Profit margins have skyrocketed threefold. The AI Rollup, which has become extremely popular in Silicon Valley, aims to transform traditional companies into "AI factories".

Recently, a new trend has emerged in Silicon Valley called AI Rollup.

To put it simply, investment institutions help AI application companies acquire a batch of traditional small companies, then integrate AI technology into them to make them more efficient and profitable, and then scale them up in batches.

A typical example is Crescendo, an AI customer service company.

Last year, it acquired PartnerHero, a customer service outsourcing company, absorbing more than 2,800 employees and over 200 customers at once. It also established a "AI + human" hybrid model: 90% of routine inquiries are handled by AI, and complex issues are taken over by human customer service representatives.

Company executives revealed that this model has quadrupled the profit margin compared to traditional call centers. The Annual Recurring Revenue (ARR) has exceeded $100 million, and the company has achieved profitability.

For investors, there is an additional bonus to this model: the valuation gap. The valuations of many traditional enterprises are still calculated based on profit margins before the emergence of AI. Once they are acquired and AI efficiency is added, both the profit margin and cash flow will improve rapidly, which is like getting a bargain.

That's why a large amount of capital is pouring in. For example, General Catalyst has allocated $1.5 billion from its new $8 billion fund specifically for Rollup; OpenOcean, a Finnish investment institution, has also raised 100 million euros to test the waters.

Even Elad Gil, a well - known angel investor in Silicon Valley (who has invested in Perplexity, Character.AI, and Harvey), has invested in two AI Rollup companies this year.

Today, let's talk about what the magic of the AI Rollup strategy is that can attract so many top - tier investment institutions?

01 AI Rollup is booming: Transforming traditional businesses with AI

Actually, Rollup is not something new. As early as the 1980s, private equity firms used this method to integrate scattered small companies into national platforms and made money through scale and synergy. During the e - commerce boom in 2021, VCs also tried e - commerce aggregators, but it didn't work out in the end, and they lost all their money.

The difference this time is that AI can bring real productivity improvements.

Take the accounting industry as an example. A small accounting firm charges customers $50 per month, with $35 as costs and $15 as profit, resulting in a 30% profit margin.

If a large - scale model reduces the workload by 40%, the cost immediately drops from $35 to $21. Still charging $50, the profit then rises to $29, and the profit margin doubles to 58%, leaving traditional competitors far behind.

This means that AI transformation can release huge profits and cash flows, which can in turn support further expansion, creating a snowball effect.

In this situation, investors can easily profit from the "valuation gap". Many traditional enterprises are priced based on old - fashioned profit margins before the emergence of AI. Once they are acquired and AI efficiency is added, both the profit margin and cash flow will improve immediately, which is like getting a bargain.

Now, more and more capital is starting to smell the opportunity here. For example, General Catalyst has taken out $1.5 billion from its new $8 billion fund specifically for Rollup; even OpenOcean, a Finnish investment institution, has raised 100 million euros to test the waters.

Individual investors are also not idle. Elad Gil, a well - known angel investor in Silicon Valley (who has invested in Perplexity, Character.AI, and Harvey), has invested in two AI Rollup companies this year.

Gil believes that the difference in this integration is that AI can fundamentally reshape the cost structure and improve the profit margin.

Overall, AI Rollup is more like an industrial transformation that combines capital and technology: through acquisitions, it upgrades scattered traditional businesses into scalable AI factories.

02 Profit margins are three times higher than traditional businesses; 105 companies are involved

According to statistics from foreign data websites, currently 105 startups are implementing the AI Rollup strategy. (For the specific list, please refer to: https://www.ai - rollup.fyi/companies)

Looking at the industry distribution, they are mainly concentrated in traditional service industries such as accounting, insurance brokerage, software outsourcing, law, and logistics. The common feature of these industries is that they are highly labor - dependent and have great potential for efficiency improvement.

AI Rollup has found a breakthrough in these "labor - intensive" areas. Now, a number of companies have emerged.

Law: Eudia

In the legal industry, the most representative case is Eudia, an AI legal platform.

In July this year, Eudia spent $42 million to acquire Johnson Hana, a legal service provider, bringing more than 300 legal professionals under its wing at once.

Two months later, it launched Eudia Counsel, the world's first "AI - enhanced law firm". By packaging AI and legal teams into a delivery platform, Eudia can quickly analyze thousands of documents, generate due - diligence lists for mergers and acquisitions, and accelerate contract review and risk identification.

Customer service: Crescendo

In the customer service field, Crescendo's approach is also typical.

In October last year, it acquired PartnerHero, a customer service outsourcing company with 2,800 employees serving more than 200 customers. After the acquisition, Crescendo transformed from a tool provider into a full - stack customer experience platform.

It adopts a "AI + human" hybrid model: 90% of routine inquiries are handled by AI, and complex issues are quickly taken over by human customer service representatives. The results were immediate. After deploying in a regional telecommunications company, the number of phone calls answered doubled in the first week, the gross profit margin increased to 60 - 65%, and customer satisfaction also improved significantly.

Company executives also revealed that this model has quadrupled the profit margin compared to traditional call centers. The ARR has exceeded $100 million, and the company has achieved profitability.

