# The graph companies: many of the most valuable companies in the world are graphs, of links, people, products, payments and ontologies, and the graphs never scaled for everyone else because they were built to lock in, sgit.ai > Reading my book, I keep coming back to one idea: some of the biggest and most successful companies in the world are graph companies. Google started from a graph of links and built a Knowledge Graph of things, not strings. Facebook is a social graph, LinkedIn calls itself an economic graph, Amazon builds a product graph on top of a supply chain, and Visa and Mastercard are networks that move trillions. Four of the ten most valuable companies today are, at heart, graphs, worth about $12.8 trillion together. Around them are the companies and consultancies that sell ontologies and taxonomies, a digital twin of an organisation, and the old ones too: legal research has been a citation graph since 1873. Even a spreadsheet, a relational database and a language model are graphs. Once a graph has the right connections, and you remove the nodes that do not matter, the answer becomes obvious, like the end of a whodunit. So why did graphs not scale for everyone else? Because building one took a lot of technology, and whoever built it used it to lock the customer in and keep competitors out: no shared schemas, no shared code. Fractal semantic graphs, where nobody has to share a schema, and agents that maintain them, are how they can scale now. *Source: · sgit.ai v0.7.87 · this file is generated from the same content as the page, so the two cannot drift. Every page on this site has a `.md` twin; internal links below point at them.* --- [Home](../index.md) / [Articles](index.md) / The graph companies: many of the most valuable companies in the world are graphs, of links, people, products, payments and ontologies, and the graphs never scaled for everyone else because they were built to lock in # The graph companies: many of the most valuable companies in the world are graphs, of links, people, products, payments and ontologies, and the graphs never scaled for everyone else because they were built to lock in By [Dinis Cruz](https://sgit.ai/about/index.md) · 2026-10-19 · [article v1.0.0](versions/the-graph-companies.md) · [newsroom v0.12.134](https://newsroom.sgit.ai/admin/versions.html) · graphsstrategystartupsontologiesgen-z-bookarticle ***Abstract:** Reading my book, I keep coming back to one idea: some of the biggest and most successful companies in the world are graph companies. Google started from a graph of links and built a Knowledge Graph of things, not strings. Facebook is a social graph, LinkedIn calls itself an economic graph, Amazon builds a product graph on top of a supply chain, and Visa and Mastercard are networks that move trillions. Four of the ten most valuable companies today are, at heart, graphs, worth about $12.8 trillion together. Around them are the companies and consultancies that sell ontologies and taxonomies, a digital twin of an organisation, and the old ones too: legal research has been a citation graph since 1873. Even a spreadsheet, a relational database and a language model are graphs. Once a graph has the right connections, and you remove the nodes that do not matter, the answer becomes obvious, like the end of a whodunit. So why did graphs not scale for everyone else? Because building one took a lot of technology, and whoever built it used it to lock the customer in and keep competitors out: no shared schemas, no shared code. Fractal semantic graphs, where nobody has to share a schema, and agents that maintain them, are how they can scale now.* The graph companies. Market values on 19 October 2026, with the graph at the heart of each: Alphabet, $4.17 trillion, the link graph and the Knowledge Graph; Microsoft, $3.96 trillion, which owns LinkedIn's economic graph; Amazon, $2.84 trillion, the product graph and the logistics network; Meta, $1.79 trillion, the social graph; Visa, $755 billion, and Mastercard, $538 billion, payment networks; Palantir, $469 billion, an ontology of the organisation. Four of the ten most valuable companies in the world are on the list. ## In short - **The biggest companies are graph companies.** Links, people, products, payments. - **Four of the top ten.** About $12.8 trillion between them. - **Ontologies are a business too.** A digital twin of the organisation, sold and consulted on. - **Everything is a graph.** Spreadsheets, databases, law, contracts, even language models. - **With the right connections, the answer is obvious.** Like the end of a whodunit. - **Graphs never scaled for everyone else.** They were expensive to build, and built to lock in. - **Now they can.** Fractal graphs that need no shared schema, maintained by agents. ## Graph companies Reading my book, I keep coming back to this idea: some of the biggest and most successful companies in the world, especially the ones that do not sell physical resources, are graph companies. Nodes and edges. Connecting A to B to C to D. Look at them: - **Google** began with a graph of links. Its founders' 1998 paper describes a search engine that "makes heavy use of the structure present in hypertext" ([Brin and Page](https://courses.cs.washington.edu/courses/cse454/15wi/papers/google.html)). In 2012 it added the Knowledge Graph, which "understands real-world entities and their relationships to one another: things, not strings", starting with more than 500 million objects and 3.5 billion facts ([Google, 16 May 2012](https://blog.google/products/search/introducing-knowledge-graph-things-not/)). - **Facebook** is a social graph; Mark Zuckerberg used