Amazon Games couldn't even hand in somebody else's MMO homework

Only a few months ago, Amazon Games was still insisting its MMO business had merely changed shape. The October 2025 cuts gutted first-party AAA development, specifically including MMOs, but Lost Ark and Throne and Liberty were held up as the surviving pillars. Amazon told players both would continue receiving updates and support for the “foreseeable future”. (The Verge)

By June 2026, Amazon Game Studios boss Jeff Gattis was openly talking about how expensive traditional MMOs were to build and maintain, and how poorly their audience overlapped with Amazon's new mid-core and casual ambitions. Even then, he pointed towards the two Korean MMOs as evidence that Amazon hadn't completely abandoned the genre: “We haven't walked away.” (GamesBeat)

Two months later, Amazon announced its exit from both.

Throne and Liberty moves to NCSoft and FirstSpark in Q4 2026, less than two years after its Western launch. Lost Ark follows in early 2027, with Smilegate taking over publishing and live operations in the West. Customer support, policies, account responsibilities and other services move with them, while Smilegate will even take control of Lost Ark's website and community channels. Amazon has somehow reached the point where its MMO publishing strategy ends by asking the developers to take responsibility for publishing their own games again. (Throne and Liberty) (Lost Ark)

The retreat itself has accelerated beautifully. October 2025 gave us the “foreseeable future”. June 2026 gave us “we haven't walked away”. August gives us migration instructions.

New World shuts down on 31 January 2027, which I've already written about and don't need to exhume all over again. The useful part here is the timing. Amazon's own MMO dies at almost exactly the same point that Amazon finishes removing itself from its two remaining third-party MMOs. Within months, the surviving “pillars” went from evidence that Amazon Games still had an MMO business to a list of services being transferred somewhere else.

Lost Ark was the one that actually worked:

Lost Ark makes Amazon's retreat much more interesting because the original publishing deal actually worked.

Its February 2022 Western launch was enormous. Steam concurrency peaked at 1.325 million, Amazon claimed more than 10 million new Western players, and more than 1.5 million Founder's Packs reportedly generated around $75 million, roughly £55 million at the time, before the normal free-to-play spending had really begun. Amazon threw Twitch, Prime Gaming, AWS, localisation and a huge marketing campaign behind it. For once, all that corporate machinery produced exactly what it was supposed to produce: one of the biggest MMO launches the West had seen in years. (Amazon)

The bots were completely out of control and polluted the concurrency numbers almost immediately. Amazon was already banning huge numbers of bot accounts shortly after launch, and later ban waves could wipe hundreds of thousands off the visible Steam population. Still, there was obviously an enormous real audience underneath them. Bots don't explain 1.5 million paid Founder's Packs or the revenue Smilegate subsequently booked from the Western operation. (Lost Ark)

Smilegate RPG's accounts give us a much better idea of the money involved. Overseas Lost Ark revenue tied to the Amazon publishing operation came to roughly $135 million, around £110 million, in 2022, then about $110 million, around £87 million, in 2023. By 2024 it had fallen again to roughly $88 million, around £69 million. Those are Smilegate-side revenues, so they don't tell us Amazon's own profit or exact contractual split, but they do show that the Western business was generating serious money for several years. (Inven)

Then came 2025 and 2026.

Average Steam concurrency was around 26,800 in January 2025. By July it had fallen below 12,000, December averaged roughly 8,700, and August 2026 is hovering around 5,000 to 6,000. That 1.32 million launch peak now sits at the top of the graph looking like it belongs to an entirely different game. (Steam Charts)

The wider business followed it down. Smilegate's Lost Ark revenue fell sharply again through 2025, while Amazon was cutting members of the Western team by early 2026. The trajectory had been obvious for years: huge launch, substantial early revenue, shrinking population, smaller and increasingly hardcore remaining audience, then Amazon deciding it no longer wanted the operational burden at all.

