Huang Renxun: In the next 10 years, the semiconductor market will expand 10x—has the AI super cycle just begun?

The AI boom hasn’t ended yet.

At least, that’s what NVIDIA CEO Huang Renxun believes.

Recently, Huang Renxun said in an interview that over the next roughly 10 years, the global semiconductor industry scale may need to expand to 10 times its current size in order to meet the explosive compute demand growth of AI agents (AI Agents) and the robot era.

This view has quickly sparked heated discussion in the market:

Can the super cycle of AI chips really last another 10 years?

Why is Huang Renxun so optimistic?

In the past few decades, chips mainly served:

Computers.

Smartphones.

Servers.

And Huang Renxun believes the biggest change ahead is:

Compute demand will no longer come only from humans, but from hundreds of millions of AI agents and robots.

If every AI Agent needs inference, learning, and real-time computation, then global demand for GPUs, HBM memory, networking equipment, and data centers will be far higher than today.

Has the AI super cycle really not ended yet?

From an industry logic perspective, AI is still in the infrastructure buildout stage.

At present, global tech giants are continuing to invest in:

1、AI data centers
2、GPU clusters
3、High-bandwidth storage (HBM)
4、High-speed networking equipment
5、Power and cooling infrastructure

All these areas require a large amount of semiconductor support.

In other words, more of what’s being built now is the “shovel-selling” stage, not the final application stage.

So why has the market adjusted recently instead?

Recently, AI chip stocks have shown a clear pullback.

The reason isn’t that the AI story has disappeared.

Instead, capital has started to reassess:

1、Whether the ROI of AI investment is fast enough;
2、Whether tech giants’ capital expenditures can keep growing;
3、Whether chip companies’ current valuations have already priced in performance for the next few years.

So what the market is correcting is valuation—not the long-term AI logic.

The biggest challenges for the semiconductor industry in the future

Huang Renxun also noted that even if demand continues to grow, there are bottlenecks on the supply side.

The constraints on industry development today include:

1、Insufficient supply of HBM high-bandwidth storage;
2、Data center construction speed;
3、Power resources;
4、Land and advanced packaging capacity.

This means:

In the next 10 years, the semiconductor industry will not only be a contest of technology, but also a contest of supply-chain capabilities.

Which companies may continue to benefit?

If the long-term AI trend doesn’t change, the core companies in the industrial chain are still worth ongoing attention:

1、NVIDIA—leader in AI GPUs;
2、SK hynix—core supplier of HBM;
3、Micron Technology—leader in memory chips;
4、TSMC—leader in advanced process foundry services.

In the future, whoever can continuously secure AI orders will have a better chance to ride through the industry cycle.

My view

When Huang Renxun proposed “10x growth over the next 10 years,” it sounds more like a judgment about the industry’s long-term space rather than a forecast of short-term stock prices.

The AI market won’t keep rising in a straight line.

In the middle, there will definitely be:

Valuation corrections.

Earnings validation.

Capital rotation.

But if AI agents, robots, and autonomous driving continue to land in the real world, semiconductor demand still has a chance to maintain long-term growth.

The real opportunity isn’t chasing every price increase—it’s identifying those core companies that can continuously benefit from AI infrastructure buildout.
#黄仁勋加持！Marvell股价飙升 $NVDA ‌