What the studio confirmed
Lightspeed LA, the California studio Tencent built to make a Western blockbuster, confirmed at the end of July 2026 that it was cutting about 80 jobs and changing the creative and development direction of Last Sentinel, the open-world game it announced in 2023. The studio said the decision followed extensive internal reviews and play-testing. It said affected staff would receive comprehensive support and, where possible, the chance to transfer internally, and that North America remains a cornerstone of its global gaming strategy.
Tencent has not called the project cancelled. That distinction is the reason this belongs in front of an owner rather than only in a games column. A company that says it is redirecting a programme is telling you the spending continues. A company that says it is cancelling one is telling you the spending stops. Last Sentinel is being described as the first and reported as something close to the second.
Where the two accounts split
Bloomberg reported on the same project a day before the studio spoke, and its account is harder. Citing people familiar with the work, Bloomberg put development at roughly six years and the budget in the hundreds of millions of dollars, described persistent staff turnover and an unclear creative vision, and said an internal playtest earlier this year drew overwhelmingly negative feedback. Bloomberg also reported that Tencent gave the studio until July to show something better, that Todd Papy, a veteran of the God of War series, had been brought in, and that the studio did not clear the bar.
These are not the same story told twice. One is a company statement. The other is reporting built on unnamed sources. They agree on the single fact visible from outside, which is that about 80 people lost their jobs, and they disagree about what the project now is.
That disagreement is not a reason to discount either one. It is the ordinary shape of a large write-down while it is still in progress. The reporting describes the decision. The statement describes the framing a company chooses while the accounting is still open. Anyone who has sat on the buying side of a stalled programme has heard both versions in the same week.
A review process that worked too late
The detail worth taking is not the money. It is when the evidence arrived. Lightspeed LA credits its own extensive internal reviews and play-testing for the decision, and Bloomberg places the decisive negative playtest earlier this year, which its account makes the sixth year of the project. The review process was not missing. It produced a clear verdict. It produced it after most of the money had been spent.
This is the expensive failure mode, and nothing about it is specific to games. A governance process that exists but cannot fire early is worse than an honest admission that there is none, because it supplies false comfort at every board meeting in between. The question was never whether a long programme gets reviewed. It is whether the review is scheduled, or whether it waits for someone senior to run out of patience.
What to change in your own long builds
Two changes are cheap and worth making before the next multi-year commitment. Put a kill review on the calendar at a fixed share of budget, at a third and again at two thirds, and give it a written mandate to stop the work rather than merely to comment on it. Then require that the review reads the most recent real test with users instead of the roadmap. On Last Sentinel the playtest and the plan said different things, and the plan was the one that kept getting funded.
The second change is in how you read a supplier. When a vendor tells you it is shifting the creative or technical direction of something you depend on, treat that sentence as a dated event rather than as reassurance. Ask what the last internal test showed and when it ran. If the answer describes a roadmap, you have learned what you needed to know.
Read next: DLSS 5 Ships as an Art Asset, Not a Toggle | Xbox Cuts 3,200 Jobs and Admits It Is Not Healthy
