Hasbro owns Dungeons and Dragons, which means it owns Baldur's Gate. Baldur's Gate 3 shipped on 3 August 2023 and became the first game to win Game of the Year at all five of the major award shows.
Three years later there is no sequel, no announcement, and no studio attached to one.
That is not for want of asking. Hasbro went to the two parties on earth best placed to say yes, and both of them said no. The second no came with a reason, and the reason is more interesting than the refusal.
The build that existed and then stopped existing
Larian nearly did it. Swen Vincke told PC Gamer in August 2024, on the game's first anniversary, that a Baldur's Gate 4 had been partially playable inside the studio.
He described himself as "vulnerable" after the launch, and agreeing to the sequel in that state. His account of why is the most honest thing I have read about what success does to a decision. You are "prone to do the obvious thing", he said, because a sequel "is the easiest route to take".
Then he spoke to his teams and they killed it. They should be looking, he said, at "how we can do stuff that we get excited about".
Those quotes are from 2024. They got recycled this June wrapped around a fresher story, which is normal enough, but most of the coverage forgot to say so. It is a two-year-old decision that everyone spent this summer reacting to as news.
The reason is less flattering than the story wants it to be
There is a version of this where Larian walks away on principle, refusing to dilute something perfect, and the internet has been enjoying that version a great deal. It is not quite what Larian said.
Talking to GamesRadar+ in December 2025, around the reveal of Divinity, Vincke was blunter. The team is excited about the new game, he said, "because we weren't excited about what we were making back when we were doing the D&D thing".
That is a studio that got tired of somebody else's rulebook and wanted its own IP back. Which is a perfectly good reason, and I will take the real one over the noble one every time.
Then Hasbro phoned the man who wrote Baldur's Gate 2
James Ohlen was lead designer on Baldur's Gate in 1998 and co-lead on Baldur's Gate 2 in 2000. He also led Neverwinter Nights, Knights of the Old Republic and Dragon Age: Origins.
If you are drawing up a shortlist of people qualified to make a Baldur's Gate sequel, he is first, and there is a gap before second. Until December 2025 he also ran Archetype Entertainment, a studio owned by Wizards of the Coast. Hasbro did not have to leave the building to make the call.
He told PC Gamer in June how it went. "I don't, I would fail, and here's why I would fail." Then the line everyone quoted: "Doing Exodus is hard enough, but having to compete against Baldur's Gate 3? That would be insanity."
Look at what he is not saying. He is not saying the game would be mediocre, or that the licence is cursed, or that the market has moved on. He says he would fail.
That is a strangely specific thing for a designer with four landmark RPGs behind him to say out loud about a project with a blank cheque attached to it.
Half a decade of horror, from a man shipping on Unreal
His stated reason was the technology. Building the tooling to make a game at that scale, he said, is "at least half a decade of horror, building all that stuff".
Which sits oddly beside his day job, because Exodus, the science-fiction RPG his studio is making for 2027, runs on Unreal Engine 5. He is not contradicting himself. He is being precise about where the hard part actually is.
Unreal will render you a gorgeous alien planet. What it will not hand you is a decade and a half of accumulated machinery for writing, branching, staging and testing a reactive party-based CRPG, the kind where a player does something ruinous in act one and the game has to keep track of it for another eighty hours.
That machinery is internal tools, conventions and habits. It exists only because one group of people kept building it in the same direction, year after year, while shipping other games on top of it.
Borrowing the engine is how this franchise survived the first time
So Ohlen asked the sensible follow-up: could Hasbro get Larian to license out the engine?
That question already has an answer in this franchise's own history. BioWare built the Infinity Engine for Baldur's Gate in 1998. Interplay licensed it, and Black Isle Studios used it to ship Planescape: Torment in 1999 and Icewind Dale in 2000.
Two of the most loved RPGs of that era exist because a third studio's tooling could be borrowed rather than rebuilt. Everyone involved understood that the engine was the reusable part.
Mark Darrah, who produced Dragon Age at BioWare for years, made the same argument in public on 23 July, speaking to FRVR. Get Larian to license the Divinity Engine, let someone build what amounts to an oversized expansion pack, call it Baldur's Gate 4. He floated Owlcat, who have spent years shipping large CRPGs on off-the-shelf technology.
It is a good plan and it cannot happen, for reasons that are structural rather than creative. In 1998 the licensor and the licensee were effectively one company. Black Isle was a division of Interplay, and Interplay published Baldur's Gate.
In 2026 the engine belongs to Larian, a private studio of roughly five hundred people that owns its own IP, answers to no publisher, and has said since December that it would rather be making its own thing.
Hasbro cannot make that decision. It can only ask, and it has been asking.
Trent Oster thinks it was the money, and he is not wrong
The strongest argument against everything above comes from Trent Oster, who co-founded BioWare and then spent more than a decade at Beamdog trying to get his own Baldur's Gate 3 made before Larian got there first.
He puts it down to money. Larian succeeded, he told PC Gamer in June, because it had the financial backing to chase the ambitious version without compromise. His own pitch was a fraction of the size: "It wasn't going to be a $100 million game. I think we were pitching it in the $20 million range."
He is right that the money mattered. Look at the order it arrived in, though.
Larian raised $944,282 on Kickstarter for Divinity: Original Sin against a $400,000 goal, hoped to spend three million euros, spent four and a half, delayed its tax payments and came close to not existing at all. That was 2014, with a team of about thirty. Baldur's Gate 3 was made by around four hundred people.
The money followed the tooling, because of the tooling. Hand Beamdog a hundred million dollars in 2015 and you get a bigger version of the game Beamdog already knew how to build. Nobody has yet found a way to buy their way backwards into having already spent ten years on something.
The engine Larian is already throwing away
Larian's next game is Divinity, revealed at the Game Awards in December 2025 and billed by the studio as its most ambitious yet. It is being built on a new engine, because the one underneath Baldur's Gate 3 was not good enough.
Vincke gave the specific example in an interview published on 16 December 2025. Baldur's Gate 3's city had to be split into two parts, he said, "because we didn't have streaming in it and couldn't fit it into memory". He described the game as carrying "a bunch of constraints, and a lot of smoke and mirrors", and said thank God it worked.
Set that beside Darrah's proposal and Ohlen's question. The asset everybody wants to license is the one its owner considers obsolete and is in the middle of replacing.
Larian has been building and discarding the same lineage of engine since around 2010. One for Original Sin in 2014, another for Original Sin 2 in 2017, another for Baldur's Gate 3 in 2023, and now a fourth.
That is the part Hasbro cannot buy, and it was never the engine. It is a studio that has thrown away its own tools three times and started again, each time because the tools were the thing standing between it and the next game.
You cannot commission that. There is no line to put it on in a contract. It is a habit rather than an asset, and habits do not show up on anybody's balance sheet.
A billion dollars and a fifty-six million dollar hole
None of this makes Hasbro stupid, and its position is reasonable from where it sits. Chris Cocks said in May that the company is putting a billion dollars into single-player games, mostly around D&D, Magic and Transformers, and wants a pipeline running for the next decade.
The Baldur's Gate 3 licensing deal on its own earned Hasbro somewhere around ninety million dollars. Sitting on a licence that valuable and doing nothing with it is its own kind of failure when there are shareholders in the room.
Then on 21 July the second-quarter results landed with a $56 million impairment against what Hasbro called its "refocused Digital Games portfolio for 2028 and beyond".
On the call, Cocks said that work included cancelling several games scheduled for 2028 and later, and that the company expects total digital spend to fall at least twenty-five percent a year by 2028. In the same remarks he held up Baldur's Gate 3 as one of the strong proof points for where the upside is.
Both are true at once, and together they are the story. The proof point is a game somebody else made, on tools Hasbro does not own, by a studio that has since walked off. The response is to spend less on making games.
What is actually for sale
On 25 July, Dark Horse announced a Karlach comic and tagged Larian in the post. Larian replied that it is not involved in any Baldur's Gate 3 projects, and that it would rather talk about Divinity.
It was not the first time the studio has had to say a version of that sentence this year, and it will not be the last. The licence and the people who made the good game with it now live in separate places.
Hasbro can sell the setting, the name, the tieflings and the whole Sword Coast. It has the one genuinely scarce thing in this industry, which is an audience that will show up for a word.
What it cannot sell, or buy, is sixteen years of the same people rebuilding the same engine because the last one got in their way.
Larian is about to throw its tools out for the fourth time. Watch what they build with the new ones, because that is the part nobody can license.
