Sony finally got Kojima back to espionage, then cancelled it while its first-party strategy falls apart

Sony spent years trying to convince Hideo Kojima to return to the action-espionage genre that made him famous. Fans had spent even longer asking for basically the same thing. Kojima eventually changed his mind, PlayStation loudly promised its full support, and less than two years later Sony suddenly tried to kill the project. Xbox then picked it up.

You genuinely could not script a better summary of PlayStation's first-party management right now.

Sony asked for this:

The important context around Physint is that this wasn't Kojima begging Sony to fund some giant vanity project. PlayStation wanted this game.

When Sony and Kojima announced the project in January 2024, then-PlayStation Studios boss Hermen Hulst said Kojima was taking them on an adventure he had been encouraging him to reinvent for years. Kojima described the project as a completely new action-espionage IP built on almost 30 years of history with Sony, while Kojima Productions announced that it would be made with the “full support of Sony Interactive Entertainment”, using cutting-edge technology, a major cast and near-photoreal visuals. Kojima even called it the culmination of his career. (kojimaproductions.jp)

That distinction matters because Kojima had deliberately spent his independence doing almost everything except making another Metal Gear. After leaving Konami, he wanted to create his own IP and experiment. The result was Death Stranding, a bizarre AAA delivery game about rebuilding human connections through asynchronous multiplayer systems that practically nobody else with a blockbuster budget would have attempted. Then came Death Stranding 2, while OD took that experimentation in another direction entirely.

Love those games or hate them, they are at least recognisably the products of somebody trying weird shit rather than feeding another established franchise through an executive-approved sequel factory.

Kojima himself explained that he had resisted going backwards. After becoming independent, he wanted to make new things. But for roughly eight years, he said people around the world had been asking him every day to make another Metal Gear-style game. Then he became seriously ill during the pandemic, underwent surgery, wrote a will and started thinking rather differently about how much time he had left. If people wanted another action-espionage game from him that badly, he decided he should change his priorities and make one. (gamespot.com)

So everything lined up at once: fans wanted it, Kojima finally wanted to do it, Sony had spent years encouraging him to do it, and Sony publicly offered the technology and support to do it properly. The man who helped define the modern stealth/action-espionage genre was finally returning to it with a brand-new IP rather than another contractual Metal Gear sequel.

This should have been one of the easiest prestige projects for PlayStation to understand.

Sony certainly acted like it understood. As recently as September 2025, Hulst was still appearing at Kojima Productions' tenth-anniversary event to reiterate how excited PlayStation was, describing Physint as being fuelled by the same creative energy and ambition that had powered Sony and Kojima's relationship for almost 30 years. (videogameschronicle.com)

Nine months later, Sony cancelled it.

Then Sony suddenly wanted out:

According to Kojima, Sony unexpectedly informed Kojima Productions in mid-June 2026 that it was cancelling Physint. Kojima didn't decide to walk away and there was no public creative split. Sony pulled the plug, catching him sufficiently off guard that Kojima Productions then spent three months looking for another company willing to take the project. (theverge.com)

Sony has not given a meaningful public explanation beyond the usual corporate sludge about making a difficult decision after careful consideration, so the obvious speculation is money. Alinea Analytics currently estimates that Death Stranding 2 has sold around 2.5 million copies, roughly 1.8 million on PlayStation and 700,000 on PC, generating about $170 million in revenue across the two platforms. Those are estimates rather than official Sony or Kojima Productions figures, and we still do not have a reliable final development budget or profit figure for the game. (alineaanalytics.substack.com)

Those numbers certainly aren't spectacular for a giant modern AAA production, and analysts have understandably suggested the comparatively weak financial return may have made Sony nervous about another extremely expensive Kojima project that wasn't likely to arrive until around 2030 or later. (gamesradar.com)

Maybe that is exactly what happened. We simply don't know. But even if Sony eventually confirms that explanation, it only gets them so far. The original Death Stranding was explicitly confirmed by Kojima to have recouped its development costs and become profitable, later reaching five million sales across PS4 and PC by March 2021. Death Stranding was divisive by design, yet the weird fucking walking game still became a viable new IP. (gematsu.com)

More importantly, Sony knew exactly who it was hiring. Nobody accidentally discovers in year two that Hideo Kojima likes expensive actors, cutting-edge technology, cinematic presentation and absurd production ambition. Sony literally advertised those things as reasons to be excited about Physint. It promised him its full support, access to its technology and wider film expertise, while positioning the game as a next-generation showcase.

You cannot spend years asking Kojima to build the giant ambitious Kojima game, advertise its giant ambition as the selling point, then suddenly act shocked when the spreadsheet says the giant ambitious Kojima game might cost a lot of money.

Especially when you look at what PlayStation has been spending its time and money on instead.

Meanwhile, Guerrilla spent years making a Horizon live-service game:

Horizon Hunters Gathering is almost too perfect as the comparison.