That makes Lost Ark the most damning example in Amazon's portfolio. It was handed a proven Korean game, created a gigantic Western launch and had years of meaningful revenue to work with. This wasn't Amazon discovering that nobody wanted the product. Millions of people tried it. Huge numbers spent money. Then Amazon and Smilegate spent the following years watching them leave.

Throne and Liberty was the expensive gamble:

Throne and Liberty is a much sillier deal.

Amazon signed the publishing agreement with NCSoft in February 2023, before the game had even launched in Korea. NCSoft's disclosure put the contract at more than 2.5% of its 2021 consolidated revenue, giving us a minimum value of around $44 million, roughly £36 million at the time. The actual figure was kept confidential, so Amazon committed at least that much simply to secure the publishing arrangement. (Yonhap)

Amazon then took on the international work across the Americas, Europe and Japan, covering PC, PlayStation 5 and Xbox Series consoles. Localisation, marketing, community management, customer support and live operations became Amazon's problem. NCSoft's later reporting also says the agreement generated royalty income based on a percentage of Throne and Liberty sales, so NCSoft continued participating in the upside while Amazon ran the Western publishing machine. (NCSoft)

Then Throne and Liberty launched in Korea in December 2023 and promptly failed to become the new NCSoft monster everyone had hoped for. Analyst expectations were slashed as traffic and spending disappointed. One revised estimate put Korean daily revenue at around $190,000 to $220,000, roughly £150,000 to £175,000, which would translate to somewhere around $70 million to $80 million in annual gross revenue if sustained. That sounds large until you remember this was NCSoft's next flagship MMO after years of development, running one of the most aggressive MMO monetisation models in its home market. (Business Post)

Amazon still got its Western launch spike. Throne and Liberty peaked at 333,393 Steam players and averaged about 155,000 during October 2024. December was already down to around 47,000. January 2025 averaged roughly 28,000, December was below 7,000, and August 2026 sits around 5,000 to 6,000. Console players add to that total, but they aren't going to turn that population graph upside down. (Steam Charts)

Less than two years after launch, Amazon wants out.

NCSoft gets a Western operation Amazon spent years establishing, its own PURPLE launcher joins the existing platforms, and publishing, support and live operations move back under NC control. From NCSoft's side, the sequence is remarkable: secure a publishing contract worth at least $44 million before launch, collect a percentage of subsequent sales while Amazon handles the international operation, then have Amazon hand the operation back once it no longer wants to run it. NCSoft found its whale, and it was Amazon Games.

Westernisation, Amazon style:

The most revealing part of Amazon's Korean MMO business was always its strange idea of what needed changing for Western players.

Lost Ark arrived with problems that were obvious almost immediately. Honing demanded enormous quantities of materials, progression increasingly pushed players towards maintaining alts, daily activities became chores, catching up became harder as the game moved forward and raid gatekeeping punished anyone who fell behind. The monetisation sat directly on top of that progression pressure. Spend enough time and you could grind through it. Spend enough money and a lot of that friction became considerably easier to deal with.

Amazon knew what players thought about this. By April 2023, Amazon and Smilegate were publicly discussing “Player Exhaustion - Learning & Dailies”, specifically mentioning Chaos Dungeons, Guardian Raids and Una's Tasks. Amazon's own focus-group work later produced the same complaints from people reluctant to return: gatekeeping, bots and the daily grind. (Lost Ark) (Lost Ark)

Those problems took years of gradual changes. Guardian Raids were reduced, Chaos Dungeons were shortened, honing became easier, catch-up systems multiplied, Jump-Start servers appeared and solo raids eventually gave players another route around some of the gatekeeping. Amazon and Smilegate slowly chipped away at many of the same systems players had complained about while the population was already disappearing.

Amazon showed considerably more confidence when “Westernisation” meant corporate-safe content changes.

