Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
IndustryInsight · On RadarTNW→ original

Magnific Hits $300M in Annual Revenue on Other Companies' AI Models, Without Investors

Magnific, formerly Freepik, does not build its own AI model. Instead, it assembles tools for studios and brands around third-party models such as Kling, Seedance and Wan. Annual recurring revenue (ARR) rose from $230M in April to $300M, and the company has raised no external funding rounds since 2010. For the Singapore studio Hagia Labs, a commercial for the telecom operator StarHub took two to four weeks to produce instead of two to four months of shooting. The lesson for founders: tools built on top of other companies' models can sustain a business without a race to build your own.

AI-processed from TNW; edited by Hamidun News
Magnific Hits $300M in Annual Revenue on Other Companies' AI Models, Without Investors
Source: TNW. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceTNW ↗
Editor-in-Chief
Key metric #1More than 1 million paying subscribers and over 400 enterprise clients
Key metric #2Offices in Shanghai and Singapore, with $10M earmarked for Asia until early 2027
Key metric #3Goal of $1B in revenue by the end of next year, with no IPO planned

Picture an advertising agency that needs a spot about green characters flying across Asia and landing in Japan. The classic route means business trips, film crews and a set in every country. The Singapore studio Hagia Labs produced such a campaign for the telecom StarHub using Spaces, a tool from Magnific, and it took two to four weeks from brief to delivery instead of two to four months. Stories like this one have carried the Spanish platform, formerly known as Freepik, to $300M in annual recurring revenue (ARR, the sum of subscriptions converted to a yearly figure). In April, it was $230M.

A Business Paid for by Customers

Co-founder and CEO Joaquín Cuenca told TNW that Magnific has not taken any external investment rounds since 2010, and that its shift to AI was also funded by its users. The fund EQT acquired a controlling stake in 2020, but it bought shares from Cuenca's partners and did not put new money into the company. According to the company itself, it is the fourth-largest generative AI business in Europe by ARR, after Lovable, ElevenLabs and Mistral.

The company is profitable on an EBITDA basis, but its margin is thinner than its revenue suggests, because almost everything flows back into computing. Whenever a new image or video model launches, users want it the same day, and Magnific pays for it. "We are already spending hundreds of millions of dollars on generating what users ask for," Cuenca says.

He does not rule out outside money: if capital would let the company accelerate, it may happen "possibly next year." There are no IPO plans, because public markets now expect much larger companies. The goal is $1B in revenue by the end of next year, achieved organically. Cuenca values the fact that "there is no timer on us."

Not Its Own Model, but Everything Around It

Cuenca calls competing with Google, ByteDance and OpenAI in building foundation models a "lost battle" for an independent company. His analogy: models are the processor of a MacBook, and Magnific builds the rest of the computer, including the tools, workflows and control settings.

In practice, the platform connects third-party models. Some of the best video models today are Chinese, and users choose them for their quality. Magnific works with Seedance from ByteDance, Kling from Kuaishou, Wan from Alibaba and models from MiniMax. Its suppliers used to be American, and until last year Cuenca had never been to China. Now China has entered the company's top ten markets for the first time.

That explains the office in Shanghai: being close to the teams that build the models makes it easier to integrate new releases quickly and pass user feedback back to them. Singapore will serve as the regional sales base, in a region where the company already has teams in South Korea, Japan and Thailand. Until early 2027, Magnific will spend $10M in the Asia-Pacific region. By the end of 2026 it plans to have more than 20 employees there, and double that in 2027. Enterprise usage in the region grew by 35%. Product and development remain in Málaga.

Studios, Lawyers and Automation

Magnific has more than 1 million paying subscribers and over 400 enterprise clients. Over the past year, the client base has shifted: it now consists of film studios, large marketing agencies and brand departments. The company has moved from "clients find us on their own" to a sales department: about 45 of its 370 employees work on pre-sales, account management, demos and support. The company sends "creatives on site" to clients to train their teams. Individual subscribers remain a funnel: people use the service for years, then bring it into their companies.

Studios need legal guarantees before using outputs in production. Enterprise plans include protection against third-party intellectual property claims. The output belongs to the client, and inputs and outputs are not used to train models. The exception is bad faith: "If you ask for Mickey Mouse and get Mickey Mouse, that's on you." Cuenca argues that responsibility should shift toward the output: AI can infringe rights by accident, so it is important to detect such cases and warn the user.

According to Cuenca, teams first do the same work faster, then start doing things that were previously unaffordable. Brands that used to advertise with a picture and a line of text now produce cinematic-quality spots, and studios launch fantasy and historical projects that used to die for lack of budget. The next step is automation: a team builds a workflow once in Spaces, and employees or their agents in ChatGPT and Claude run it through Magnific's MCP server.

What Follows From This

The caveats are worth remembering. The ARR figures and the European ranking come from the company itself, and thin margins and dependence on the prices of third-party computing remain the main risks. If the best model changes tomorrow, Magnific will have to pay to integrate the new one.

For product managers and founders, the lesson is practical: when models replace one another quickly, the value often sits in the layer around them. That means workflows, legal guarantees for enterprise clients and a sales team that brings in studios. A business like this can be built on customer money without chasing a proprietary model.

Story

Key Questions Answered

What are the key details on A Business Paid for by Customers?

Co-founder and CEO Joaquín Cuenca told TNW that Magnific has not taken any external investment rounds since 2010, and that its shift to AI was also funded by its users. The fund EQT acquired a controlling stake in 2020, but it bought shares from Cuenca's partners and did not put new money into the company. According to the company itself, it is the fourth-largest generative AI business in Europe by ARR, after Lovable, ElevenLabs and Mistral.

What are the key details on Not Its Own Model, but Everything Around It?

Cuenca calls competing with Google, ByteDance and OpenAI in building foundation models a "lost battle" for an independent company. His analogy: models are the processor of a MacBook, and Magnific builds the rest of the computer, including the tools, workflows and control settings. In practice, the platform connects third-party models. Some of the best video models today are Chinese, and users choose them for their quality.

What are the key details on Studios, Lawyers and Automation?

Magnific has more than 1 million paying subscribers and over 400 enterprise clients. Over the past year, the client base has shifted: it now consists of film studios, large marketing agencies and brand departments. The company has moved from "clients find us on their own" to a sales department: about 45 of its 370 employees work on pre-sales, account management, demos and support. The company sends "creatives on site" to clients to train their teams.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Private AI Builders Community · Vibe Club

Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews

Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.

Join Vibe Club →Explore Hamidun Academy
What do you think?
Loading comments…