How Genspark Grew to $200M ARR in 11 Months After Abandoning 5M Users

Genspark hit $10M ARR in nine days — the fastest in AI history — but only after killing its first product and walking away from 5M users. The forced-pivot playbook, told honestly.

TL;DR

• Genspark's Super Agent reached $10M ARR in nine days — the fastest commercial curve in AI product history at the time — and the company crossed a $200M+ run rate within 11 months of that launch.
• The growth came after a brutal pivot. Genspark spent nine months as an AI search engine, reached 5M+ registered users with almost no disclosed revenue, then abandoned that product to launch an agent — riding a competitor's viral wave three weeks late, on purpose.
• The arc has no latent period. Day 0 was a $60M pre-revenue seed and a TechCrunch headline, engineered around founder Eric Jing's pre-existing reputation in Chinese tech — the company started the clock at its peak attention moment.

The Numbers

The $10M and $22M figures are media-reported (AISecret, VentureBeat); $36M, $50M, $100M and the $200M+ run rate are company-disclosed. For comparison, Lovable took two months to $10M ARR and the next-fastest dev-tools peers took six — Genspark did it in nine days with a 20-person team and zero paid acquisition.

What they did differently

Move 1: Fire everything into Day 0 when Day 0 is your peak

Genspark inverts the standard SaaS launch pattern. The conventional curve has a long quiet period before a catalyst kicks in vertical takeoff. Genspark had no latent period at all.

On June 18, 2024, Genspark and its Sparkpages search product went live. The same morning, TechCrunch and SiliconANGLE broke a $60M pre-revenue seed at a $260M post-money valuation, led by Lanchi Ventures. Product, funding and founder narrative all fired in a single news cycle. The structural reason it worked: the press coverage was pre-loaded. TechCrunch's hook was not a hot product or a hot revenue number — it was "former Baidu VP launches a Perplexity-Wikipedia hybrid with $60M from US and Singapore VCs." Founder Eric Jing's reputation — former Baidu VP, Xiaodu CEO, early Microsoft Xiaoice technical lead, founding Bing Asia search engineer — made the round coverable as news on its own.

The hidden cost of this trade is real, and worth saying plainly: when Day 0 is your peak attention moment, you start the clock at the top. Every quarter without a second event, attention decays. Genspark got in months 18 through 24 what a normal company gets — except it got it on Day 1, and then had to manufacture the next event before the first one faded.

Move 2: Make the pivot the story instead of hiding it

From June 2024 to early 2025, Sparkpages reached 5M+ cumulative registered users. The valuation re-priced 5.3x in eight months on a still-thin revenue base. But almost no ARR data exists for this period — the most plausible read of the silence is that Sparkpages' monetization was weak, the same trap Perplexity was struggling with at the same time.

Then in March 2025, Manus — a different China-origin agentic AI product — went viral on X. The "give an AI a task and it executes for hours" paradigm exploded. On April 2, 2025, Genspark launched Super Agent: a mixture-of-agents architecture with 8–9 LLMs, 80+ tools, and a killer feature Manus did not have — "AI Call For Me," where the agent dials real phone numbers. Genspark did not bury what this meant for the old product. It publicly redirected Sparkpages into the new Workspace experience, writing off 5M registered users in the open.

"20 people, 45 days, $36M ARR" — the number triple Wen Sang now repeats in every podcast appearance. The team-size-over-revenue framing is the move: a memorable team-efficiency claim travels further than any standalone ARR number.

— From the Genspark growth teardown

Most companies hide a pivot of this scale. Genspark turned "5M users walked away" into "we had the conviction to make the right call." Reframing the abandonment as conviction is what kept the pivot from reading as failure.

Move 3: Ride a competitor's viral wave on purpose, three weeks late

Genspark's Super Agent launch was not a coincidence of timing. It was wave-riding, executed deliberately.

