The Warning

In October 2024, OpenAI CEO Sam Altman delivered an unusual warning to his company's investors. According to multiple sources familiar with the matter, Altman advised them to avoid investing in five specific companies due to competitive concerns. Among those five names: Glean, the enterprise AI search startup founded by former Google Distinguished Engineer Arvind Jain.

The warning was significant. OpenAI rarely acknowledges direct competitors publicly, and Altman's investor communications typically focus on partnership opportunities rather than competitive threats. But Glean's rapid ascent—from stealth mode in 2019 to a $7.2 billion valuation by June 2025—had caught the attention of the world's most valuable AI company.

By the time of Altman's warning, Glean had achieved what few enterprise software companies accomplish: $100 million in annual recurring revenue within three years of commercial launch, customer contracts with Databricks, Duolingo, Reddit, and Instacart, and recognition as Fast Company's number one Most Innovative Company in Applied AI for 2025. The company had raised $410 million across six funding rounds, with its Series F in June 2025 valuing the business at $7.2 billion—a 57% increase from its $4.6 billion valuation just nine months earlier.

For Arvind Jain, the 47-year-old founder who spent a decade building Google's search infrastructure before co-founding cloud backup unicorn Rubrik, the competitive acknowledgment validated years of conviction: enterprise search remained unsolved, and the organization that cracked it would reshape how billions of knowledge workers access information.

This is the story of how a Google search engineer built the enterprise AI search platform that spooked Sam Altman, why $7.2 billion might still undervalue the opportunity, and what Glean's trajectory reveals about the future of enterprise AI.

The Search Engineer Who Saw What Google Missed

Arvind Jain joined Google in late 2003 as employee number 1,000-something, when the company was still private and small enough that Larry Page and Sergey Brin interviewed most engineering hires personally. He arrived with a BTech in Computer Science from the Indian Institute of Technology, Delhi, a Master's from the University of Washington, and prior stints at Microsoft, Akamai, and Riverbed Technologies.

At Google, Jain's career trajectory followed the classic path of the company's technical elite. He worked on search ranking algorithms, the core technology that made Google's search engine superior to Yahoo and AltaVista. He helped design the infrastructure that would eventually process billions of queries daily. By the late 2000s, he had expanded into Maps, leading the MapMaker project that used community contributions to map 190 countries, including India.

Jain's work on YouTube infrastructure and search teams demonstrated his ability to scale systems across different domains. Google promoted him to Distinguished Engineer, a title reserved for the top 1% of the company's technical staff. Only a few dozen engineers held the designation at any given time, placing Jain in the same tier as Jeff Dean, Sanjay Ghemawat, and other Google infrastructure legends.

But by 2013, after more than a decade at Google, Jain observed a paradox that would define his next decade: while Google had perfected consumer search, enterprise search remained fundamentally broken. Inside Google itself, employees struggled to find information across internal wikis, documents, and communication tools. The irony was not lost on the company's search engineers.

"We had built the best search engine in the world for the open internet," Jain told Fortune in a March 2025 interview. "But when I looked at how we searched for information inside Google, or at Rubrik when I went there, it was terrible. You'd spend hours looking for a document you knew existed, or trying to figure out who worked on a project two years ago."

In 2014, Jain left Google to co-found Rubrik with Bipul Sinha, Soham Mazumdar, and Arvind Nithrakashyap. The data security and cloud backup company raised $553 million before going public via direct listing in April 2024 at a $5.6 billion valuation. Jain served as VP of Engineering and Co-Founder, leading the technical architecture that enabled Rubrik to compete with Veeam, Commvault, and legacy backup providers.

The Rubrik experience taught Jain two critical lessons. First, enterprise infrastructure markets could support multiple billion-dollar companies if the incumbents had failed to solve fundamental customer problems. Second, even at a hyper-growth startup like Rubrik, internal knowledge management remained chaotic—engineers couldn't find technical specifications, sales teams couldn't locate competitive intel, and executives struggled to access the institutional knowledge locked in Slack messages and Google Docs.

