Invisible Technologies company profile: valuation, funding & revenue

This Invisible Technologies company profile covers the company's valuation, revenue, funding rounds, investors and history. Invisible Technologies works in artificial intelligence, with a focus on AI training and enterprise AI services. It was founded in 2015 and is headquartered in San Francisco, California, United States. Its latest valuation on record is More than US$2 billion (reported) (September 2025). Every figure links to a public source, and anything the company hasn't disclosed is marked as such.

Revenue

About US$134 million — Revenue, 2024 (reported), as of 2024. Source: Bloomberg (via Bloomberg Government).

Funding history

Invisible Technologies timeline

Key dates from the sources on this page, oldest first. Metric snapshots are listed after the funding events.

• 2015 — Invisible Technologies is founded in San Francisco by Francis Pedraza (founder); Matt Fitzpatrick (CEO).
• Sep 2025 — Growth round: US$100M, valuing the company at More than US$2B (reported). Investors: Vanara Capital (lead).
• September 2025 — Total funding: About US$144 million after the 2025 round (as reported).

Investors

Overview

Invisible Technologies trains AI models and helps enterprises deploy AI. It combines a software platform with a marketplace of experts who organise, clean, label and evaluate data, and who provide the human feedback that AI labs use to improve their models, a process often called reinforcement learning from human feedback (RLHF). It also works with large companies to turn AI into working systems for their operations. In September 2025, Bloomberg reported that Invisible raised US$100 million at a valuation of more than US$2 billion, and that its revenue roughly doubled to about US$134 million in 2024.

Francis Pedraza founded Invisible in 2015, initially to provide operations work for businesses using a mix of software and human workers, often described as "work as a service". The company has operated as a largely distributed business. Matt Fitzpatrick, a former senior partner at McKinsey who led its QuantumBlack AI division, became CEO and has steered the company towards AI training and enterprise AI work.

Products and services

• AI training: expert data creation, evaluation and feedback for AI labs building frontier models.
• Expert marketplace: a network of specialists in fields such as coding, mathematics, science, law and languages.
• Enterprise AI: services and software that help companies apply AI to their data and processes.
• Data platform: tools to organise, clean, label and map enterprise data for AI.

Funding

Invisible raised relatively little outside capital for most of its history. In September 2025 it raised US$100 million in a round led by Vanara Capital, a new firm that spun out of TPG, which Bloomberg reported valued the company at more than US$2 billion. Reports put total funding at about US$144 million. Because the valuation came from people familiar with the matter rather than the company, we label it as reported.

Business model

Invisible earns revenue from AI labs paying for training data and evaluation work and from enterprises paying for AI implementation services and software. Like other AI data companies, it manages large numbers of contractors and experts, and its margins depend on platform efficiency and the mix of higher-value expert work. Bloomberg reported that the company said it had trained more than 80% of the world's top AI models, a claim that hasn't been independently verified.

Competition

Invisible competes with Scale AI, Surge AI, Turing, Mercor, Handshake and others in AI training data, and with consulting and IT services firms in enterprise AI. Its strengths include long experience managing mixed human and software work, and a leadership team with consulting background for enterprise clients.

Invisible Technologies at a glance

Invisible Technologies is an AI training and enterprise AI company founded in 2015 by Francis Pedraza and led by CEO Matt Fitzpatrick. It provides expert data and feedback for AI labs and helps enterprises deploy AI. It raised US$100 million in September 2025 led by Vanara Capital at a reported valuation of more than US$2 billion, after revenue of about US$134 million in 2024.

Timeline

• 2015: founded as a work-as-a-service company.
• 2022–2023: grows AI training work as generative AI takes off.
• 2024: revenue roughly doubles to about US$134 million.
• September 2025: raises US$100 million at a reported valuation above US$2 billion.

From training AI to deploying AI

Invisible's strategy links two markets. AI labs need expert human input to improve models, and enterprises struggle to turn those models into reliable systems because their data is messy and their processes are complex. Invisible argues that the skills needed to train models, such as organising data, defining tasks and evaluating outputs, are the same skills enterprises need. Moving into enterprise work could diversify revenue away from a small number of AI labs, though it also brings competition from large consulting firms.

Who uses Invisible

Invisible's customers include AI labs and large enterprises in sectors such as financial services, healthcare and consumer goods.

Valuation context

A reported valuation above US$2 billion on 2024 revenue of about US$134 million implies a revenue multiple in the mid-teens, reflecting fast growth. Our guide to AI training data companies compares it with peers.

Risks

• Customer concentration: AI lab spending can shift between vendors quickly.
• Synthetic data: models generating their own training data could reduce demand.
• Competition: many firms target the same labs and enterprises.
• Unverified claims: some company claims about its reach are hard to confirm.

Questions for buyers

• How are experts selected and their work audited?
• How is confidential data protected?
• What results have enterprise deployments achieved?

What RLHF means in practice

Reinforcement learning from human feedback is a technique in which people review a model's responses, rank them or write better answers, and those judgements are used to train the model to prefer helpful, accurate and safe outputs. Early RLHF work often involved general reviewers judging everyday conversations. As models have become more capable, labs need feedback from experts on harder tasks, such as checking a proof, debugging code or assessing a legal argument. Each piece of expert feedback is expensive, so providers compete on recruiting the right experts, designing efficient tasks and checking quality so that the data actually improves models. Invisible's long history of managing mixed human and software workflows is relevant here: much of the value lies in orchestrating many people and quality checks reliably at scale.

Demand for this work is driven by competition among AI labs. As long as labs see expert feedback as a way to improve their models faster than rivals, spending on companies like Invisible is likely to stay strong.

Who funds Invisible

Vanara Capital, which led the 2025 round, was formed by investors who spun out of TPG, and the deal was its first public investment, according to Bloomberg.

For enterprises evaluating Invisible, the most useful evidence is likely to be case studies showing measurable results, such as reduced processing times or improved accuracy, in organisations similar to their own.

Latest developments (October 2026)

Invisible Technologies' September 2025 round of US$100 million, led by Vanara Capital at a valuation of more than US$2 billion, remains its latest, according to Bloomberg Government. Demand for expert data to train AI models has continued to grow, benefiting competitors such as Mercor, Surge AI and Handshake. See our AI training data guide.

What to watch

Watch enterprise growth, AI lab relationships, new funding and competition from Scale AI and others.

Takeaways

AI labs and enterprises can use Invisible for training data and AI deployment. Founders can see how a long-standing operations business found a new market in AI. Investors should note that its valuation is reported.

Sources

Every figure on this page comes from the public sources below. Where sources disagreed, we used the company's own announcement or a primary news report. Private-company data changes; check the date beside each figure.

1. Bloomberg (via Bloomberg Government)
2. CB Insights