Parag Agrawal was the CEO of Twitter for exactly 13 months before Elon Musk fired him. That was 2022. Since then, he’s been building Parallel Web Systems, a startup focused on something most people haven’t thought about but every AI company desperately needs: web search infrastructure built specifically for AI agents. Last week, Parallel closed a $100 million Series A at a $740 million valuation, co-led by Kleiner Perkins and Index Ventures. Spark Capital, Khosla Ventures, and First Round Capital also participated.
The valuation is massive for a Series A. The bet VCs are making is even bigger: that the way AI accesses the internet is fundamentally broken and Parallel is the company that fixes it.
The Problem Parallel Is Solving
When you ask ChatGPT, Claude, or Gemini a question about something current, those models need to search the web. The problem is that today’s web search was designed for humans. Google returns ten blue links optimized for click-through rates. An AI agent doesn’t want ten links. It wants structured, verified, real-time data it can use immediately to complete a task.
Parallel builds APIs that let AI systems search the live web for up-to-date information. Their customers use these APIs to power AI agents that write code, analyze sales data, assess insurance risk, conduct research, and make decisions — all tasks that require fresh, accurate information from the internet. The infrastructure handles the messy work of crawling, indexing, extracting, and structuring web data in a format AI models can consume directly.
Think of it like this: Google is a restaurant where you read the menu yourself. Parallel is a kitchen that prepares exactly the ingredients an AI chef needs, already measured, cleaned, and ready to cook with.
Why VCs Are Paying $740M for an Infrastructure Company
The AI agent market is projected to reach $47 billion by 2028, according to Markets and Markets. Every one of those agents needs fresh data. Every one of them needs to access the web. The companies building agents — OpenAI, Anthropic, Google, and thousands of startups — don’t want to build web crawling infrastructure themselves. They want to buy it.
That’s Parallel’s position: the picks-and-shovels play for the AI agent gold rush. When Kleiner Perkins leads your Series A, the signal is clear — this is being treated as foundational infrastructure, not a feature.
The revenue model is API-based. Customers pay per query. As AI agents proliferate, query volume grows. It’s the same model that made Twilio and Stripe worth tens of billions — developer infrastructure that scales with usage.
The Open Web Problem: Parallel’s Biggest Challenge
Here’s where it gets complicated. Websites don’t want to be crawled by AI. Publishers, e-commerce sites, and content platforms are increasingly blocking AI crawlers because they take content without driving traffic. The New York Times sued OpenAI over this. Reddit charges for API access. The open web is closing, and Parallel needs it to stay open.
Agrawal’s answer is an “open market mechanism” — an economic model that gives websites financial incentive to keep their content accessible to AI. The details are thin, but the concept is essentially: instead of AI companies scraping content for free, they pay websites through Parallel’s marketplace. Websites that opt in get revenue. AI companies that use Parallel get access to data they can’t get otherwise.
Whether this works depends on whether the economics are compelling enough for publishers. Google pays Apple billions to be the default search engine on iPhones. If Parallel can create a similar value exchange between AI companies and content creators, it could become the economic backbone of AI-web interaction. If it can’t, it’s just another crawler fighting an increasingly hostile web.
Agrawal’s Advantage: He’s Been Here Before
Before becoming CEO, Agrawal was Twitter’s CTO. He joined the company in 2011 as the person responsible for scaling Twitter’s search and data infrastructure. He redesigned how Twitter indexed real-time content from hundreds of millions of users. The technical problem Parallel solves — making vast amounts of web data searchable and structured in real-time — is essentially the same problem at a different scale.
His CEO tenure at Twitter was short and largely overshadowed by the Musk acquisition. But the technical credibility is real. VCs are betting on the engineer, not the executive.
The Verdict
Parallel Web Systems is a bet that AI agents will need their own web infrastructure, separate from the human-optimized search engines we use today. The $100M raise and $740M valuation say the smart money agrees. The execution risk is real — building a sustainable content marketplace is harder than building the API — but the timing is right. Every major AI lab is building agents. Every agent needs real-time web data. Parallel is positioning itself as the plumbing. And if there’s one lesson from the last tech cycle, it’s that the plumbing companies outlast the flashy apps built on top of them.