"At Rubrik, we had maybe 500 people at one point, and already the knowledge management problem was overwhelming," Jain recalled in a 2024 podcast interview. "I kept thinking: if Google's search engineers can't solve this internally, and if a fast-moving startup can't solve it, then nobody has solved it. That's usually a good sign there's a real opportunity."

In March 2019, Jain left Rubrik to start Glean with two other former Google engineers: T.R. Vishwanath (who had worked on Google's distributed systems) and Tony Gentilcore (who had led engineering at Twitter and Pinterest after Google). The founding team's combined expertise spanned search algorithms, distributed infrastructure, and consumer-grade product design—precisely the skill set needed to bring Google-quality search to the enterprise.

The Enterprise Search Graveyard

Jain's conviction that enterprise search remained unsolved was backed by decades of failed attempts. The graveyard of enterprise search startups stretches back to the 1990s, littered with well-funded companies that promised to bring Google-like search to corporate data.

Autonomy, founded in 1996, raised hundreds of millions and went public before being acquired by HP for $11.1 billion in 2011—a deal that later became one of tech's most notorious acquisitions when HP wrote down $8.8 billion of Autonomy's value amid accounting fraud allegations. Fast Search & Transfer, a Norwegian company founded in 1997, sold to Microsoft for $1.2 billion in 2008. Google itself launched Google Search Appliance in 2002, a physical server that companies could install to search their internal documents; Google discontinued the product in 2016 after 14 years of modest adoption.

More recent attempts fared no better. Coveo, founded in 2005, went public via SPAC in 2021 at a $2 billion valuation and promptly crashed 75% as customers complained about implementation complexity and poor relevance. Elastic, which powers search for Uber, Slack, and Microsoft, focuses primarily on log analytics and technical search rather than knowledge worker use cases. IBM Watson's enterprise search ambitions collapsed amid overpromising and underdelivering on AI capabilities.

The fundamental problem, Jain realized, was that enterprise search faced challenges consumer search never encountered. Google could crawl the public web, index standardized HTML pages, and rely on PageRank's link analysis to determine relevance. Enterprise search required connecting to hundreds of different applications (Google Drive, Slack, Salesforce, Jira, Confluence, GitHub, and dozens more), each with different APIs, authentication schemes, and data structures. Relevance couldn't rely on links—a critical email from the CEO had no inbound links, but was infinitely more important than a random wiki page.

Even more challenging, enterprise search needed to respect complex permission systems. An engineering document visible to the infrastructure team but not to sales couldn't appear in sales reps' search results. Consumer search engines like Google never faced this problem—the web is largely public. But enterprise applications have Byzantine permission hierarchies, with access determined by role, team, project, seniority, and often manual overrides.

Previous enterprise search vendors had failed, Jain concluded, because they approached the problem like a consumer search engine with an authentication layer. They crawled documents, built keyword indices, and bolted on permission checks. The result was slow, irrelevant search that returned hundreds of results with no understanding of which mattered.

"The biggest mistake everyone made was thinking enterprise search was a crawling and indexing problem," Jain told Sequoia Capital in a 2025 podcast. "It's actually a knowledge graph and personalization problem. You need to understand not just what documents exist, but who created them, who uses them, what projects they're related to, and what the person searching cares about. That's a completely different architecture than consumer search."

This insight—that enterprise search required an Enterprise Knowledge Graph rather than a document index—became the technical foundation for Glean.

Building the Enterprise Knowledge Graph

Glean's technical architecture diverges fundamentally from previous enterprise search systems. At its core is the Enterprise Knowledge Graph, a proprietary technology that maps not just documents, but the relationships between people, content, and context across an organization.

When Glean connects to a customer's systems—say, Google Workspace, Slack, Salesforce, Jira, and GitHub—it doesn't simply index the text in documents. Instead, it builds a graph of entities and relationships: which people work on which projects, which documents relate to which initiatives, which Slack channels discuss which topics, which code repositories connect to which product features.

The graph continuously updates as employees create documents, send messages, commit code, and close tickets. It learns patterns: when an engineer searches for "authentication bug," the system knows to surface not just documents containing those keywords, but the specific Jira tickets, GitHub pull requests, Slack discussions, and design docs related to authentication issues that person's team has worked on.