the term when Facebook opened to developers in 2007. Its own researchers measured it in 2011: 721 million active users and 69 billion friendships, with almost everyone in one connected piece ([Ugander and others](https://arxiv.org/abs/1111.4503)). Meta now counts 3.6 billion people a day across its apps ([Meta, July 2026](https://investor.atmeta.com/investor-news/press-release-details/2026/Meta-Reports-Second-Quarter-2026-Results/)). - **LinkedIn** calls itself the Economic Graph, "a digital representation of the global economy", with more than a billion members ([LinkedIn](https://economicgraphchallenge.linkedin.com)). Microsoft paid $26.2 billion for it in 2016. - **Amazon** is a gigantic graph of products, sellers and supply chain. Its scientists have published how they build a product knowledge graph automatically, across thousands of product types ([AutoKnow, 2020](https://www.amazon.science/publications/autoknow-self-driving-knowledge-collection-for-products-of-thousands-of-types)). Last year Amazon's sales were $716.9 billion. - **Visa and Mastercard** are networks. Visa processed 257.5 billion transactions in its last financial year, over 4.9 billion payment credentials ([Visa 10-K](https://www.sec.gov/Archives/edgar/data/1403161/000140316125000089/v-20250930.htm)). By market value on 19 October 2026, Alphabet, Microsoft, Amazon and Meta are four of the ten most valuable companies in the world, worth about $12.8 trillion together ([companiesmarketcap.com](https://companiesmarketcap.com/)). And there is an argument for some of the others: a chip that trains models is a machine for computing graphs. ## Logistics is a graph Any logistics company is the graph that connects the dots. It is the graph that enables the profitability, that tells you what to do at that moment in time. UPS's route optimisation system, ORION, is a graph problem, and UPS said it would save about 100 million miles and 10 million gallons of fuel a year once fully deployed ([UPS, 2015](https://app.3blmedia.com/news/ups-accelerates-use-routing-optimization-software-reduce-100-million-miles-driven)). ## Ontologies are a business Then there are the companies that are very successful because of their ontologies and their taxonomies. Palantir is the poster child these days. It describes its Ontology as "an operational layer for the organization" which, "in many settings, ... serves as a digital twin of the organization" ([Palantir](https://www.palantir.com/docs/foundry/ontology/overview/)). Its revenue grew 56% in 2025, to $4.5 billion. It is not alone. LSEG publishes PermID, permanent identifiers for organisations, instruments and people, as open linked data ([PermID](https://permid.org/about)). Neo4j, a graph database company, passed $200 million of annual recurring revenue in 2024 ([Neo4j](https://neo4j.com/press-releases/neo4j-revenue-milestone-2024/)). ServiceNow agreed to buy data.world in 2025, and described its core as "a semantic layer or a knowledge graph" ([TechTarget](https://www.techtarget.com/searchitoperations/news/366623687/ServiceNow-digs-deeper-into-data-for-AI-with-Dataworld-buy)). And a lot of consulting projects are, at heart, bringing ontologies and taxonomies into an organisation. ## Everything is a graph A lot of software is just connecting data. A spreadsheet is a graph with a couple of dimensions. A relational database, the model [Codd described in 1970](https://dl.acm.org/doi/10.1145/362384.362685), is a graph, probably not in the most efficient way. A language model is a graph: the frameworks that train them record every operation as one ([PyTorch](https://docs.pytorch.org/docs/stable/notes/autograd.html)). Law is a gigantic graph. Legal research has been a citation graph since Shepard's Citations started in 1873, and researchers have measured the importance of precedents by analysing the network of 26,681 US Supreme Court majority opinions ([Fowler and others, 2007](https://www.cambridge.org/core/journals/political-analysis/article/network-analysis-and-the-law-measuring-the-legal-importance-of-precedents-at-the-us-supreme-court/D9B2F13BE2594E1527780AAFDD6BF164)). Contracts are graphs. Governments would be a lot more efficient if they thought in graphs and connected one piece of the puzzle to another, and so would most businesses. ## The answer becomes obvious There is a very cool thing in design: once you have the right questions and the right information, the answer is obvious. It is the end of a whodunit. All the clues were there; what changes at the end is that the right connections are made, the irrelevant ones fall away, and only one answer is left. The answer becomes obvious. On the left, a graph with every node and edge: people, places, events, evidence, and no answer in sight. In the middle, the irrelevant nodes removed, the relevant connections kept. On the right, what is left: a short path from the question to the answer, which now looks obvious. The same move works for a supply chain, a contract, a regulation, or a search result. That is what a graph does once it has the right connections. You remove the nodes that are not relevant, you focus on the ones that are, and the answer becomes obvious. It is why search worked, why a feed knows what to show you, and why a logistics company knows which van goes where. ## Why graphs never scaled So you can argue that graphs have already won. But they won for a few. For everyone else, they did not scale, and I think there are three reasons. Why graphs never scaled, and how they can now. Before: building a graph took a lot of technology; whoever built it used it to lock in the customer and keep competitors out; there