US Tech Stocks Suffer a Black Thursday, Wiping Out $1 Trillion Overnight! Jensen Huang Drops Out of the Top Ten Richest as Markets Reprice AI’s Political Costs

On November 5, all three major US stock indices fell sharply. The Dow declined 3.44%, the S&P 500 dropped 4.36%, and the Nasdaq slid 5.52%. The heaviest selling was concentrated in technology: the Philadelphia Semiconductor Index closed down 12.25%, SanDisk plunged more than 22%, AMD, Micron and Western Digital fell more than 17%, SK Hynix dropped over 13%, Broadcom declined more than 10%, and Nvidia lost 9.26%.

The immediate trigger was the “blue wave” in the midterm elections. Democrats regained control of both the House and Senate and made breakthroughs in major data centre construction hubs, including Texas and Georgia. Markets were concerned that Democrats would impose stricter conditions on the physical expansion of AI infrastructure, including environmental approvals, power connections, land-use permits and even temporary moratoriums. UBS estimates that approximately 30% of planned additional US data centre capacity could be affected.

This marks a shift in the logic behind valuations. Previously, markets priced AI on the assumption that demand was unlimited and supply would catch up, effectively assigning no weight to political factors. After November 5, political factors could no longer be treated as zero.

Whose Business Models Depend Most on Regulatory Green Lights?

Nvidia closed down 9.26%, shedding $447 billion in market value in a single day. Its shares fell 24.1% from their record closing high on October 6, officially entering a technical bear market. Jensen Huang’s net worth declined by $15.1 billion to $150.9 billion, pushing him from ninth to thirteenth place on the Bloomberg Billionaires Index.

Musk’s losses were even greater. Tesla closed down 10.77% and SpaceX fell 9.05%, reducing his wealth by approximately $79.7 billion in one day to $785 billion. The specific risk facing Musk centres on Texas: the victorious Democratic governor-elect has threatened to freeze the construction of new data centres, while the AI infrastructure plans of both SpaceX and Tesla are closely tied to the state’s approval environment.

Ellison’s situation is more structural. Oracle shares have fallen 39% so far this year and more than 53% from their November 2025 peak. Over the past year, Ellison’s wealth has fallen by $159 billion, nearly halving his net worth to $164 billion. On November 5 alone, he lost another $13.2 billion. One of Oracle’s core businesses is selling computing capacity through cloud services, meaning the pace of data centre approvals directly determines how quickly it can expand that capacity.

Viewed together, the three share a clear characteristic: their business empires are deeply tied to the large-scale construction of data centres to meet AI computing demand. Nvidia supplies the chips, SpaceX builds its own computing capacity, and Oracle sells computing capacity. When the approval environment for that construction shifts from an assumed clear path to one requiring a wait at every junction, the valuation denominator rises across all three types of assets.

With control of both chambers, Democrats gain investigative powers, oversight of funding allocations and authority to scrutinise appointments. They cannot simply legislate a ban on data centres. Most ordinary legislation still requires 60 votes in the Senate to overcome a filibuster and remains subject to a presidential veto. However, committee hearings and investigations can create persistent policy uncertainty, delay approval processes and raise compliance costs. For an asset whose valuation depends on computing capacity doubling over the next three to five years, slower approvals do not need to become a “ban”. A “delay” alone is enough to change the outcome of discounted cash flow calculations.

At a deeper level, the sell-off exposed an assumption embedded in the AI trade before November 5: computing capacity could expand indefinitely, constrained only by chip supply and capital expenditure. The midterm elections changed that assumption. Electricity, land, environmental permits and local political willingness were added to the list of constraints on computing expansion. These constraints cannot be resolved immediately by capital spending. Their timelines are measured in years rather than quarters.

Leverage and Crowding: The Technical Factors Amplifying the Decline

Both gains in the AI hardware supply chain and the concentration of institutional holdings have reached extreme levels this year. When expectations reverse, automated stop-loss selling, leveraged position unwinding and retail selling can converge within a short period, amplifying declines. Michael Monaghan, a portfolio manager at Founder ETFs, noted that hedge funds operating with high leverage and tighter risk limits were forced to exit quickly as share prices began falling, intensifying intraday volatility.

CoreWeave deserves particular attention. Its shares fell approximately 10% that day, while its five-year credit default swap spread rose to a record high. Wider CDS spreads indicate that credit markets have begun repricing the risks facing AI computing operators. This points in the same direction as the equity sell-off: investors are becoming less willing to let the narrative of scarce computing capacity overshadow rising financing and regulatory costs.

What to Watch Next

One key checkpoint will be Nvidia’s earnings report later this month. Monaghan expects Nvidia’s revenue to grow another 70% next year and believes data centre construction is “likely to continue”, keeping demand for chips strong. Given the sharp decline in its share price, he sees a buying opportunity. David Wagner of Aptus Capital Advisors offered a more cautious framework: Nvidia’s competitive moat—its chip performance, software ecosystem and developer lock-in—remains wide, but its valuation is heavily tied to AI infrastructure spending, and markets believe that enormous expenditure will eventually slow.

The long-term AI trend has not been invalidated. What is being repriced is the speed and path of expansion. Moving from green lights all the way to red lights at certain junctions does not mean traffic stops, but it does mean reaching the destination takes longer. For assets valued primarily on future cash flows, a longer timeline itself implies a lower valuation.

At its core, the November 5 sell-off marked the first time markets incorporated the political costs of AI infrastructure into their pricing models. Huang falling out of the top ten, Musk losing nearly $80 billion, and Ellison’s wealth nearly halving over a year are outcomes, rather than causes. The underlying cause is the market’s realisation that the list of bottlenecks to computing expansion includes a ballot, alongside chips and electricity.

The next thing to watch is how quickly approval delays translate into changes in the computing capacity actually brought online at data centres, and whether Nvidia’s earnings can offer a new balance that markets accept between strong orders and a longer path to future growth. Until the first data centre is actually delayed by approval issues, the current decline is still pricing in a probability rather than an event that has already occurred.