Personalization happens at query time. The same search query—"Q3 roadmap"—returns different results for a product manager, an engineer, and a sales executive, because Glean understands their roles, teams, and recent activity. The product manager sees the product roadmap. The engineer sees the technical implementation plan. The sales executive sees the go-to-market strategy.

This approach requires fundamentally different infrastructure than traditional search engines. Glean processes permissions in real-time for every query, checking the searcher's access rights across dozens of connected systems and filtering results accordingly. The company claims sub-300-millisecond query latency despite this complexity, a performance benchmark achieved through aggressive caching, predictive pre-computation, and distributed architecture refined over six years.

The Enterprise Knowledge Graph also enables Glean's AI assistant, launched in 2023 and significantly enhanced in 2024. Rather than simply retrieving documents, the assistant can answer questions by synthesizing information across multiple sources. A query like "What's our competitor strategy for Q4?" might pull data from sales documents, competitive intel in Salesforce, recent Slack discussions, and analyst reports—all filtered by the user's permissions and personalized to their role.

In February 2025, Glean announced Glean Agents, the company's entry into autonomous AI agents for enterprise workflows. Unlike the AI assistant, which responds to user queries, Glean Agents proactively monitor data sources and take actions based on triggers. An agent might monitor GitHub for security vulnerabilities, automatically create Jira tickets when issues are detected, notify the relevant engineers via Slack, and escalate to management if unresolved after 48 hours.

The agents leverage the same Enterprise Knowledge Graph to understand context and permissions. An agent tasked with "notifying relevant stakeholders when deals above $100K are at risk" needs to know who the stakeholders are, which deals they own, what constitutes "at risk" based on Salesforce activity, and how to prioritize notifications based on deal size and stakeholder seniority.

"The Knowledge Graph is what makes agents possible," Jain explained in a Citi Gen AI Summit fireside chat in early 2025. "Without understanding the structure of your organization—who, what, where, when, why—you just have a chatbot that hallucinates plausible-sounding nonsense. The graph gives agents grounding in reality."

Glean's technical moat lies in the depth and accuracy of this graph. The company claims it takes 12-18 months of real usage for the graph to fully mature for a large enterprise customer, as the system learns organizational patterns, team structures, and knowledge flows. This creates natural switching costs—a competitor would need to rebuild that organizational understanding from scratch, losing personalization and relevance during the transition.

The $100 Million ARR Sprint

Glean launched commercially in 2020, emerging from stealth after a year of product development and design partner testing. The timing proved fortuitous: the COVID-19 pandemic forced companies into remote work, amplifying the enterprise knowledge management crisis as tribal knowledge from office hallway conversations evaporated.

Initial sales focused on technology companies and high-growth startups—organizations with distributed teams, rapid employee growth, and knowledge scattered across dozens of SaaS applications. Early customers included Databricks, Confluent, and Grammarly, tech-forward companies where engineering leaders understood the search problem intimately.

The sales motion was bottom-up, starting with free trials for small teams. A product team at Databricks might adopt Glean to find design docs and customer feedback. As search quality improved with usage, adoption spread organically to engineering, then sales, then the entire company. Glean converted these grassroots deployments into enterprise contracts once usage proved value.

By 2022, Glean's customer base had expanded beyond tech. Duolingo, the language learning app with 500+ million users, deployed Glean across product, engineering, and operations teams. Reddit, navigating explosive growth and a 2024 IPO, used Glean to help employees navigate the company's institutional knowledge as headcount doubled. Sony Electronics, a 50,000-person organization with decades of accumulated documentation, adopted Glean to modernize knowledge access for global teams.

The company reached a critical milestone in its fiscal year ending January 31, 2025: $100 million in annual recurring revenue. This represented a doubling of revenue from the previous fiscal year and placed Glean among the fastest SaaS companies to reach the benchmark. For context, Snowflake took four years to reach $100M ARR, Databricks took five years, and Salesforce took six years. Glean achieved it in approximately three years from commercial launch.

