What’s the key to maintaining brand integrity at scale? A lot goes into it, but the teams that get it right have one thing in common: they know how to talk to AI models.

As teams speed up and the required creative volume balloons, communication matters more than ever— both with LLMs, and with the people around you. Creative teams need to focus on ideas, concepts, craft, while technology augments their manual execution demands.

Join our discussion on this evolution with Ari Murray, Chief Digital Officer at Salt and Stone, and get 5 tips for expert LLM prompting with ready to run prompts.

Register and get prompting

👋 Hi, it’s Rohit Malhotra and welcome to the FREE edition of Partner Growth Newsletter, my weekly newsletter doing deep dives into the fastest-growing startups and S1 briefs. Subscribe to join readers who get Partner Growth delivered to their inbox every Wednesday morning.

If you’re new, not yet a subscriber, or just plain missed it, here are some of our recent editions.

_{Interested in sponsoring these emails? See our partnership options here.}

Subscribe to the Life Self Mastery podcast, which guides you on getting funding and allowing your business to grow rocketship. 

Previous guests include Guy Kawasaki, Brad Feld, James Clear, Nick Huber, Shu Nyatta and 350+ incredible guests.

Firecrawl raised a $75 million Series B led by Smash Capital, with participation from Altos Ventures, Nexus Venture Partners, Y Combinator, Freestyle and Offline Ventures, announced on September 22. The round's valuation was not disclosed. It comes 13 months after a $14.5 million Series A led by Nexus in August 2025. The company began by solving its own problem. While building Mendable, a docs chatbot, the team found that clean web data was the hardest layer of the stack. Today it serves 1.5M+ developers and 150,000+ companies including Shopify, Apple, Lovable and Canva.

On the same day, it launched Alexandria. This is one interface agents can use to find, inspect and pull from the live web, official data providers and Firecrawl's own indexes. Across 845 tasks, agents using it scored 21% higher on answer quality than agents using built-in web tools.

That launch is the tell. This is not a scraping round. It is a bet on who pays for knowledge.

Firecrawl already pays Wikimedia Enterprise for data. Part of the Series B will fund payments to researchers, publishers, experts, creators and other data providers whose material flows through Alexandria. Scraping is a commodity. A paid supply side is a moat.

$75M Series B, valuation undisclosed. The stack underneath it matters more: roughly $91M raised in total, with $16.2M of that before this round. A $14.5M Nexus-led Series A in August 2025. Now a round 5.2x larger, 13 months later, with Nexus and Y Combinator writing checks again.

Growth is the engine. Firecrawl grew 15x in the year before the Series A and was profitable when it closed. It hit eight figures of ARR in year one and more than doubled that in year two. Five months into 2026, it had more than double the paying customers it had in all of 2025.

The adoption is where the argument lives. 350,000 developers at the Series A. 1.5M+ today, a 4.3x jump in 13 months. GitHub stars went from 48,000 to 180,300. 150,000+ companies use it, including Shopify, Apple, Lovable and Canva. This is not a scraper riding agent demand. It is the default pipe agents already pull the web through, now pointed at paid sources through Alexandria.

The valuation is the part to sit with. There isn’t one. A company with these numbers that stays quiet on price is choosing which number tells the story. Investors are not paying for the scraper. They are paying for the toll booth on knowledge.

Smash Capital led the round. It is a growth-stage firm co-founded by Disney veterans including Kevin Mayer. Nexus and Y Combinator, both in the Series A, came back. Altos, Freestyle and Offline are new names. This round is not insiders doubling down. It is outsiders paying up to get in, with insiders protecting their position.

No lead investor has published a thesis yet. The numbers wrote it for them. Firecrawl was profitable at its Series A, which is rare for infrastructure this young. Its customers became its backers. Zapier integrated it in an afternoon and then invested. Shopify’s Tobi Lütke wrote a personal check. Nobody was underwriting a pitch. They were underwriting a product the buyers chose twice: once as users, once as investors.

So the 5.2x step up from the Series A is not exuberance. It is the price of entering after 1.5M developers already have.

The forward bet is bigger than scraping. Scraping risks becoming a feature that larger platforms absorb. Alexandria is the hedge. It moves Firecrawl from fetching the open web to paying for what is not on it, starting with Wikimedia Enterprise.

That is what $75M actually buys. Not the crawler Firecrawl started as. The licensing layer between knowledge and the agents that need it.

