I was clearing phone storage recently and noticed NetEase Cloud Music taking up 3.8 GB.
I write code for a living. I know what 3.8 GB means — a mid-size game install. A full development environment. And this is a "music app."
Occupational hazard kicked in, so I ran Charles to intercept NetEase's network requests. Opening the home screen — without tapping anything — fires off over twenty API calls: recommended playlists, livestream listings, short-video feeds, social updates, AI-generated playlist commentary, shopping mall data, banner ad slots. The requests actually related to "playing music" accounted for maybe a third.
A music app where two-thirds of the home screen requests have nothing to do with music. Wild.
These Apps Through a Technical Lens
Let's start at the foundation.
I've spent years at big tech companies. I know exactly how a product ends up like this. Nobody sets out to build a Frankenstein's monster — it's the result of KPIs stacking up year after year.
Year one: a product manager says "we need social features" because social boosts retention. So they add comment sections, activity feeds, "Listen Together." Year two: another PM says "we need livestreaming" because livestreaming monetizes. Year three: yet another PM says "we need short video" because ByteDance is eating screen time. Year four: AI's hot, "we need AI playlist reviews and AI-curated lists." Every annual OKR targets "breakthrough in XX domain." Nobody's OKR is "make the playback experience flawless."
Because optimizing a music player doesn't count as innovation. It doesn't tell a compelling story at a promotion review.
Here's one thing though: Spotify built a custom audio engine with adaptive bitrate switching across network conditions — it plays smoothly even on 2G. Apple Music's CoreAudio framework is deeply integrated with the OS, achieving 30%+ better decoding efficiency than third-party solutions. These are genuine technical moats.
In China? Most player cores are FFmpeg with a wrapper, audio rendering pipelines untouched for years. But the social matching system has been through three iterations, the recommendation algorithm team has twenty engineers, and the livestream gift animations have a dedicated GPU rendering pipeline. Where the technical resources go tells you everything.
QQ Music's membership system is a product design disaster — Green Diamond, Black Diamond, Purple Diamond, Luxury Green Diamond, Super Member. As an engineer used to clean RBAC permission models, every time I see this I picture a chaotic authorization matrix and a backend engineer who's been driven insane by business stakeholders. The system exists for one reason: price discrimination — selling the same content at different prices through different packaging. Trivial to implement, catastrophic for user experience.
Apple Music: China vs. International Are Two Different Products
Many people say "Apple Music is great," but I need to be specific — which region?
International Apple Music (US/Hong Kong/Taiwan/Singapore) is my daily driver. Global catalog exceeding 100 million tracks. Full feature set — Apple Music 1, Apple Music Hits, Apple Music Country (three 24/7 live radio stations), human-curated regional stations, Top 100 playlists, Friends feature, complete editorial content. The Dolby Atmos spatial audio library is extensive — mainstream Western and K-pop albums almost universally have Atmos mixes. Monthly cost around $10-11 USD (US region), but the experience is complete.
China-region Apple Music is a different product entirely.
The catalog is gutted. Apple's China site describes "tens of millions of songs." The global description says "over 100 million." An order-of-magnitude gap. The reason is simple: licensing in China is split between Tencent Music and NetEase. Apple Music isn't a major player domestically and can't secure all rights.
Features are gutted too. The Radio tab has only algorithm-generated genre stations — Apple Music 1 and the live stations are absent. Top 100 playlists are gone. The Friends feature is gone. Explicit-tagged content either doesn't appear or arrives months late — Eminem's album was available on US release day, but China had to wait through content review, sometimes for months.
The price is cheap — 11 RMB monthly for individual. But what you get for 11 RMB and what someone pays $10.99 USD for aren't remotely the same product. China-region Apple Music is a catalog-reduced, feature-stripped, content-reviewed edition. It's cleaner than NetEase — true. But calling it "perfect" means you haven't used the international version.
I use the Taiwan region myself — 165 TWD per month, about five bucks, catalog nearly identical to the US with better Chinese music coverage. Creating a foreign Apple ID isn't hard — look it up.
Spotify: The Algorithm Is Genuinely Elite
Spotify has no China region — using it domestically requires workarounds I won't detail here. But purely as a product, Spotify's recommendation system is the best I've encountered.
I have a rough understanding of the tech stack — collaborative filtering + NLP analysis of lyrics and reviews + audio feature extraction (BPM, key, energy levels, etc.) + user behavior sequence modeling. Discover Weekly surfaces tracks you haven't heard but will probably love because it doesn't just look at "people who listened to the same songs also listened to X" — it analyzes the acoustic signatures of the audio itself for cross-domain recommendations.
NetEase's early "Daily Recommendation" used similar collaborative filtering and was genuinely solid. But it's visibly degraded in the past two years. You can feel the recommendations aren't purely about what you'd enjoy anymore — they're what the platform "wants you to hear." Likely songs with paid promotion agreements, or new artists who need streaming numbers pumped. Users complain that annual listening reports don't match their actual habits. At the system level, the objective function has drifted from "user satisfaction" to "platform revenue maximization."
