Why Did US Tech Stocks Shed $1 Trillion Overnight on Black Thursday? Jensen Huang Falls Out of the Top Ten Richest as Markets Reassess AI’s Political Costs

Technology stocks bore the brunt of a sharp sell-off across all three major US stock indices on 5 November. Losses stood at 3.44% for the Dow, 4.36% for the S&P 500 and 5.52% for the Nasdaq. The Philadelphia Semiconductor Index ended 12.25% lower as selling intensified across the sector. SanDisk tumbled more than 22%; AMD, Micron and Western Digital each lost more than 17%; SK Hynix retreated over 13%; Broadcom shed more than 10%; and Nvidia declined 9.26%.

Concerns over tighter restrictions on AI infrastructure expansion followed the midterm elections’ “blue wave”, the immediate catalyst for the sell-off. Alongside retaking both the House and Senate, Democrats secured breakthroughs in key data centre construction hubs, including Texas and Georgia. Investors feared stricter requirements covering environmental approvals, power connections and land-use permits, potentially extending to temporary moratoriums. Approximately 30% of planned additional US data centre capacity could be affected, according to UBS estimates.

The basis for valuing AI assets has consequently changed. Earlier pricing assumed limitless demand and an eventual supply response, leaving political considerations with effectively no weight. Following 5 November, markets could no longer assign those factors a value of zero.

Which Business Models Are Most Exposed to Regulatory Approvals?

A 9.26% fall at the close wiped $447 billion off Nvidia’s market value in a single day. Having declined 24.1% from their record closing high of 6 October, its shares officially entered a technical bear market. The decline cut Jensen Huang’s net worth by $15.1 billion to $150.9 billion, moving him down the Bloomberg Billionaires Index from ninth to thirteenth.

For Musk, the wealth erosion was larger still: approximately $79.7 billion in one day, leaving his fortune at $785 billion as Tesla closed 10.77% lower and SpaceX declined 9.05%. Texas is central to his particular exposure. The incoming Democratic governor, having won the election, has threatened a freeze on new data centre construction. Both SpaceX and Tesla have AI infrastructure plans that depend heavily on the state’s approval environment.

The challenge for Ellison is more structural in nature. With Oracle shares down 39% this year and more than 53% below their November 2025 peak, his wealth has contracted by $159 billion over the past year to $164 billion, nearly half its earlier level. His loss on 5 November alone was another $13.2 billion. Selling computing capacity through cloud services is a core Oracle business, so the speed at which data centres receive approval directly governs how rapidly that capacity can grow.

A common dependency runs through all three business empires: large-scale data centre construction to serve demand for AI computing. Their roles differ, with Nvidia providing chips, SpaceX developing its own computing capacity and Oracle selling that capacity. Yet all three asset types face a higher valuation denominator when construction approvals cease to be an assumed clear passage and instead require a wait at every stage.

Control of both chambers gives Democrats powers to investigate, oversee funding allocations and scrutinise appointments, but it does not allow them simply to pass a law banning data centres. Overcoming a Senate filibuster still requires 60 votes for most ordinary legislation, which also remains subject to a presidential veto. Even so, hearings and committee investigations can prolong policy uncertainty, slow approvals and increase compliance costs. Where an asset’s valuation assumes a doubling of computing capacity over the next three to five years, an approval slowdown need not turn into a “ban”. A “delay” is sufficient to alter the discounted cash flow calculation.

Underlying the sell-off was a reassessment of a premise that had supported the AI trade before 5 November: computing capacity could grow without limit, with chip availability and capital expenditure the only constraints. The midterm results disrupted that premise by adding electricity, land, environmental permits and local political willingness to the constraints on expansion. Capital spending cannot remove these obstacles immediately; resolving them takes years, not quarters.

How Are Leverage and Crowded Positions Magnifying the Sell-Off?

This year has seen extreme gains across the AI hardware supply chain alongside extreme concentration in institutional holdings. A reversal in expectations can therefore bring automated stop-loss sales, the unwinding of leveraged positions and retail selling together in a brief window, deepening the decline. According to Michael Monaghan, a portfolio manager at Founder ETFs, highly leveraged hedge funds with tighter risk limits had to exit rapidly once share prices started falling, adding to intraday volatility.

The signals from CoreWeave warrant close attention. On the same day that its shares lost approximately 10%, its five-year credit default swap spread climbed to a record high. That widening in CDS spreads suggests credit markets have started reassessing the risks attached to AI computing operators. The message aligns with the equity sell-off: investors are increasingly reluctant to allow the scarcity-of-computing-capacity narrative to obscure higher financing and regulatory costs.

Which Developments Should Markets Watch Next?

Nvidia’s earnings release later this month will provide an important test. Expecting another 70% revenue increase next year, Monaghan believes data centre construction is “likely to continue” and sustain strong chip demand. He regards the sharp share-price fall as a buying opportunity. A more guarded assessment came from David Wagner of Aptus Capital Advisors. In his view, Nvidia retains a wide competitive moat through chip performance, its software ecosystem and developer lock-in. Its valuation, however, depends heavily on AI infrastructure spending, and markets believe that expenditure on such an enormous scale will eventually slow.

The repricing concerns the pace and route of AI expansion, not the validity of its long-term trend. A journey with green lights throughout is different from one with red lights at some junctions: traffic still moves, but arrival takes longer. When future cash flows account for most of an asset’s value, that additional time alone points to a lower valuation.

In essence, 5 November was the first sell-off in which markets factored the political costs of AI infrastructure into pricing models. Huang’s exit from the top ten, Musk’s loss of nearly $80 billion and the near-halving of Ellison’s wealth over a year were consequences, not drivers. Behind them lay the recognition that computing expansion faces a ballot as well as chip and electricity bottlenecks.

Attention now turns to how soon approval delays affect the computing capacity that data centres actually bring online. Another question is whether Nvidia’s earnings can establish a balance markets accept between strong orders and a longer route to future growth. Until approval issues actually delay the first data centre, however, the sell-off remains a repricing of probability rather than of an event that has already taken place.