was no movement to share schemas or the code; so each graph stayed inside one company, and the advantage stayed with the few. Now: fractal semantic graphs, where each part keeps its own vocabulary and nobody has to share a schema; open source and open schemas; agents that build and maintain the graph; so graphs can be built by anyone, for any domain. First, creating a graph never scaled. It required a lot of technology, and a lot of people to build and maintain it. Second, lock-in. Many of the players were motivated, consciously or not, to lock the customer in. There was no big movement to share schemas, and none to share the code that got created. So whichever company got there first and built the graph gained a competitive advantage, and then used the graph to keep competitors out. Third, some of the nodes in those graphs were treated as commodities when they should not have been, the customer among them, which is what [when the customer becomes the commodity](when-the-customer-becomes-the-commodity.md) is about. The value in graphs is enormous; it is just not evenly distributed, or not very effectively. The contrast is striking. Analysts put the whole market for knowledge graph software at about a billion dollars in 2024 ([MarketsandMarkets](https://prnewswire.co.uk/news-releases/knowledge-graph-market-worth-6-938-4-million-by-2030--exclusive-report-by-marketsandmarkets-302347820.html), an estimate). The companies built on graphs are worth trillions. The graphs were the moat, not the product. ## Now they can scale I think that is changing. [Fractal semantic graphs](introducing-fractal-semantic-graphs.md) are built on the premise that everything connects to everything and nobody has to share a schema: each part keeps its own vocabulary, and the bridges between them are edges somebody declares. Agents are very good at building and maintaining graphs. And open source and open schemas, which already exist for the web in [schema.org](https://developers.google.com/search/blog/2011/06/introducing-schemaorg-search-engines) and for European public services in the [EU's core vocabularies](https://interoperable-europe.ec.europa.eu/collection/semic-support-centre/solution/core-vocabularies), mean a graph does not have to be a moat. So the business case is this. The biggest companies, the most successful ones, and an enormous amount of money are in graphs. What has been missing is a way for everyone else to build them. That is the opportunity for the next generation of founders. ## What this does not claim - **"Graph company" is my lens.** These companies also run ads, cloud services, devices and stores. - **Market values move daily.** These are from 19 October 2026. - **Correlation, not cause.** Graphs are at the heart of these companies; they are not the only reason for their value. - **The market-size figures are estimates.** Analysts define the market differently. - **UPS's savings are its own figures.** ## Where this comes from A voice memo from me on 19 October 2026. Sources as linked: Brin and Page (1998); Google's Knowledge Graph announcement (2012); Ugander and others, The Anatomy of the Facebook Social Graph (2011); LinkedIn's Economic Graph; Amazon's AutoKnow (2020) and its 2025 results; Visa's 2025 10-K; Palantir's documentation and 2025 results; PermID; Neo4j; UPS ORION; Fowler and others (2007); Codd (1970); PyTorch's autograd notes; schema.org; the EU's SEMIC core vocabularies; MarketsandMarkets; companiesmarketcap.com, read on 19 October 2026. Related: [introducing fractal semantic graphs](introducing-fractal-semantic-graphs.md), [when the customer becomes the commodity](when-the-customer-becomes-the-commodity.md), [the graphs books](the-graphs-books.md). Written up and checked by this site's agent, newsroom.writer; the editorial responsibility is mine. ## Threads Graphs & knowledgeStartups & strategy[This article as a graph →](graphs.md#the-graph-companies) ### Builds on - [When the customer becomes the commodity: where your customer sits on your Wardley map, why Google and Amazon lost my trust, and why local delivery should be infrastructure](when-the-customer-becomes-the-commodity.md) Put your customer on your Wardley map. Google and Amazon moved theirs to commodity and lost trust; airports moved back. Local delivery should be infrastructure. - [Fractal Semantic Graphs: everything connects to everything, and nobody has to share a schema](introducing-fractal-semantic-graphs.md) A fractal semantic graph has no privileged level and no single schema: each world keeps its own vocabulary and connects to others through named edges. - [The graphs books: three books written by agents in a day, the altitudes, the board and the book as a graph, what they got right, and why they stopped under review](the-graphs-books.md) Three books on graphs.sgit.ai, written by agents in a day: the altitudes, the board, the book as a graph, and why they stopped under review. [All articles](index.md) · [All graphs](graphs.md) Article No. 215 · [start a reading list with it](https://newsroom.sgit.ai/for/make.html#r=215) **Posting this article on LinkedIn?** The cover is [the-graph-companies.jpg](banners/the-graph-companies.jpg) (1920×1080, title and key ideas on it). Upload it as the article cover, paste the title, then select and copy the body from this page. **Want the next issue by email.** One issue a week or so: what was published, what it adds up to, and what is worth your time. [Subscribe to the SGit Newsroom →](../subscribe/index.md) [← All articles](index.md) --- *[Site index for agents](../llms.txt) · [HTML version](https://newsroom.sgit.ai/articles/the-graph-companies.html)*