Manus had spent three weeks educating the market on what a general AI agent is. Genspark walked into a category that no longer needed explaining and shipped the comparable product with the missing feature. VentureBeat published the "ups the ante in the general AI agent race" framing the same day — and from then on, every subsequent piece treated Genspark and Manus as a paired story. On YouTube, Julian Goldie and AI-commentary channels cut "Genspark vs Manus" demos that landed 100K+ views each, and the "AI Call For Me" phone-dialing scene became the screenshot of the era. Eric Jing's @ericjing_ai account, dormant since November 2023, reactivated around the launch with a "Seeing AGI" essay series — founder IP activated exactly when the second product needed a face.

The bundled-milestone discipline ran the same way at scale. November 20, 2025 was the single highest-density media day of the arc: the $275M Series B, the AI Workspace launch, and a Microsoft Agent 365 launch-partner announcement — three announcements on one day, where the typical AI-era bundle is two. Bloomberg broke the unicorn news nine days early, priming the official announcement and multiplying the surface of the same budget.

What you can copy

1. When Day 0 will be your peak, fire all your assets at once. If your founder pedigree, funding and product can all be news on the same day, do not atomize them across separate announcements — that wastes the press window.
2. Make pivots legible instead of burying them. When a product fails, reframe the abandonment as conviction. A public, narrated pivot reads as decisiveness; a quiet one reads as a cover-up when it surfaces later.
3. Treat a competitor's viral moment as a calendar item, not an emergency. Three weeks is enough to ship a comparable product with a missing feature. Do not try to be early — be three weeks late and one feature better, into a category someone else already explained.
4. Lead with a killer demo, not a roadmap. "AI Call For Me" — the agent dials real phones — did more for the brand than any benchmark number. Pick the one before/after a creator can record in 60 seconds.
5. Stack three milestones in one window. Funding plus product launch plus a strategic partnership on the same day compounds, because each story validates the others.

What probably won't work for you

The playbook is reusable, but four preconditions kept it from being available to most teams.

You need a founder with regional pre-existing reputation. Eric Jing's Xiaoice and Xiaodu history — credentials the press recognizes by name — is what made a $60M pre-revenue seed coverable as news. Without that, Day 0 simply does not fire, and the engineered-visibility move is unavailable.

The pre-revenue mega-seed window has narrowed. $60M pre-revenue rounds were available to a narrow band of AI founders in 2024. That window has tightened since — the capital that made Genspark's Day 0 possible is harder to access now.

You need a live competitor's viral wave to ride. Manus's March 2025 X moment was the runway. The Super Agent launch sequence does not work without an existing wave to step into — you cannot manufacture the category education a competitor did for you.

Three-region capital and customer access is hard to replicate. Genspark raised from US, Korean (LG Technology Ventures) and Japanese (SBI Investment) strategics in the same round, with Japan as its #1 paid market. A pure-US competitor cannot replicate this structure on demand. And the honest limits matter: Super Agent's $10M-in-nine-days is a top-of-funnel triumph — six-month and 12-month retention are undisclosed, the GAIA 87.8% benchmark is self-reported with no independent reproduction, and Reddit threads show real credit-billing and refund friction. Copy the launch moves; do not assume the curve held up as cleanly as the headline.

Sources & references

• Genspark is the latest attempt at an AI-powered search engine — TechCrunch
• AI search engine startup Genspark raises $100M at $530M valuation — SiliconANGLE
• Genspark's Super Agent ups the ante in the general AI agent race — VentureBeat
• Record-breaking launch: Genspark Super Agent reaches $10M ARR — AISecret
• Genspark becomes newest AI unicorn after winning LG, SBI funding — Bloomberg
• Genspark raises $275M Series B, launches AI Workspace — BusinessWire
• Genspark launches AI Workspace 2.0 as it crosses $100M ARR — BusinessWire
• Full timeline: The Genspark Growth Story — every round, pivot, and milestone, charted

This case study is part of GrowthHunt's growth teardown series. For another AI rocket built on bundled rounds, read the Lovable teardown; for the dev-tools end of the curve, the Cursor teardown. Track the fastest-growing AI founders live on GrowthHunt Velocity.

More from the growth playbook.

See it run in the wild.