One. Timing that fixes itself. The free web is closing, and it is closing on its own. Cloudflare, which handles roughly 20% of web traffic, began default-blocking mixed-use AI crawlers on September 15, one week before this round. News and media sites block training crawlers at around 81%. OpenAI crawled 1,700 pages for every referral it sent back in 2025. Publishers are not leaving the web. They are putting a price on it. Every new block raises the value of a licensed route, and hands Firecrawl the wedge for free.

Two. The scraper ceiling, called. Scrapers were supposed to stay scrapers. Firecrawl started at URL-to-markdown, then moved into search, then into its own indexes covering research papers, developer docs and government law. It has 1.5M users. The bet is that the ceiling was never real, only untested. Alexandria is the test: one interface across the open web and the sources behind the wall.

Three. The bet that justifies $75M. Access is the entry point, not the destination. Paid-crawl models cover about 5% of top sites, up from about 1% in 2024. Firecrawl already pays Wikimedia Enterprise. Whoever owns agent-side demand gets to set the price on the supply side. Land the crawler, own the toll.

One. AI clouds have the motive and a fresh precedent. In February, Nebius agreed to buy Tavily, an agentic search API, for $275M, with earnouts that could take the deal to $400M. The logic was bundling: sell inference and web access as one workflow instead of two vendors. Firecrawl is the same layer with more surface area: crawl, extract, search, and now licensed data. The catch is price. Tavily went for roughly 98x its 2025 revenue. Firecrawl is bigger, still growing, and just raised $75M on its own terms. Every year it stays independent makes the check worse.

Two. Cloudflare owns the gate. It sits in front of about 20% of web traffic, blocks AI crawlers by default, and runs pay-per-crawl. Firecrawl owns the demand side. Together they would control both sides of the toll. But a company that sells itself as the publisher’s protector buying the biggest scraper is a hard story to tell. The likelier path is partnership: Firecrawl as the best-behaved paying crawler on Cloudflare’s rails.

Three. The real buyer is a model lab, if Alexandria works. OpenAI and Google already bundle search into their own platforms. Crawling they can build. Signed contracts with thousands of data providers they cannot. Strategics buy features. Platforms buy marketplaces.

One. That the revenue comes from depth, not just downloads. 1.5M users and 180,000 GitHub stars make a funnel, not a business. The filing has to show how many of those users pay, how fast free and self-hosted users convert, and net revenue retention. If agent companies that came for scraping now pay for search, indexes and Alexandria, the platform story holds. If revenue is mostly small credit packs, it is a developer tool wearing an infrastructure costume.

Two. That Alexandria is a business, not a cost centre. Paying Wikimedia and future providers puts payouts inside cost of revenue. The S-1 has to break out licensed-data revenue, the take rate, and what flows back to providers. A marketplace margin is not a SaaS margin, and the multiple depends on which one this is.

Three. The unit economics under the crawl. Every default block from Cloudflare raises the cost of a successful page through proxies, rendering, retries and paid access. Gross margin has to show that Firecrawl can absorb a more closed web without its costs rising faster than its prices.

Four. Concentration and dependency. Its biggest customers are AI platforms that could build this in-house, and labs already bundle search natively. The risk factors section will say how much revenue rides on companies that could become competitors, and how much scraping litigation it could face.

Firecrawl did not raise on a story. It raised on adoption it never had to buy: 1.5M developers who chose it because it fixed a problem they kept hitting. That is the part competitors cannot spin. When agent companies route their web access through you, you are not selling a scraper. You are becoming the pipe.

The lesson underneath is older than AI. Plumbing was supposed to be a small prize, too technical and too easy to copy. Firecrawl went where nobody was looking and found the supply chain for agent knowledge waiting.

So the $75M is not a bet on crawling. It is a bet on who pays for knowledge once the free web closes, made before Cloudflare or the labs set that price themselves. The risk is the same as the thesis: they still own the gate and the demand.

Firecrawl has the slope. The open question is whether Alexandria proves it is a marketplace before the platforms prove it is a feature.

Here is my interview with Sam Marchant, a Partner at 21 Ventures, backing pre-seed and seed founders across MENA and Southeast Asia with $500K–$1.5M cheques in AI, FinTech, and Industrials.

If you enjoyed our analysis, we’d very much appreciate you sharing with a friend.

Here are the options I have for us to work together. If any of them are interesting to you - hit me up!

And that’s it from me. See you next week.

What do you think about my bi-weekly Newsletter? Love it | Okay-ish | Stop it