Nvidia, Broadcom and Marvell shares all fell in the latest session as concerns about the durability of AI infrastructure spending rippled through the semiconductor sector, TradingView reported.

The three names rarely move on separate logic anymore. Nvidia sells the GPUs and systems that anchor AI data centers; Broadcom supplies custom AI accelerators and networking silicon to the largest cloud operators; Marvell designs custom silicon for the same hyperscaler customers. When investors start questioning whether that spending pipeline holds, all three tickers answer at once.

That is what happened. A trading-day selloff hit each of the three stocks, and the shared trigger was unease about AI capital expenditure — specifically, whether the multi-year buildout commitments from cloud providers and sovereign buyers will keep growing at the pace the market has already priced in.

Why one worry drags three stocks?

The three companies sit on the same revenue stream from different angles. Nvidia's data center business has made it the most valuable chipmaker on the market. Broadcom has built a franchise co-designing custom accelerators for hyperscalers who want alternatives to merchant GPUs. Marvell plays in the same custom-silicon lane, plus optical interconnects that tie AI clusters together.

So the trading logic is simple. If AI spending decelerates, merchant GPU orders slow at Nvidia, custom accelerator programs get stretched out at Broadcom and Marvell, and networking upgrades defer. One assumption, three income statements.

What changed for investors?

According to TradingView's report, the selloff reflects growing concerns about AI spending rather than any single company disclosure. That distinction matters. No reset of guidance, no lost customer, no fab hiccup — just a repricing of how long and how steep the AI capex curve runs.

That kind of sentiment-driven move cuts both ways. These same stocks have led the market's gains through the AI build cycle, and drawdowns driven by spending skepticism have repeatedly reversed once hyperscalers confirmed their budgets in subsequent earnings calls. Investors have, in effect, been re-running the same debate each quarter: peak capex, or not yet.

What comes next?

The next hard evidence arrives with quarterly earnings and capex guidance from the major cloud customers, plus each chipmaker's own outlook. Until then, the three stocks are likely to trade as a single bet on AI infrastructure durability — falling together on doubt, recovering together when the spending numbers hold up.

By Rebecca Stone, Chip Dispatch