Stability AI raises $76 million from the record labels that once sued it

The labels that dragged Stability AI to court are now on its shareholder register — the clearest sign yet that licensed training data has become the industry's business model, not just its talking point.

Stability AI has raised $76 million in a Series B round backed by the three major record labels — Universal Music Group, Warner Music Group and Sony Music — alongside gaming giant Electronic Arts and investment firms AMD Ventures and Pacific Alliance Ventures. The round, first reported by Variety, lands two years after UMG and Sony were among the plaintiffs in a copyright lawsuit against the company over models trained on their catalogs without permission. Today they're investors, with deals already signed to build new generative-music models from their own IP.

The money follows last week's launch of Stable Audio 3.0, a family of music-generation models trained entirely on licensed data, available through a plugin for digital audio workstations or directly on StableAudio.com. CEO Prem Akkaraju says the capital will fund the product suite, applied research, and a professional-services arm; Coatue co-founder Thomas Laffont is joining a board that already includes James Cameron and Sean Parker. The raise brings total funding under Akkaraju to $232 million since he took over in 2024.

The strategic flip is the story: the same rights holders who spent 2023 and 2024 arguing in court that training on copyrighted work was infringement now have equity in an AI company — and a reason to want its models to succeed. It's the licensing-first playbook we watched take shape as Suno hit 100 million users while UMG and Sony sought $9 billion, and Stability just became its biggest proof point. If the labels' bet pays, "trained on licensed data" stops being a compliance label and becomes the moat — and unlicensed rivals like Suno face partners-turned-competitors with every incentive to squeeze them.

What to watch: whether EA's involvement extends the same model to game audio and sound design — and whether the lawsuit against Stability is formally dropped or simply left to fade.

Would you pay more for creative tools trained only on licensed data? Tell us in the comments.

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Read more

Houmo moves compute inside memory — and A20 enters production

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Brain organoids classified Braille at 83% — in a trio, not alone

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute