At its core, paid growth was a form of traffic arbitrage. The principle was simple: as long as LTV exceeds CAC, you can invest indefinitely. At the top of the funnel sits impressions — when a brand buys impressions, it's buying user attention, which is to say, buying time. And user time online is finite. When arbitrage succeeds, you pour more into ads to capture more of that time, leaving less for competitors. Simple. Blunt. Effective.

I spent twelve years in paid user acquisition, from stat-heavy SLG games to the open-world Genshin Impact, I know this logic well and I used it to produce results across global markets that I'm proud of.

The funnel isn't failing because ads got more expensive or users got more selective. It's failing for a more fundamental reason: AI has radically compressed how quickly users can gather information.

Given the same one-week decision window, a user used to encounter maybe ten information sources. With AI assistance, they can now access thousands — or more. Users no longer wait passively for information to reach them. They can summon a near-complete picture of the market, across time and geography, on demand.

Ask an AI: "Recommend an open-world game with great visuals." It returns an answer. How much you spent on ads, how many impressions you bought — none of that has meaningful direct influence on that outcome. What does matter is whether the AI's understanding of your product is accurate, and whether there's enough information density to support a recommendation in the first place. That's the new battlefield.

To understand this new battlefield, you need to understand how AI perceives the world. It maps every concept — products, needs, attributes — into coordinates in a vector space. When a user expresses a need, the AI finds the product whose coordinates sit closest to that need.

Whether you get recalled depends on two things: whether your coordinates are positioned near the right need dimensions, and whether you stand out from competitors along those dimensions.

Standing out in vector space requires two things working together. First, signal accuracy: large volumes of content consistently reinforcing the same core dimensions, so that AI builds a clear, coherent, noise-free understanding of what you are. This is brand content's job. Second, signal strength: enough real-world content about you existing broadly enough in AI training data that your signal is harder to ignore than a competitor's. This is paid growth's job — by driving user scale, users generate authentic content, and authentic content amplifies signal strength.

This is why brand and performance must converge — not as a methodological preference, but as an objective requirement for building vector coordinates. They are two sides of the same thing. Running them in silos is building half a coordinate.

Genshin Impact launched in 2020. It didn't win because it outspent the market on ads. It won because from day one, every distribution action aligned around the same core dimensions: visual quality, open-world freedom, elemental combat depth, character design and narrative. These weren't just product features — they were the signal axis running through every touchpoint. Ad creative emphasized them. Launch events emphasized them. Community management emphasized them. Every piece of content pushed in the same direction. The result: signals were extremely accurate, and once paid growth amplified them, signals became extremely strong. Authentic content erupted globally in a very short window and sustained for years.

For Honkai: Star Rail, I ran an experiment: I opened Gemini and asked for recommendations for turn-based mobile games. The first result was Honkai: Star Rail. The AI's recommendation reason — without mentioning advertising, without citing acquisition data — characterized it as a model of industrial-scale content production, with stable and consistent narrative and map updates every version. miHoYo's sustained, high-density content output has formed a signal so clear that AI can accurately summarize it. Version after version, year after year, the signal grows more accurate and more powerful.

From Genshin to Star Rail, the methodology hasn't changed: identify the dimensions where you have an advantage, build accurate signals, use paid growth to amplify them. What's new is that in the AI era, this approach earns compounding returns it didn't before.

Paid growth isn't dead. But growth has undergone structural change at two levels.

First, the user ecosystem has shifted. The way users make decisions has been rebuilt. We now actively consult AI, let AI organize our requirements, let AI offer reference points. This means the growth battleground has moved — from competing for user attention to competing for prominence in AI's vector space.

Second, how practitioners work has changed. The old growth organization divided brand teams from performance teams, each with separate KPIs, separate data systems, and occasional alignment meetings. Under the new logic, that structure is a systemic disadvantage. Vector coordinates require signal accuracy and signal strength to coexist. Running them separately builds half a coordinate. Future growth organizations built on AI capabilities won't divide by brand and performance. The organizing unit will be the signal unit.

Each signal unit corresponds to a need dimension. To outperform competitors on that dimension, you need to do two things simultaneously around it: build accurate signals and amplify them. These two tasks cannot be handed to separate teams running separate logic. They have to operate in coordination under a unified framework.

What does this mean for practitioners? The growth professional of the future won't be a "paid specialist" or a "brand specialist." They'll be someone who can identify which dimensions are worth building signals around — and orchestrate content and media buying to work together around the same signal unit. That capability is considerably harder than optimizing CAC. But it's the growth capability that will actually matter in the AI era.

The real question isn't whether paid growth is dead. It's whether you've rethought what growth means now that AI is part of every user's decision process.