Spotify also has commercial recommendations (Discovery Mode), but they're explicitly labeled and don't replace your personal recommendation slots. Transparency is on a different level entirely.
Cross-platform support is another blowout — Windows, macOS, iOS, Android, Web, PlayStation, Xbox, car infotainment, smart speakers. Apple Music on non-Apple devices is painful — the Android client is sluggish, and the Windows experience is worse.
The Copyright Swamp
In 2021, Tencent Music was hit with antitrust action and forced to relinquish exclusive licensing. Everyone assumed platform catalogs would converge.
They didn't.
"Exclusive licensing" died; "exclusive first release" was born. After first-release windows expired, "priority recommendation" emerged. Early 2025, NetEase published an open letter saying SM Entertainment wouldn't renew — EXO, aespa, NCT, potentially gone overnight. Memberships you're paying for, playlists curated over years, greyed out by a single notice.
As an engineer, I'd frame it this way: China's music copyright system lacks a global authorization middleware layer. Internationally, Performance Rights Organizations (PROs) handle unified licensing for compositions — streaming platforms sign with a handful of PROs and cover most catalogs. In China? Rights holders negotiate one-on-one with platforms. Any rights holder can unilaterally decline renewal, causing catalogs to bounce between platforms like a game of musical chairs.
The result: users are forced into two or three simultaneous subscriptions. Jay Chou on QQ Music, some indie bands on Qishui Music, others only on NetEase. Annual membership costs totaling 400-500 RMB for a worse experience than a single Spotify Premium subscription.
This isn't copyright protection. It's copyright hostage-taking.
Why This Happens — A Big-Tech Insider's View
Having written code at large companies for years, my deepest takeaway is: what a product looks like isn't determined by user needs — it's determined by organizational structure and KPIs.
NetEase Cloud Music internally almost certainly looks like this: an "Online Music" business unit, a "Social Entertainment" business unit, probably an "Innovation" business unit. Each has its own P&L, its own KPIs, its own promotion track. Social Entertainment's KPI is user screen time and tipping revenue. Innovation's KPI is feature adoption rates. Online Music's KPI is paying subscriber count.
Nobody's KPI is "minimize time from app open to first song playing." Nobody's KPI is "seamless gapless playback." Nobody's KPI is "percentage of recommendations the user genuinely enjoys."
These metrics are too "basic." They don't tell a story at OKR reviews. They won't help you reach the next engineering level.
So what you see as NetEase Cloud Music isn't one product — it's three or four departments' outputs stitched together. The top half of the home screen belongs to the Music team. The livestream card in the middle belongs to Social Entertainment. The bottom navigation tabs are territory each department fought for — whoever contributes more DAU gets a tab. This isn't product design. It's departmental politics rendered as a frontend.
Why can Spotify stay clean? Because Daniel Ek established one rule from day one: Spotify is a music company. Every feature orbits three core scenarios: discover music, play music, share music. Podcasts were added later, but as a separate tab that doesn't contaminate the music recommendation feed. This product restraint is fundamentally organizational restraint — refusing to sacrifice core experience for any business unit's KPIs.
Can Chinese companies do this? No. Because a pure-music business model doesn't work — copyright costs are high, willingness to pay is low, ARPU is tiny. They have to stuff in livestreaming, social features, and e-commerce to cover losses. I get the business logic. But as a user — and as an engineer who cares about product design —
It's 2026. Wanting to just listen to a song in peace shouldn't be this hard.
My Setup
Here's how I solved it.
Primary: Apple Music (Taiwan region). 165 TWD per month — about five bucks. Complete catalog, full features, solid Dolby Atmos spatial audio. Chinese music coverage has caught up in the past two years — Jay Chou, Mayday, JJ Lin, Jolin Tsai all present. K-pop fully covered. Downsides: poor non-Apple-device experience, needs a foreign Apple ID.
Secondary: Spotify. The recommendation algorithm is genuinely unmatched — Discover Weekly has surfaced so many gems for me. Cross-platform crushes everything — Windows, Android, Web, car systems, smart speakers, all native clients. Where Apple Music's recommendations fall short, Spotify picks up the slack.
Chinese apps: all gone. If I need a track only available domestically, the web version does the job. Not worth keeping a 3.8 GB app installed.
It's kind of sad, honestly. Someone like me who just wants to listen to music in peace — there's literally nothing usable in China's music app market. My solution ended up being: sign up for a foreign account and pay five bucks a month for the privilege of not being bombarded.
No going back. Once you've used something clean, the alternative is unbearable. Try two uninterrupted hours on Apple Music's Taiwan region, then do the same on NetEase — the gap isn't "slightly better" or "slightly worse." It's two different eras.
It's not that China lacks good musicians. These apps turned the simplest possible act — listening to music — into a survival game of subscriptions, ad-dodging, upsell-avoiding, and social-feature hostage situations.
What's a music player's job? Play music.
One more thing, as an engineer: a good player should have less code than features, not the other way around.