Guerrilla finally revealed its Horizon multiplayer project in February 2026 as a “monumental step” for the studio, proudly explaining that it had grown through extensive testing, patching and feedback iteration. The game had the full live-service vocabulary: replayable missions, character builds, social spaces, gear progression, ongoing story content and a deliberately stylised art direction that somehow managed to make Horizon look like it had been squeezed through the same corporate-safe multiplayer filter as half the industry. (guerrilla-games.com)

That art direction wasn't some late compromise either. Guerrilla had been advertising a “unique stylized look” for its Horizon online project years earlier, meaning the chunky proportions, bright colours, exaggerated faces and squeaky-clean character silhouettes were very much intentional. For a series whose main games are partly sold on extremely detailed environments, machines and character rendering, the result looks less like Horizon naturally evolving and more like Sony desperately trying to manufacture another broad, marketable multiplayer cast.

Then people actually played it.

Closed testing began in February, another round followed in May, and by June Guerrilla had reportedly begun rebooting the project after poor player feedback. The live-service structure is now reportedly being removed, the scope reduced, staff reassigned and the remaining team given until December to convince Sony executives that the revised game is worth continuing. Bloomberg's reporting also says the majority of Guerrilla had been working on Hunters Gathering, while only a tiny team was plotting out the next mainline Horizon game. (gematsu.com)

This project can be traced back to at least 2021. Sony and Guerrilla therefore spent years building it, publicly revealed it after talking up all that testing and iteration, let outside players properly touch it, and within roughly four months decided that the fundamental live-service structure needed ripping apart.

Now they get to spend even more time rebuilding it.

And all of this happened because Sony decided one of its major single-player studios needed to spend years making a live-service Horizon game in the first place. That is the real waste. The next proper Horizon game remains years away, while a huge amount of Guerrilla's development capacity went into a multiplayer experiment that still had no release date when Sony finally revealed it and now needs fundamental reconstruction.

This comes after Concord, The Last of Us Online, Bluepoint's cancelled God of War live-service project and the rest of the Jim Ryan-era wreckage I've already covered. Sony's grand internal live-service experiment is basically sitting at 0.5/5, and the half point is only because Marathon still technically has functioning servers.

Its Steam peak was 88,337 players at launch. By September it is regularly sitting at roughly 2,000 to 3,000 concurrent players. Six months after release, Sony's remaining live-service trophy is a game I've already described as a dead game walking, and the corpse hasn't started moving any faster. (steamdb.info)

Apparently these were the investments worth protecting.

But don't worry, Sony has a $305 million Marvel game:

Maybe live service was the wrong answer. Fine. PlayStation built its modern reputation on enormous single-player games, so perhaps Sony simply wants to concentrate on those again.

Marvel's Wolverine is a rather spectacular demonstration of the problem with that strategy too.

Reviews landed on 10 September, and at the time of writing the game is sitting around 77 on Metacritic, its weakest reception for an original Insomniac game since Song of the Deep in 2016. The broad response isn't that Wolverine is some catastrophic piece of shit. It's arguably worse for the economics involved: it's fine. Gorgeous presentation, decent performances, repetitive combat, linear environments and a surprisingly conventional game underneath all that production money. (gamesradar.com)

Ordinarily, a 77-rated superhero game would barely be worth mentioning here. Then you remember the money.

Internal Insomniac projections exposed in the 2023 ransomware breach put Wolverine at around $305 million in total costs. The same projection expected 10 million lifetime sales to produce roughly $85 million in profit, with break-even around five million copies. These were internal forecasts from years before launch rather than confirmation of Wolverine's final 2026 expenditure, but they show the scale Sony and Insomniac were already planning around. (gamingbolt.com)

Ten million sales for a projected $85 million profit is a fairly incredible demonstration of where AAA economics have ended up, and this isn't even Sony's IP. Sony and Insomniac are taking on the production burden of another enormous blockbuster built around Marvel's Wolverine, pouring hundreds of millions into somebody else's globally established superhero property. Marvel still owns Wolverine, the X-Men and the underlying cultural asset that all of Sony's development work is advertising.

Compare that with Physint, where Sony had the chance to establish a new prestige espionage IP around the creator whose name already effectively sells the genre.

Apparently that was where the financial risk became unacceptable.

And then there is whatever the fuck Fairgame$ is now:

Haven's Fairgame$ is still alive too, although even Sony seems reluctant to say its name particularly loudly anymore.

The project was announced in 2023, lost studio founder Jade Raymond, then lost its game director, suffered delays and reportedly struggled during testing to simply “find the fun”. Earlier this year it was being tested under the codename Project Espresso, while Sony filed a trademark for Break In, believed to be a potential new title. (insider-gaming.com)

September 2026 then supplied another punchline when new Project Espresso testing began and footage immediately leaked online. The basic concept still appears to be teams entering a map, breaking into vaults, stealing loot and extracting, while early reactions are once again criticising how generic the thing looks. (insider-gaming.com)

So while Sony abruptly decided it could no longer support the action-espionage game it spent years asking Kojima to make, Fairgame$, Break In, Project Espresso or whatever it eventually gets called is apparently still crawling towards release.

Excellent prioritisation.