Property management: Dwelly

In the UK, Dwelly has introduced AI into property management. So far, it has acquired six institutions, and after fully deploying the AI platform, the EBITDA profit margin has doubled.

For tenants, the experience has also improved significantly. For example, open - house events can be coordinated automatically throughout the process, and the waiting time for repairs has been reduced by 40%. A better experience has led to higher profitability and has also expanded the scale of the properties managed by Dwelly.

Accounting: Crete PA

In the accounting industry, Crete PA's actions are also eye - catching. Supported by Thrive Capital, this company plans to invest more than $500 million in the next two years to acquire accounting firms across the United States.

Different from traditional financial mergers, the core of Crete PA is to deeply integrate AI and automation tools into daily accounting work, such as data mapping and writing audit memos, thereby reducing repetitive labor and improving efficiency.

From law to customer service, from property management to accounting, more and more companies seem to be proving the feasibility of this model.

Why do so many startups choose this path?

Because it can help enterprises build a unique moat in the competition dominated by AI technology: faster expansion speed + data advantages in vertical industries.

On the one hand, AI Rollup companies can often quickly capture market share with low prices. In industries with high customer stickiness such as accounting and insurance, low prices are not only highly competitive but also can expand the potential customer base, significantly enlarging the market space.

On the other hand, acquisitions can also bring data. Having first - party data means having a valuable set of training materials. Nowadays, proprietary data is the most important moat for vertical AI. However, the reality is that not all software companies can easily obtain customer data, especially in the context of increasingly strict privacy regulations. If a company acquires another company that has already accumulated a large amount of data, it is equivalent to directly obtaining exclusive model - training resources.

Not only can investors make money, but startups can also build a moat while expanding more quickly. In this situation, AI Rollup naturally becomes a choice for both investors and startups to "meet halfway".

03 How to play AI Rollup? GC has provided the answer

Compared with traditional venture capital, the implementation of AI Rollup is obviously much more complex. Because it not only involves capital investment but also requires a large number of mergers and acquisitions and business transformations.

Not long ago, Marc Bhargava, the managing director of General Catalyst (GC), revealed the operation manual and transformation logic of this strategy in an interview.

GC's core strategy is to provide founders with a complete "toolbox". They will form a cross - border team that can help companies quickly enter the market, assist enterprises in obtaining the data needed for training models, and establish an efficient feedback loop.

Currently, GC has invested in eight projects in this area and has personally incubated five of them, including Longlake in the HOA (Homeowners Association) management field and Crescendo in the call - center field.

Specifically, implementing AI Rollup can be roughly divided into three steps:

Step 1: Screen high - value industries.

GC's team studied 70 service industries and finally selected 10 priority areas, all of which have an automation potential of over 30%.

For example, in the accounting field, Kick, a company invested by GC, has achieved 80% automation; in the call - center field, Crescendo has proven that 50 - 70% of processes can be handled by AI agents and large - language models.

As the capabilities of large - scale models improve every 3 - 6 months, these percentages are still rising.

Based on past practices, GC divides "automated tasks" into four categories:

Customer service and support: Answering common questions and communicating with a large number of customers.

Content creation: Writing marketing copy, creating presentations, and drafting emails.

Repetitive data tasks: Filling out forms, verifying information, such as processing insurance claim documents in the insurance industry.

Logic and reasoning: AI can already assist in more complex judgments such as underwriting and pricing. In the "human - machine collaboration" mode, it can provide suggestions close to the final answer for experts to refer to.

Step 2: Form a suitable team.

A key aspect of GC's incubation model is to bring two completely different types of talents to the same table. One type is technical experts who understand AI applications and product development; the other type is "veterans" familiar with mergers and acquisitions and industry operations. Both capabilities are indispensable.

Step 3: Build AI products and services.

The team doesn't immediately completely disrupt the existing work methods but adopts an "overlay" strategy, seamlessly integrating AI into the existing processes.

For example, an HOA manager needs to prepare a board presentation. He still sends an email as before, but now the request is first processed by an AI agent, and then the result is output. This method not only reduces resistance to change but also immediately shows the results.

After the product verifies its feasibility in a small - scale pilot and accumulates 10 - 15 customers, the company enters the merger and acquisition integration stage. A typical example is Crescendo. After proving that its solution can significantly increase the automation rate, it began to acquire larger - revenue call centers to accelerate expansion.

From GC's practice, AI Rollup is not simply about throwing money at acquiring companies but a combination of "capital + technology + team".

Capital provides the ammunition for acquisition and expansion, AI technology brings efficiency improvements, and the cross - border team ensures that these transformations can actually be implemented. All three are indispensable.

Its value lies in not only rejuvenating traditional "labor - intensive" industries but also validating a new business model: AI is not just a tool but a productivity factor directly embedded in the business to drive results.

In the future, as the capabilities of large - scale models continue to improve and more industries join, this type of "AI factory" - style Rollup may become a new trend in the capital market. For investors, this is a new opportunity for scaling; for startups, this is a practical path to break through in fierce competition.

This article is from the WeChat public account "Crow Intelligence Talk", author: Intelligent Crow. Republished by 36Kr with authorization.