Infrastructure3DNews AI→ original

Nvidia takes TSMC's crown: Apple is no longer the top customer (and that makes sense)

For decades, Apple was TSMC's golden child, first in line for the best process nodes. The AI boom has flipped that. Jensen Huang says openly that Nvidia is becoming the Taiwanese giant's largest customer. This is more than a matter of prestige; it is a fundamental shift. Demand for Blackwell chips is so enormous that Nvidia is ready to buy capacity in any volume, pushing even the makers of the iPhone aside. The center of gravity in tech has finally moved from consumer gadgets to the server racks used to train neural networks. Where TSMC once lived off your wallet, it now lives off the future of superintelligence.

AI-processed from 3DNews AI; edited by Hamidun News
Nvidia takes TSMC's crown: Apple is no longer the top customer (and that makes sense)
Source: 3DNews AI. Collage: Hamidun News.
Listen to article

If you want to know who really runs the tech world, you don't need to look at smartphone sales charts or app download counts. It is enough to look at the order book of TSMC, the only company in the world able to print the most complex chips. For a long time Apple topped that list, and it seemed an unshakable constant. But times have changed. Jensen Huang, the man in the unchanging leather jacket, now says with full confidence that Nvidia is ready to become, and may already be, the Taiwanese giant's largest customer. It is a historic moment that marks the end of the "mobility era" and the start of a full-fledged "AI era."

Let's recall the context. For years Apple was not just a customer for TSMC but a kind of sponsor of progress. Cupertino paid for the development of new process nodes, whether 5-nanometer or 3-nanometer chips, so it could be first to get them for its iPhones and Macs. Everyone else, Nvidia included, waited obediently in line, making do with leftover capacity or older technologies. That hierarchy seemed eternal, because nobody could match Apple in volume and margins. But generative AI made its own adjustments, turning compute accelerators into the scarcest commodity on the planet.

Why is this happening right now? The answer lies in the appetite of modern neural networks. Training a notional GPT-5 or new models from Meta* takes tens or hundreds of thousands of chips, each costing as much as a decent car. Nvidia doesn't just order wafers with dies; it demands sophisticated CoWoS packaging, over which a real war is now being fought. Demand for the Blackwell architecture exceeds supply so far that Nvidia can afford to buy up entire production lines without looking at the price tag. Apple, for all its stature, is boxed in by the consumer market: people are not ready to buy an iPhone at the price of a server.

This shift in TSMC's priorities has colossal consequences for the whole industry. Now it is Nvidia's requirements that will determine what the next generations of transistors look like. Where engineers in Taiwan once racked their brains over saving a milliwatt for a smartphone battery, their main task now is pushing terabytes of data through a chip without melting it. The focus of development is shifting from energy efficiency to raw performance and memory bandwidth. Nvidia is effectively dictating the rules of a game that everyone else, Intel and AMD included, will have to play tomorrow.

Of course, Apple will not vanish from the list of priority partners, but its status as a "special customer" is now in question. In a world where every other startup and every state dreams of its own sovereign AI, TSMC's capacity is becoming more important than oil. And Nvidia is today the main steward of that resource. Jensen Huang understands this better than anyone. His confidence is backed not only by financial reports but by the fact that without his chips, progress in AI would simply stall. That gives him leverage that Tim Cook could once only dream of.

The irony is that Nvidia was once just "one of" the makers of graphics cards for gamers. Today it has become the architect of global computing infrastructure. The handover of the top-customer title from Apple to Nvidia is not just a change of names in a table. It is official recognition that training models and processing data have become more important than making gadgets for end users. We are entering a world where hardware is built first of all for machines, not for people.

Bottom line: if Nvidia really does overtake Apple in orders at TSMC, it will mean that the shortage of AI capacity will become chronic, and priority access to new technologies will go to server solutions for good. Can Apple win back the lead with its own AI ambitions?

*Meta is designated an extremist organization and banned in Russia.

Story
ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Unified AI Workspace · Nomad

Test Frontier LLMs & Autonomous Agents in One Workspace

Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.

Try Nomad AI →Launch Nomad Agent
What do you think?
Loading comments…