Customer expansion drove much of this growth. Glean's net revenue retention—the percentage of revenue retained from existing customers after accounting for churn, contraction, and expansion—reportedly exceeds 130%, according to investors familiar with the company's metrics. This means existing customers expanded their Glean usage by 30%+ annually, either through seat growth (more employees using Glean), feature upgrades (adding AI assistant or agents to basic search), or increased consumption (as query volume grew).

Pricing follows a per-seat model, with reported costs ranging from $20-$40 per user per month depending on features and contract size. Enterprise customers with 1,000+ seats typically negotiate custom pricing, often with volume discounts and multi-year commitments. For a 5,000-employee company paying $30 per seat, Glean could generate $1.8 million in annual revenue from a single customer—making enterprise sales highly lucrative once Glean proved ROI.

That ROI case, according to customer references, centers on time savings. Glean claims its search and AI assistant save knowledge workers 2-4 hours per week by reducing time spent finding information, asking colleagues for context, and searching across multiple applications. For a company paying $100,000 in fully-loaded annual cost per knowledge worker, recovering even 2 hours per week (5% of a 40-hour workweek) represents $5,000 in productivity gains per employee annually.

At $30/month ($360/year), Glean's ROI calculation suggests customers receive $5,000 in value for $360 in cost—nearly a 14x return. While these calculations rely on assumptions about productivity conversion, they provide a compelling business case for procurement departments evaluating Glean alongside other productivity tools.

By mid-2025, Glean served hundreds of customers globally. The company declined to disclose exact customer counts, but investor presentations reportedly cite 400+ paid customers as of June 2025, up from approximately 250 in January 2024. Major customer wins in 2024-2025 included T-Mobile (bringing Glean to tens of thousands of employees in telecommunications), BILL (a fintech company with complex financial data governance), and Samsara (an IoT platform with distributed engineering teams).

International expansion accelerated in 2024, with Glean establishing regional teams in Europe and Asia-Pacific. Approximately 25% of revenue now comes from outside the United States, according to sources familiar with the business, as European enterprises adopt Glean to comply with data sovereignty requirements while maintaining global knowledge access.

The Funding Blitz

Glean's revenue growth attracted increasingly aggressive investor interest, culminating in six funding rounds across five years that valued the company from tens of millions to $7.2 billion.

The company raised a $4.5 million seed round in 2019 from General Catalyst, Sequoia Capital, and other early-stage investors betting on Jain's Google pedigree and the Rubrik team's execution track record. Series A ($15 million, 2020) and Series B ($35 million, 2021) followed at modest valuations, primarily from Sequoia and General Catalyst doubling down.

The funding trajectory changed dramatically in 2022-2023 as Glean's ARR growth accelerated. Series C ($100 million, November 2022) valued Glean at $1 billion, making it the latest unicorn. Kleiner Perkins and Lightspeed Venture Partners joined as new investors, signaling that established enterprise VCs viewed Glean as a category-defining company.

Series D ($200 million, February 2024) at a $2.2 billion valuation reflected the explosion of enterprise AI interest following ChatGPT's launch. Investors bet that Glean's enterprise focus and permission-aware architecture positioned it to capture AI assistant spending that OpenAI and Anthropic couldn't easily address with general-purpose chatbots.

The valuation more than doubled seven months later. In September 2024, Glean raised $260 million in a Series E led by Altimeter Capital and DST Global at a $4.6 billion valuation. The round included new investors DST Global, Craft Ventures, Sapphire Ventures, and SoftBank Vision Fund 2, alongside existing backers Sequoia, General Catalyst, and Kleiner Perkins.

The Series E's $4.6 billion valuation represented a 2.1x increase from Series D just seven months earlier—an unusual markup velocity even in frothy AI markets. According to PitchBook, the valuation implied a revenue multiple of approximately 46x (assuming $100M run-rate ARR at the time of the round), placing Glean among the most expensive enterprise software companies by multiples.

For context, Databricks raised at a 65x revenue multiple in 2021, Snowflake's IPO implied a 100x multiple, and ServiceNow trades at approximately 15x revenue in public markets. Glean's 46x multiple suggested investors expected hyper-growth continuation and eventual market dominance justifying premium pricing.

Just nine months later, in June 2025, Glean raised another $150 million in a Series F at a $7.2 billion valuation. Wellington Management led the round, with participation from existing investors including Capital One Ventures, Altimeter, Citi, Coatue, and DST Global.

The Series F's rapid follow-on suggested Glean was either funding aggressive expansion, building a war chest against competition, or opportunistically raising at attractive valuations while investor appetite remained strong. In a May 2025 interview with PitchBook at Web Summit Vancouver, Jain characterized the fundraising as "more of a statement than a necessity," suggesting Glean's cash flow situation didn't require additional capital but the company accepted funding to accelerate product development and market expansion.

Total funding across six rounds reached approximately $765 million (seed through Series F), though the exact seed and early-stage amounts remain undisclosed. At a $7.2 billion valuation, Glean's dilution implied the company had sold approximately 10-15% equity across all rounds, leaving founders and employees controlling significant ownership—unusual discipline for a company that raised three quarters of a billion dollars.

The investor syndicate reads like a who's who of enterprise software investing: Sequoia Capital (OpenAI, Snowflake), General Catalyst (Stripe, Databricks), Kleiner Perkins (Amazon, Google), Lightspeed Venture Partners (Snap, Affirm), Altimeter Capital (Snowflake, MongoDB), DST Global (Facebook, Airbnb), and SoftBank Vision Fund (DoorDash, Uber). This investor quality provides Glean with not just capital but strategic relationships, customer introductions, and M&A expertise if the company eventually pursues acquisition discussions.

The Competitive Siege

Glean's rapid ascent attracted competition from three directions: horizontal AI platforms (OpenAI, Anthropic), big tech incumbents (Microsoft, Google), and specialized enterprise search startups (Perplexity, GoSearch, Coveo).

The horizontal AI threat emerged most clearly in October 2024, when Sam Altman warned OpenAI investors against funding Glean. The warning likely reflected OpenAI's own enterprise ambitions with ChatGPT Enterprise, launched in August 2023. ChatGPT Enterprise offers similar capabilities to Glean—searching across internal documents, synthesizing information from multiple sources, and answering questions based on company data.

However, ChatGPT Enterprise faces architectural disadvantages in enterprise search. OpenAI's product uploads customer documents to its systems for processing, raising data governance concerns for regulated industries. Glean, by contrast, never moves customer data from its original location—it indexes metadata and permissions but retrieves actual content at query time from customers' own Google Drive, Slack, or Salesforce instances.

This architectural choice allows Glean to navigate complex compliance requirements in financial services, healthcare, and government sectors where data residency and sovereignty matter. A European bank can use Glean while keeping all data in EU-based Google Workspace, satisfying GDPR requirements. ChatGPT Enterprise's data ingestion model makes this significantly more complex.

Microsoft Copilot represents an even more direct threat. Built into Microsoft 365, Copilot integrates with Outlook, Teams, SharePoint, and the entire Office suite, giving it distribution advantages Glean can't match. For the approximately 345 million paid Microsoft 365 seats globally, Copilot offers "good enough" enterprise search without requiring a separate purchase decision.

Yet Glean customers report that Copilot's Microsoft-only focus creates gaps. Most enterprises use Google Workspace alongside Microsoft 365, Slack alongside Teams, and numerous SaaS applications outside Microsoft's ecosystem. Copilot can't search Salesforce, Jira, GitHub, or the dozens of other critical enterprise systems where knowledge lives. This multi-platform reality gives Glean an opening despite Microsoft's distribution power.

Google's enterprise search offerings—Google Cloud Search and its integration with Workspace—suffer similar limitations. Google Cloud Search works well within Google Workspace but poorly with Microsoft, Salesforce, and third-party applications. Google's enterprise focus has historically been weak compared to its consumer dominance, allowing Glean to compete effectively despite Google's search expertise.

Perplexity, the AI search startup valued at $14 billion following its June 2025 funding round, represents a different competitive angle. In December 2024, Perplexity acquired Carbon, a retrieval engine that connects external data sources to large language models. This acquisition positioned Perplexity to offer enterprise search alongside its consumer product.