The Artist class became the obvious Lost Ark example. Amazon specifically announced costume changes to “better fit Western norms”, including shorts underneath short skirts and longer trousers or tights on other outfits. Apparently the acceptable amount of visible anime thigh was a Western-market issue that could be identified and dealt with rather efficiently. (Lost Ark)

Then Amazon prepared Blue Protocol and somehow became even more explicit about it. Franchise lead Mike Zadorojny openly discussed changes being made for “cultural, legal, and regulatory” reasons and to target a Teen rating. Amazon altered outfits on younger-looking characters, removed the character creator's infamous breast-physics “shake” button and removed the smallest female body option from its Western build. This was all being discussed long before Western players ever had the game in their hands. (GameSpot)

Amazon spent a remarkable amount of time telling people how it intended to sanitise Blue Protocol for Western release. Then the actual MMO bombed in Japan, Bandai Namco killed it, and Amazon's heavily prepared Western version never launched at all. Years of localisation work and arguments over character models ultimately produced nothing because the game itself couldn't survive long enough to reach the audience Amazon was supposedly protecting. (The Verge)

That pattern is what makes the whole corporate-safe approach so ridiculous. Amazon's supposed expertise was adapting Asian MMOs for Western players, yet so much visible energy went towards deciding what Western players were allowed to see while the games underneath remained saddled with the exact progression, monetisation and retention problems Western players kept complaining about.

Amazon did soften parts of the Korean business models, but neither Western version was remotely cleaned of the monetisation tied to the grind players complained about. Lost Ark continued selling Royal Crystals, Crystalline Aura Plus and bundles stuffed with card packs, honing materials and other progression resources, with Amazon itself advertising packs as a way to get a “significant head start” in Tier 4. Throne and Liberty went even further by making Lucent both its real-money premium currency and the currency of the player Auction House, while the same Lucent could pay for the in-game shop, Battle Pass and Leveling Pass. So Amazon's grand Westernisation still left players able to spend real money around the same progression friction, materials and time sinks burning them out. The cash shop apparently fitted Western norms just fine. The skirts needed intervention.

Covering a skirt was easy. Making the game worth sticking with was apparently the difficult part.

Built by the boardroom:

Amazon Games' MMO history makes far more sense when you look at it as a corporate expansion plan rather than something built around any particular love or understanding of the genre.

Amazon had AWS, Twitch, Prime, an enormous marketing reach and access to more capital than almost any conventional games publisher could dream of. That bought studios, developers, licences, advertising, infrastructure and repeated attempts. It couldn't buy the judgement required to turn all of those pieces into a healthy MMO portfolio.

New World gave Amazon complete control over its own game and is now shutting down. Lost Ark gave it an established Korean success with a gigantic Western launch and is being handed back. Throne and Liberty gave it another major Korean publishing partnership, cost at least $44 million just to secure the deal, and is being handed back less than two years after launch.

Blue Protocol added another variation. Amazon spent years preparing Bandai Namco's huge anime MMO for Western release, including loudly explaining the content it intended to censor or alter for Western sensibilities, only for the Japanese game to collapse before Amazon could release its version. The Lord of the Rings MMO provided yet another expensive attempt before disappearing during the wider MMO retreat.

These projects all had different problems, which makes the overall record worse. Amazon managed to fail with its own development, a proven external hit, a speculative external bet, a major Japanese partnership and one of the largest fantasy licences on Earth. There wasn't a single technical mistake waiting to be fixed across all of them. The common factor was Amazon Games itself.

Gattis more or less revealed where management ended up in June. Traditional MMOs are expensive, difficult to maintain and don't overlap particularly well with the more casual audience Amazon now wants. Amazon's priorities have moved towards Luna, smaller and more approachable experiences and projects that slot more neatly into the wider corporate machine. (GamesBeat)

That makes the previous decade look even more absurd. Amazon charged into one of the most demanding genres in gaming because MMOs looked like an attractive long-term business. It repeatedly threw money and corporate structure at the problem, imposed its own ideas about what Western players wanted, then discovered that running these games required years of awkward, specialised work after the launch marketing stopped mattering.

The genre was treated like another market Amazon could enter through scale. Players got the results.