By Olivia Hart

A 5.5% Nasdaq Composite plunge last week exposed the market’s central weakness: nominally diversified portfolios still depend on the same AI spending cycle.

Nvidia lost 9.3% and AMD 17.3% as investors reassessed the durability of technology spending. The decline spread overseas, with Japan’s Nikkei and South Korea’s Kospi down around 6% the following day. Investors moved into government bonds as concerns about AI spending triggered a broader retreat from risk.

AI has moved from an emerging theme a few years ago to the market’s dominant driver, pushing technology stocks to record valuations. TradingView wrote: “But the latest selloff has exposed a very big problem: we’re all hooked and there’s no simple way out.” The warning matters because one spending assumption now links US index weights, Asian manufacturing and global financing conditions.

Can four different assets remain one concentrated bet?

Yes. An investor holding an S&P 500 ETF, Nvidia shares, a global technology fund and Asian semiconductor stocks may look diversified on a statement.

All four assets carry heavy exposure to the same driver: continued AI spending. Nvidia, Microsoft and Alphabet command enormous weight in US indices, while other technology giants reinforce the sector’s influence.

The supply chain widens the same exposure. Taiwan manufactures advanced chips, South Korea supplies memory, and European and Japanese companies produce essential semiconductor equipment. A change in investment expectations can hurt businesses across that chain at once.

The portfolio resembles four suppliers to the same AI build-out: one order-book shock can travel through all four. TradingView defines the resulting danger as concentration risk, meaning too much exposure to one source of potential losses despite holdings across different companies, funds or countries.

How does borrowed capital magnify the AI trade?

CoreWeave, the Nvidia-backed cloud provider expanding computing capacity, has accumulated approximately $50 billion in debt. That balance makes the neocloud particularly exposed when financing costs rise or projects run late.

CoreWeave reported a $104.2 billion revenue backlog in the second quarter. The figure represents contracted future business, not cash available today, so investors must distinguish promised demand from current cash generation.

Higher borrowing costs can make future expansion more expensive. If construction slows while interest payments continue, companies may struggle to generate enough cash to meet their obligations.

The chain reaction starts with expensive financing, delays data centers and reduces equipment orders. Weaker growth expectations then make investors less willing to finance further expansion, tightening funding for the next round of capacity.

Chipmakers face a different risk. Nvidia does not depend on borrowing to the same degree as heavily indebted infrastructure operators, but part of its revenue growth depends on those customers continuing to buy its products.

Why do bonds, currencies and commodities also move?

When investors lose confidence in a dominant theme, they often cut exposure across portfolios. The selling can spread to unrelated stocks as funds raise cash, reduce leverage or rebalance positions.

Government bonds may attract demand as investors seek safer assets. Currencies react to shifts in risk appetite, while copper can respond to expectations for infrastructure investment.

These markets retain separate drivers, yet AI has become large enough for a sudden sentiment shift to affect multiple asset classes simultaneously. Holding a mixture of investments therefore does not guarantee protection when they share the same underlying trigger.

Does strong AI demand remove the valuation risk?

No. AI demand remains strong, and the technology could deliver substantial economic benefits. Neither point guarantees that every company in the sector will produce attractive returns.

CoreWeave supplies computing infrastructure at the center of the AI boom, yet its stock has fallen nearly 80% from its record. Nvidia’s stock is also 25% below its record, leaving investors to reassess how much future growth their prices already capture.

Investors need answer three questions before treating the rout as an entry point:

• How much future growth is already reflected in share prices?
• Can companies finance expansion without adding excessive debt?
• How quickly will infrastructure investment generate cash?

The selloff exposes two linked vulnerabilities: capital concentrated in one popular theme and expected growth dependent on expensive, debt-financed infrastructure. TradingView closed with the investor question: “Long-term reshuffling or a buy-the-dip opportunity?”

The evidence calls for balance-sheet and exposure checks, not an automatic dip purchase. Investors should map direct and indirect AI holdings, test debt-service capacity and separate contracted backlogs from available cash before adding risk.