Perplexity Enterprise Pro, launched in 2024 at $40 per user per month, combines conversational AI with dual-source search (public web data and internal documents). The product leverages GPT-4, Claude 3, and proprietary models to answer questions with citations from both internet sources and company data.

However, Perplexity approaches enterprise search as an extension of consumer search—web search that can also query internal documents. Glean inverts this model: enterprise knowledge first, with optional web search for context. For organizations where internal knowledge is more valuable than external information (most enterprises), Glean's approach delivers better relevance and personalization.

Specialized competitors like GoSearch, Coveo, and Elasticsearch target similar use cases to Glean but with different architectural choices. GoSearch focuses on startup and mid-market customers with simpler needs and lower price points ($15-25 per user per month). Coveo emphasizes customer-facing search (ecommerce, support portals) alongside employee search, spreading its product focus across use cases. Elasticsearch requires significant technical implementation and primarily serves developers rather than knowledge workers.

None of these competitors have matched Glean's Enterprise Knowledge Graph sophistication or personalization capabilities, according to analyst reports and customer evaluations. This technical differentiation, combined with Glean's product-led growth motion and enterprise sales execution, has allowed the company to maintain competitive positioning despite facing some of tech's most formidable companies.

The question for Glean is whether its lead is sustainable. OpenAI, Microsoft, and Google have effectively infinite capital and distribution reach. Perplexity raised more than Glean at a higher valuation. The enterprise search market is large enough for multiple winners, but being the independent player competing against platform providers carries existential risks.

The Agent Future

Glean's February 2025 launch of Glean Agents represents a strategic pivot from passive search to autonomous AI—a shift Jain described as "the biggest product evolution since we started the company" in a Fortune interview.

Glean Agents allows customers to build custom AI agents without coding. A sales operations team can create an agent that monitors Salesforce for deals that haven't been updated in 14 days, automatically sends reminders to account executives, and escalates to sales management if no action occurs within 48 hours. An engineering team can build an agent that monitors GitHub for pull requests awaiting review, identifies the best reviewers based on code ownership and availability, and sends targeted Slack notifications.

The agents run continuously in the background, checking data sources based on defined schedules or event triggers. They leverage the Enterprise Knowledge Graph to understand context—who owns which accounts, which engineers have expertise in which codebases, which documents relate to which projects—and take actions based on that understanding.

Glean claims customers are on pace to execute one billion agent actions by the end of 2025, less than 12 months after the agents' launch. This would represent an extraordinary adoption rate—billions of automated workflows executing based on Glean's intelligence layer.

The agent strategy positions Glean to capture more value from enterprise customers. A company using Glean for search might pay $25 per user per month for 1,000 employees, generating $300,000 in annual revenue. If that same company builds 50 agents automating workflows across sales, engineering, operations, and support—with each agent processing hundreds of actions daily—Glean can justify consumption-based pricing atop seat-based fees.

Agent pricing reportedly starts at $1,000 per agent per month for active agents processing significant action volumes, with volume discounts for customers deploying dozens of agents. This pricing model could generate millions in additional revenue from large customers who build comprehensive agent ecosystems.

The strategic bet is that enterprise AI will evolve from "ask AI questions" (ChatGPT, Claude, Gemini) to "AI takes actions" (agents, automation, workflows). Glean's Knowledge Graph provides the understanding necessary for agents to act appropriately—knowing not just what data exists but what it means, who it's relevant to, and what actions make sense in context.

However, agents also introduce new risks. Autonomous AI taking actions on behalf of humans creates accountability challenges: who's responsible if an agent sends incorrect information, escalates inappropriately, or takes an action based on misunderstood context? Glean addresses this through audit logs (tracking every agent action), approval workflows (requiring human confirmation for high-stakes actions), and scoped permissions (limiting what agents can do).

The agent market is crowded. UiPath, the robotic process automation company valued at $12 billion at its 2021 IPO, has pivoted to AI-powered agents. Microsoft offers Power Automate for workflow automation. Zapier, Workato, and dozens of integration platforms enable similar capabilities. Glean differentiates through its knowledge understanding—agents that know your organization, not just API endpoints.

But the agent future also exposes Glean to a deeper threat: if foundation model providers like OpenAI and Anthropic build robust enterprise knowledge understanding into their platforms, they could replicate Glean's core value proposition. The Knowledge Graph is defensible today, but if GPT-5 or Claude 4 develops native organizational understanding through extended context windows and multimodal reasoning, Glean's moat narrows.

The Enterprise Software Endgame

Glean's trajectory illuminates broader trends in enterprise software's evolution under AI pressure. Three dynamics stand out: the death of seat-based pricing, the collapse of product categories, and the migration of Google talent.

Traditional enterprise software charged per seat—$100 per user per month for Salesforce, $50 per user per month for Slack, $30 per user per month for various productivity tools. This model assumed linear value: twice as many employees meant twice as much revenue. AI breaks this assumption. An AI agent serving 1,000 employees might deliver more value than 10 human analysts, but it's just one "seat" consuming API tokens.

Glean's hybrid model—seats for search, consumption for agents—represents a transitional pricing structure. The company charges for human users accessing search and AI assistant features, but adds consumption-based fees for agents automating workflows. This allows Glean to capture value as automation replaces human labor without alienating customers who expect seat-based transparency.

Over time, consumption-based pricing will likely dominate enterprise AI, with customers paying for outcomes (queries answered, workflows automated, decisions supported) rather than seats. Glean's early move toward hybrid pricing positions the company for this transition, but traditional seat-based SaaS companies (Salesforce, Workday, ServiceNow) face business model disruption as their per-seat economics erode.

The second dynamic—category collapse—threatens every vertical SaaS company. Historically, enterprises bought specialized point solutions: Jira for project management, Confluence for documentation, GitHub for code hosting, Slack for communication. Each category generated billions in revenue for dominant players.

AI-powered knowledge platforms like Glean collapse these categories. If Glean can answer "What's the status of the authentication project?"—pulling data from Jira tickets, GitHub commits, Slack discussions, and Confluence docs—why do users need to navigate four separate applications? The knowledge layer becomes the interface; underlying systems become data stores.

This dynamic favors horizontal AI platforms (Glean, Microsoft Copilot, Google Workspace AI) over vertical SaaS. Point solution providers must either build compelling AI layers themselves or risk becoming commoditized backends invisible to end users. Atlassian (Jira, Confluence), Salesforce, and other vertical SaaS giants face existential pressure to develop AI experiences competitive with horizontal platforms.

The third dynamic—Google talent migration—explains why so many enterprise AI companies trace lineage to Google Search. Arvind Jain (Glean), Aravind Srinivas (Perplexity), and dozens of other enterprise AI founders spent formative years building Google's search infrastructure. They internalized lessons about ranking, relevance, indexing, and infrastructure that don't exist in textbooks.

This knowledge transfer from Google to startups mirrors previous tech generations: Microsoft alumni founding cloud companies in the 2000s (Amazon AWS led by former Microsoft managers), Facebook alumni founding social and mobile companies in the 2010s (Instagram, WhatsApp). The 2020s will be remembered as the era when Google's search expertise dispersed across enterprise AI, applying consumer search lessons to corporate knowledge.

For Google, this talent exodus represents strategic loss. The company trained the engineers now building competitive search products in enterprise markets Google never dominated. Glean, Perplexity, and other search-adjacent AI companies benefit from Google's R&D investment without contributing to Google's bottom line.

The $7.2 Billion Question

Is Glean worth $7.2 billion? The valuation implies aggressive expectations: investors are betting Glean will reach multi-billion-dollar revenue within 5-7 years, defend margins against competition, and either IPO at a premium or get acquired by a strategic buyer at a significant markup.

The bullish case is straightforward. Enterprise knowledge management represents a total addressable market exceeding $100 billion annually (combining enterprise search, productivity tools, collaboration software, and workflow automation). If Glean captures even 3-5% of this market, it generates $3-5 billion in revenue. At 10x revenue multiples typical for high-growth SaaS, that supports a $30-50 billion valuation—4-7x higher than today's $7.2 billion.

Customer adoption supports this narrative. Glean's $100M ARR in three years, 130%+ net revenue retention, and expansion from tech startups to global enterprises demonstrate product-market fit. The company's differentiation—Enterprise Knowledge Graph, permission-aware architecture, personalization—creates defensible moats competitors struggle to replicate quickly.

Agent adoption could accelerate revenue growth. If Glean's one billion agent actions by end of 2025 translate to hundreds of millions in incremental revenue (at $1,000+ per active agent monthly), the business could reach $200-300M ARR by fiscal 2026, supporting aggressive valuation multiples.

The bearish case centers on competitive threats and market structure uncertainty. Microsoft Copilot's distribution through 345 million Microsoft 365 seats gives it overwhelming reach. Even if Copilot delivers inferior search quality, "good enough" integrated into existing workflows beats "excellent" requiring separate adoption. Glean's independent positioning becomes a liability if enterprises consolidate AI spending with platform providers.

OpenAI and Anthropic represent wildcards. Both companies have effectively unlimited capital, foundation model advantages, and ambitions to serve enterprise customers. If they solve data governance, permission management, and organizational understanding—challenges Glean already solved—they could offer enterprise search as a feature of ChatGPT Enterprise or Claude for Work, undercutting Glean's pricing through bundling.

Acquisition risk also clouds valuation. At $7.2 billion, Glean is buyable by Microsoft, Google, Salesforce, or even OpenAI (which has raised $20+ billion). A strategic acquisition at a 20-30% premium ($8.5-9.5 billion) would reward investors but cap upside compared to an independent path to $30-50 billion valuations.

Historical comparisons provide mixed signals. Slack sold to Salesforce for $27.7 billion at 30x revenue. Tableau sold to Salesforce for $15.7 billion at 15x revenue. GitHub sold to Microsoft for $7.5 billion at an estimated 25x revenue. These multiples suggest Glean's $7.2 billion valuation at ~50-70x revenue (depending on current run-rate assumptions) is expensive but not absurd for a hyper-growth enterprise AI company.

The Path Forward

Glean's next 12-18 months will determine whether the company achieves its potential or succumbs to competitive pressure. Three initiatives will be decisive: international expansion, vertical specialization, and M&A.

International expansion offers significant growth. Glean currently derives approximately 75% of revenue from the United States, leaving European, Asia-Pacific, and emerging markets largely untapped. European data sovereignty requirements (GDPR, digital sovereignty initiatives) favor Glean's architecture over US-centric cloud providers. Expansion into financial services, healthcare, and government sectors—where data governance concerns are acute—could unlock billions in TAM.

Vertical specialization could deepen moats. Glean currently offers horizontal search across all industries. Building industry-specific knowledge graphs—understanding healthcare workflows, financial services regulations, manufacturing processes—would create switching costs and justify premium pricing. A healthcare-specific Glean understanding HIPAA compliance, clinical terminology, and hospital workflows would be difficult for horizontal competitors to replicate.

M&A could accelerate capability development. Glean's $7.2 billion valuation and $765M in funding provide currency for acquisitions. Buying workflow automation platforms, data integration startups, or vertical SaaS companies could expand Glean's footprint. Potential targets include integration platforms like Workato or MuleSoft (if Salesforce divests), vertical collaboration tools, or even struggling enterprise search competitors like Coveo.

The ultimate question is whether Glean can maintain independence. History suggests most enterprise software companies eventually get acquired or go public—staying private indefinitely is rare. Glean's path likely leads to one of three outcomes: IPO at $15-20 billion valuation in 2026-2027, acquisition by Microsoft/Google/Salesforce at $10-15 billion, or continued private growth toward a $30-50 billion private valuation before eventual public exit.

For Arvind Jain, who spent a decade building Google's search infrastructure before founding two unicorns, the opportunity to reshape enterprise knowledge management represents a career-defining challenge. If Glean succeeds, it will validate his core thesis: that enterprise search required a fundamental architectural rethinking, not incremental improvements on failed approaches.

And if Sam Altman's warning to avoid investing in Glean serves as any indication, OpenAI's CEO already considers that thesis validated—and the competition serious.