Google made AI music generation free for 800 million Gemini users on February 18, 2026. Then it capped every track at 30 seconds and walked away from the professional market entirely.
This isn’t a beta limitation. It’s the business model. Lyria 3 ships with professional-grade audio specs — the kind musicians actually use in studios — but refuses to build the features those same musicians need to finish a song. No stem export. No MIDI control. No tracks longer than a TikTok clip. Google looked at AI tools targeting creative work and decided to fragment the market instead of dominating it.
The question isn’t whether Lyria 3 works. It’s whether “good enough for TikTok” just became the ceiling for AI-generated music.
Google shipped studio-quality audio and gave it away for free
Lyria 3 generates tracks in under 20 seconds, according to Google’s official announcement. That’s faster than Suno’s multi-minute workflow. And it’s available immediately in 8 languages — English, German, Spanish, French, Hindi, Japanese, Korean, Portuguese — not the English-only rollout most AI tools ship with.
The accessibility is the land grab. No waitlist. No paywall. Just an 18+ age gate and a Gemini account. Google trained this model with input from actual musicians and producers, which means it understands musicality differently than competitors who optimized purely for prompt-following. The result: professional-grade fidelity without professional-grade friction.
But here’s the thing. Google didn’t just make music generation accessible — it made it *disposable*. Thirty seconds is all you get. That’s the entire product.
The 30-second cap isn’t a bug — it’s the entire strategy
Google isn’t competing with Suno. It’s conceding the pro market to focus on creators who need 15-30 second clips for YouTube Shorts and TikTok. TechCrunch reported the 30-second hard cap at launch with no workaround and no premium tier announced. For creators chasing viral AI-generated tracks on short-form platforms, 30 seconds is all they need.
Suno V5 offers multi-track editing, full song structure, and longer tracks. Lyria 3 offers speed. The limitation is the moat. Google is betting casual creators outnumber professionals 100-to-1, and that most people generating AI music aren’t trying to release an album — they’re trying to score a reel.
Dream Track, Google’s YouTube-integrated music tool, expanded globally the same week Lyria 3 launched. The integration *is* the product. Google doesn’t need to build pro features. It needs to own the distribution channel where 30-second clips live.
The copyright problem Google won’t solve and the pronunciation errors it can’t
Google’s artist mimicry policy is vague. The company says it won’t outright copy artists but will generate “similar style or mood” if you name them. Then it admits, according to the same TechCrunch report, “it’s not clear if generation will make it easier for others to decode the music style of a particular artist.” That’s not transparency. That’s liability hedging.
And SynthID watermarking, Google’s answer to authenticity, is embedded in every track — but imperceptible to listeners. While platforms banning AI music cite authenticity concerns, Google’s watermarking approach shifts verification burden to listeners who have no incentive to check. You can pass AI music off as human unless someone uploads it to Gemini for verification. Which nobody will.
Then there’s the stuff it just gets wrong. Early users reported pronunciation errors in self-generated lyrics — no editing UI exists to fix them. Occasional gibberish singing. Ignored prompt instructions. These aren’t catastrophic failures, but they’re honest ones. Google shipped a tool that works most of the time, not all of the time.
Professional musicians need tools Lyria 3 will never build. Casual creators need speed Suno can’t match at this price. The AI music market just split in two. Google’s betting the casual side gets bigger. Suno’s betting the professionals pay more. Both might be right.
