or call center agents can be made by a handful of man who manages AI systems.

Third, for buyers and public markets, turn out a new Mint-made AI effective services company, because a person has turned a heavy service business. Traditional BPOs trade in 6x EBITDA, program companies order 20x or more.

It is a bright arbitration on paper. In practice, it is a way. Stands in a fundamental category error: Improving surprising surprising operation to work with the transformation of the work model. Yes, AI can make work flow more efficient. No, it does not become a program company to a program company.

Indeed, five years ago, now the notable AI company fled this exact experience and went. The findings should serve today’s believers as a warning. Let’s go in.

69x Assessment Canyon

The most damaged evidence against the AI ​​Rollp dissertation is hiding in a straightforward landscape in public markets. Today’s “AI-Variable” companies that cause automation of BPO companies Conscription, Carpactand infosys-trading from 5-23x EBITDA (EBITDA in the enterprise). Pure software colleagues like Righteousness, Serviceenowand working day, 22-92x Home / EBITDA command assessment. Here is a graph to tell the story:

This is not an indication that can be bridled with press releases about Openai, Anthropic or Gemini Partnership. It is a fundamental difference in how the market is worth the dependent workplaces, which are against the true program platforms.

Consider the concentric that is often as a BPO transformation success story. Although in 2024 and now that there is a great impetus to launch Gen-AI products placement More than 1,000 customers remain in a low figure in the company’s home / EBITDA, and its margin of EBITDA still hounds around 10%. The market message is clear: Automation Workflows do not change your fundamental business model.

Polyi prophecy

In 2019, the leading spoken AI company investigated whether the current person was operated by the current person to accelerate the growth of a six-month. After visiting the opportunity by visiting more than 10 contact centers, the response was not open to establishing relations with three large BPOs and contacting industrial consultants.

“Business process firms are not rewarded for innovations, innovations and not allowed to innovate,” Read the Board’s deck.

Structural barriers that are changed today:

• Illusion of control: BPO does not mean to be the owner of the work you support. You lease the right to supply labor to the conditions of the customer. Technical teams, processes and approvals remain strong in the customer’s hands. AI placement still requires their permission, integration and control. You are not in control; An altering seller.
• Price trap: Project with hours of most service businesses. Get income to improve efficiency reducing paid hours. As the Polyai is found, BPOs promise innovation to win contracts, then return to maximize the calculated hours to protect the margins. It is a substantial work model with automation.
• Zero transition costs: 10 years of service contracts were once in the norm, it is already increasingly widespread to see three years or less. This is not the fact that the frontal AI investments, especially the customer lock, network effect and no moat

Polyay chose to remain a program company with BPO, than to get them. This is the day valuable More than $ 500 million with customers like PG & E, Marriott Fedex. Meanwhile, BPOs thought they were still thinking of buying a digital multitude.

Why isn’t this different this time

Here are what investors are missing: Services enterprises are not accidentally ineffective. Are ineffective by design. As a result of the product. Customers will blame comfort, customization and when everyone is wrong.

Automating the human automating costs not only reduces costs, radically changes what you sell. BPO technology has never been a limit. Customers who want the program will now receive the program.

The most successful services firms understand this. They use AI to increase their people and replace them. They protect the edges with price strength and relationships, not as a result of price reduction. As a result, they still trade in services, because this is.

Classes of history

AI Rollp thesis represents a familiar example of technology in the field of investment: the confusion of technological capacity by conversion of business model. We’ve seen this movie before.

In the early 2000s, the believers think e-commerce will change retailers. Amazon Sears or Barnes and Noble and changed and changed the local digital retailer, established them by building a local retailer. In 2010, investors believed that the program would eat the traditional areas. The winners have already set up new software than to return old.

The same lesson is applied today, but with a narrower coverage. AI, especially existing companies, especially existing companies, may change the corners of professional services, as new tools and new tools that promote new tools. We saw this as sectors in sectors, PE companies as vocational and financial services that are adopted by AI-landslides. However, it is different from the AI ​​Rollp thesis that the low-margin of VCs can be converted to the program similar platforms. For these companies, the transformation will not have a service layer. The new, EU-native companies will come from companies with a completely different economy.

Subline: Software, not service

AI rollup thesis is an attempt to arbitrate a lot of space between the services and programs between enterprise capital. But this space is available for a reason. Services enterprises, even highly automated, various limited restrictions, different customer relations than different economic and software companies.

Polyay saw this in 2019. Public markets see now. The AI ​​revolution is real. The ability to develop service facilities with AI is real. The idea that this improvement has become their software companies? As in 2019, it is difficult to be real today.

AI rollups can still deliver income, but not the underwriting of type VCs. In the best case, technological asset capital is: it is impossible to operate heavy, assessment and scale as a program.

Reviews expressed in Fortune.com are the opinions of the authors and do not necessarily reflect the ideas and beliefs Fortune.

Read more: