Parag Agrawal's Parallel Web Systems hits $2B valuation as Sequoia leads $100M Series B for AI agent infrastructure
Palo Alto-based Parallel Web Systems, the AI infrastructure company founded by former Twitter chief executive Parag Agrawal, has raised $100 million in a Series B round led by Sequoia Capital at a $2 billion post-money valuation, the company announced on April 29.
The round more than doubles Parallel's $740 million valuation from its $100 million Series A in November 2025, taking the total raised to $230 million since the company's first $30 million seed in January 2024. Sequoia partner Andrew Reed will join the board as part of the deal. Every existing investor doubled down, including Kleiner Perkins, Index Ventures, Khosla Ventures, First Round Capital, Spark Capital, Terrain Capital and Abstract Ventures.
Parallel sells APIs and a proprietary web index purpose-built for "machine retrieval", an alternative to Google-style search results that are ranked and rendered for human readers. Where a conventional search engine returns a list of links, Parallel returns structured, grounded outputs that drop directly into an AI model's context window, which the company says reduces hallucination and operational cost for enterprise deployments. Its product suite spans a Search API, a Task API for open-ended autonomous research, a Monitor API for tracking changes across the live web, and an Extract API. The company maintains SOC-2 Type II certification and offers zero-data-retention deployments for regulated buyers.
In a post on X announcing the round, Agrawal said the bet underpinning Parallel was that "agents will use the web a thousand times more than humans ever have, and that most of that work will happen in the background." A second image he shared, styled as a logical map of the agentic web modelled on the 1977 ARPANET diagram, carried the line: "Agents are here. The web's second user is online."
Reed, in a statement accompanying the announcement, said long-horizon agents were beginning to redefine products across every industry, and that the ability to use the web was a shared core function across all of them. "The best AI teams around the world are choosing Parallel," he said, "and we're excited to partner with the Parallel team as they build the company that invents the future of web infrastructure for AI."
Customer momentum has run ahead of the pace usually expected at this stage. More than 100,000 developers are building on the platform, and Parallel's enterprise roster spans both AI-native firms and Fortune 500 incumbents. Legal AI company Harvey uses Parallel to ground its reasoning in public legal documents across more than 60 jurisdictions, having selected the infrastructure after concluding that conventional search engines could not provide the granular site-level control required for agent-grade legal research. Notion's AI agents run background research, analysis and stakeholder follow-ups for millions of users. Opendoor has cut its automated HOA research workflow from roughly 10 minutes of manual work per property to about 2 minutes of verification. Two of the largest US property and casualty insurers, working through a partnership with Genpact, have reduced claims processing times by 50%, with 55% of cases now flowing touchlessly. Other named customers include AI-native CRM Attio, marketing platform Profound (which itself recently hit unicorn status with a $96 million round at a $1 billion valuation), serverless cloud Modal, financial intelligence platform Rogo, and public-sector sales platform Starbridge.
Harvey co-founder and president Gabe Pereyra, quoted by the Wall Street Journal, said it was no longer enough to give AI agents access to Google Search and that Parallel provided the more granular control required for production legal workflows.
The investor thesis is that the next infrastructure category in AI will not be models or compute, but the connective tissue that lets autonomous systems read, retrieve and act on the open internet without breaking. First Round Capital partner Josh Kopelman, who has backed Parallel since seed, described the company's premise as "a completely different infrastructure layer, built from scratch, for a customer the web never had before." He also offered rare colour on Agrawal's path between Twitter and Parallel, writing that the founder "had no shortage of options" after leaving the social network but "passed on all of them," instead spending months trying to identify a problem worth committing a decade of his life to.
Bloomberg has previously reported that during this period, Agrawal was reading research papers at a Blue Bottle coffee shop in downtown Palo Alto and slowly returning to writing code, telling the publication: "I'm not a person that can enjoy the beach in that moment."
Agrawal's biography sharpens the bet. He joined Twitter in 2011 as a software engineer after completing a PhD in computer science at Stanford under Jennifer Widom, having earlier graduated from IIT Bombay with an All India Rank 77 in the IIT-JEE entrance examination. He rose to chief technology officer in 2017, where he led the recommendation algorithms behind Twitter's home timeline, and was named chief executive in November 2021. His tenure ended abruptly the following year when Elon Musk acquired the platform for $44 billion in October 2022 and dismissed senior leadership on day one.
Agrawal was denied severance and joined other former executives in a multi-year lawsuit against Musk over roughly $200 million in unpaid stock vesting, a case that survived Musk's motion to dismiss in late 2024. He co-founded Parallel with Travers Nisbet in 2023, with the company emerging publicly in 2024 and launching its first commercial APIs in late 2025.
Parallel's pitch lands at a moment when investor appetite for agentic infrastructure has sharpened decisively. AI venture funding reached roughly $242 billion in the first quarter of 2026, and the global AI agents market, valued at $5.4 billion in 2024, is projected to grow to more than $139 billion by 2033, according to MarketsandMarkets, with a compound annual growth rate of close to 44%. The agentic stack, the layer between foundation models and real-world workflows, is where much of the new capital is converging.
The competitive set is consolidating quickly around it. Exa Labs, the closest analogue to Parallel, raised $400 million at a $2.4 billion valuation in early 2026, backed by Nvidia, putting it slightly ahead of Parallel on paper but with a different focus on semantic search and dedicated indexes for people and company data. Tavily, the Israeli search-API startup that competed in the same category, was acquired by Amsterdam-based Nebius in February 2026 in a deal reported at $275 million and rising to as much as $400 million on milestones, after raising just $25 million in venture funding.
Smaller specialists Firecrawl and Brave Search API are competing for developer mindshare with bundled crawl-and-search products and independent indexes, respectively. Perplexity continues to build a consumer-facing product alongside its Sonar API for developers. Hovering over the entire category is the prospect that OpenAI, Anthropic and Google could internalise web search for their own agents, a risk Sequoia and Parallel's other backers appear comfortable underwriting on the bet that specialist infrastructure will outperform in-house solutions, much as Stripe did against bank-built payment APIs and Twilio against telco-built communications stacks.
Parallel's longer-term ambition extends beyond search APIs. Agrawal has spoken publicly about building market mechanisms that give publishers and data owners a direct economic stake in AI's use of their content, an economic model intended to keep the open web commercially viable as agents become its dominant traffic source. "How many jobs are there where we could turn off web access and ask you to do the same job fully?" Agrawal told Reuters at the time of the Series A. "You can't deprive an M&A lawyer from not being able to use the web, why would you deprive their agents?" A growing share of websites is deploying measures to block AI scrapers and gate content behind logins or paywalls, and Parallel has positioned itself as a broker between content owners and AI systems rather than another extraction layer.
The new capital will be deployed into expanding the web index, scaling enterprise sales and marketing under recently appointed go-to-market lead Graham Moreno, deepening partnerships including a recently announced strategic alliance with Genpact, and continuing investment in research and development.
For Sequoia, the investment is one of the firmest commitments yet to the agentic infrastructure thesis. For Agrawal, who in Kopelman's account spent months reading research papers at a Palo Alto coffee shop before settling on Parallel as his next chapter, it is the clearest signal yet that the second act after Twitter has found its market.