When Netflix Stopped Trying to Make Anime and Actually Made Anime
Put this on while you read
November 19, 2021. Netflix drops its live-action Cowboy Bebop adaptation. They’ve spent somewhere between $60 and $70 million on it. Ten episodes. One of the most beloved anime of all time as the source material.
Shinichirō Watanabe, the director who made the original, watched one scene and stopped. His comment, when it got out: “It’s Clearly Not Cowboy Bebop.”
The show was cancelled on December 9. Liner Note The live-action Cowboy Bebop aired November 19, 2021 and was cancelled December 9, 2021. It dropped 59% in viewership between its first and second week. Critics score on Rotten Tomatoes: 46%. Budget: estimated $60-70M for 10 episodes. Source ↗
Netflix put seven million dollars per episode into a property, handed the original director a cameo credit and nothing else, built something that felt exactly like what happens when a corporation decides it understands a thing without understanding why the thing works. Then cancelled it before the month was over.
This was the Netflix anime strategy: volume, IP, speed. Chase what people already love, adapt it broadly so it doesn’t alienate anyone, ship it fast and move on. Death Note in 2017, Ghost in the Shell the same year, Cowboy Bebop in 2021. The pattern was identical every time: acquire beloved property, sand down the edges, discover the edges were the whole thing.
By 2022, anime producers had a phrase for the Netflix treatment. None of it was polite.
The Different Bet
In June 2020, while the Cowboy Bebop production was still underway, CD Projekt RED announced a partnership with Netflix and Studio Trigger on a standalone anime set in the Cyberpunk 2077 universe.
Studio Trigger was founded in 2011 by Hiroyuki Imaishi and Masahiko Otsuka after they left Gainax to build something smaller and more deliberately strange. Their debut original was Kill la Kill (2013): a show about a girl fighting a sentient sailor uniform that controlled the school. Their reputation was built entirely on swinging for specific, personal visions instead of broad commercial appeal. Imaishi had directed Gurren Lagann before founding Trigger, the show that defined a generation of mecha anime by refusing to be modest about anything. This is not a studio you hire when you want something safe.
Director: Hiroyuki Imaishi. The show would be called Cyberpunk: Edgerunners. Ten episodes. Netflix would let Trigger make it.
That last part is the part that matters.
There’s no public document listing what Netflix gave up. No press release about creative autonomy. What’s documented is the outcome: Imaishi made the show he wanted to make, in the visual language Trigger had spent a decade building, with an ending that nobody at a streaming platform should have cleared if they were optimizing for renewal. They cleared it.
The show came out September 12, 2022.
What Happened
Liner Note Cyberpunk: Edgerunners hit 14.88 million hours viewed in its first week on Netflix, ranking top 5 in over 30 countries. It holds a 100% critics score on Rotten Tomatoes, one of the highest-rated Netflix originals in any category. Source ↗ Number one in over thirty countries. 100% on Rotten Tomatoes from critics.
Then the game numbers came in.
Cyberpunk 2077 had launched in December 2020 as one of the more spectacular disasters in recent gaming history: broken on consoles, pulled from the PlayStation store, a running joke about how badly you could miss your own release. By 2022 it had been patched into something functional. Steam players were hovering around 10,000 to 15,000 a month. Respectable for a two-year-old game.
The week Edgerunners dropped, concurrent Steam players peaked at Liner Note Steam concurrent player data via SteamDB shows Cyberpunk 2077 peaking at 136,724 players on September 25, 2022, up from a monthly average of roughly 10,000-15,000 before the anime. Daily active players hit one million during peak week. CD Projekt revenue rose 70-96% in the quarter. Source ↗ . Daily players hit one million during peak week. CD Projekt revenue was up somewhere between 70 and 96 percent in the quarter. The game crossed 20 million copies sold.
A ten-episode anime series revived a dying $60 game better than two years of patches and marketing.
The Ending They Didn’t Change
David Martinez dies in the finale. Rebecca gets killed two scenes earlier. Lucy survives, alone on the Moon, replaying a fake memory of a beach date with someone who is gone.
No rescue. No loophole in the show’s internal logic. Imaishi spent ten episodes making David and Rebecca people worth caring about, then used that investment exactly the way Fujimoto used Aki Hayakawa: as a weapon against the audience’s assumption that caring about a character means the story will protect them.
Netflix announced Season 2 after the viewership numbers came in. New story, new characters. David is not coming back.
This is where the product lesson lives. The easy read is “Netflix learned to trust creators,” which is true but incomplete. The full read is that Netflix made a specific bet: not a philosophy shift, not a rebrand. One show. They made it by selecting the right studio, then getting out far enough that the team could kill the protagonist without routing that decision through a committee.
Most streaming platforms would have flagged that ending. Somebody in a meeting would have asked whether David could survive in a really damaged way, some condition that preserves the possibility of return. The note would have come from genuine concern about renewal viability, not cynicism. That’s the dangerous version of interference: it sounds reasonable. Imaishi’s version got through.
The reason Edgerunners hit 14.88 million hours in week one is the same reason it revived the game. People finished it in a sitting because they couldn’t look away, then told everyone they knew because the ending meant something. That emotional weight comes entirely from the creative decisions that looked risky: the character deaths, the unresolved ending, the refusal to leave a sequel hook. Those decisions required someone to not send the note.
The Handoff Problem
The Cowboy Bebop failure wasn’t a budget problem or a talent problem. It was a delegation problem. Netflix acquired an IP, formed a creative team, stayed close enough to ensure the result felt like Netflix. The result felt like Netflix. Nobody wanted that.
Edgerunners worked because the delegation was real. Real delegation means the team can make decisions you would not have made. It means the ending surprises you. It means if you find out at the last minute that the protagonist dies and the show doesn’t set up a sequel, you don’t send the note. Because the note is about your comfort, not the product.
Most PMs understand this in theory. The failure mode is all the ways you can believe you’ve delegated while actually maintaining control: approval gates on every decision, reframing check-ins as “support,” asking to see early cuts for “context.” Each one is reasonable on its own. Together they reconstruct the original relationship and the team stops making the calls that make the product worth caring about.
The Handoff Test
Delegation that doesn’t change the outcome isn’t delegation. Before handing a bet to a team, run these.
Are you hiring for vision or execution? If you already have the vision and they’re building to spec, you haven’t delegated. You’ve outsourced. Real delegation means the team can see things you can’t and make calls you wouldn’t have made. Netflix handing Trigger a universe and stepping back produced Edgerunners. Netflix hiring animators to execute a Netflix concept would have produced something else.
Can they change the ending? This is the concrete test. If every major decision still routes through you for approval, the team is executing under constraints, not leading. Imaishi could kill David. That’s what real handoff looks like. If your team can’t ship a decision you’d personally regret without your sign-off, you haven’t handed it off.
What does failure look like for them versus for you? If the team defines success by the quality of what ships and you define it by the quarterly engagement number, the first time those diverge you’ll override them. Align on what winning means before the work starts. Mismatched failure definitions guarantee interference at exactly the moment it does the most damage.
Are you measuring the output or the process? Asking to see early work is a form of control. Sometimes it’s necessary. But if you’re watching how the team works rather than what they ship, you’ve already reclaimed the authority you said you were giving up. Trust is established in the brief, not rebuilt from sprint reviews.
Netflix spent years and considerable money proving that volume and IP acquisition are not the same as making something people love. Then they made one bet differently: the right studio, the actual authority to make their show, an ending nobody would have recommended from a business case perspective.
The business outcome was downstream of the creative decision that looked like risk. It almost always is.