【TMGM Financial Recap】Why Does Goldman Sachs View the November 5 Tech Sell-Off as a 1998-Style Liquidity Flush Rather Than a 2000-Style Bubble Burst?

Markets came under severe pressure in the November 5 sell-off, with the scale of the damage evident across major indicators. A 6.4% fall in the Nasdaq 100 erased approximately $2.3 trillion in market value in a single day, marking its steepest daily decline since March 2020. The Philadelphia Semiconductor Index slumped 12.25%, while a 70% jump took the VIX to its highest level since early April. After falling 9.26%, Nvidia stood 24.1% below its October high, officially putting it in a technical bear market. For Goldman Sachs, however, the size of the losses was secondary to a question directly relevant to investment decisions: did the selling reflect a liquidity-driven liquidation in an otherwise healthy growth story, or the first signs that its underlying economic rationale was flawed? Its assessment favoured the former.

Does the Sell-Off Resemble 1998 More Than 2000?

After weighing the present decline against two historical episodes, Goldman Sachs concluded that 1998 offered the closer parallel.

The case for that parallel rests on three points. To begin with, the disruption came from outside the economy rather than from deteriorating fundamentals. Russia’s default and the collapse of LTCM triggered the 1998 shock, even as US economic growth held up. The current uncertainty about data centre approvals, following the Democrats’ sweep of both chambers of Congress, similarly constitutes a political shock, not the disappearance of AI demand. Next, unlike the companies lacking earnings support in 2000, today’s technology leaders generate real revenues and cash flows and have robust balance sheets. Finally, postponing data centre construction does not remove the need for computing power. Capacity constraints could, in fact, strengthen the economics of assets already in place.

The vast spending commitments and uncertain returns are the most troubling echoes of 2000, Goldman Sachs conceded. Even so, its preferred interpretation is that investors are liquidating positions in a healthy growth story facing obstacles around financing schedules and construction timelines, rather than uncovering weaknesses in its underlying economics.

For investors, the distinction has a clear implication: should the 1998 parallel prove valid, comparable liquidity shocks have historically given rise to buying opportunities. Yet Goldman Sachs made an important qualification elsewhere. Federal Reserve interest rate cuts enabled the swift recovery in 1998; no equivalent support exists for the market today. That absence is the biggest difference between the historical template and present conditions.

In Goldman Sachs’ view, forced selling by leveraged ETFs before the close intensified the falls in semiconductors and the Nasdaq. Dealers’ negative gamma exposure, meanwhile, set off a liquidation cycle that fed on itself. More detailed estimates came from Nomura’s Charlie McElligott: leveraged ETF rebalancing generated approximately $47 billion in sell orders ahead of the close, volatility control funds accounted for around $61 billion, and dealers’ negative gamma exposure added approximately $7 billion in selling pressure for every 1% decline in the market. Across these channels, mechanical selling comfortably surpassed $100 billion.

Positioning, however, explains why the liquidation escalated from a “technical correction” into a “stampede”. Goldman Sachs’ prime brokerage data showed no underlying buying demand from long-only institutions. Instead, multi-strategy funds, quantitative funds, long-only institutions and retail investors were selling together. The difficulty lay in “Everyone being on the same side”.

Execution activity described by John Flood of Goldman Sachs’ trading desk reinforced that picture. The session on November 4, the day after the election, ranked among Goldman Sachs’ busiest of the year. Having cut risk before the election, long-only investors were surprised by Wednesday’s rally and had to chase rising prices to rebuild their holdings. By November 5, they were selling the hyperscaler stocks purchased only a day before, alongside hedge funds. For a second successive session, brokers with substantial retail client bases reported a pronounced tilt towards selling. At Goldman Sachs, there was “almost no real buying”. Correlations were rising, the momentum factor was underperforming, selling in the residual leverage factor persisted, and investors sought low-beta and defensive assets. Taken together, these signals were far more consistent with forced de-risking than with a reassessment of valuations based on fundamentals.

The fall was magnified further by “Front-running”. Morgan Stanley’s midday estimate put mechanical sell orders ahead of the close at approximately $52 billion. Once that estimate circulated quickly across Wall Street, discretionary traders opted to sell first, planning to repurchase at the “usual low” between 2 pm and 3 pm. That low did not arrive. Instead, the front-runners added to the downward pressure, leaving the market to finish at its lowest point of the day.

How Does Goldman Sachs Assess the Political Shock?

Notably, Goldman Sachs did not read the Democrats’ win as an ideological repudiation of AI. In assessing the political shock, it specifically referred to remarks by Texas Governor-elect Gina Hinojosa: “We can achieve economic development while protecting Texans. The two are not mutually exclusive,” as well as her explicit assertion that “I do not hate data centres.” For Goldman Sachs, these comments point more towards a regulatory reset than ideological opposition, leaving scope for actual spending to surpass the expectations embedded in current market discounts.

Political developments have nevertheless altered the basis on which markets price assets, Goldman Sachs acknowledged. Its trading desk explicitly observed that discussion of the AI trade now centres on “the speed and cost of deployment”, having previously focused on “long-term demand for AI computing power”. Sentiment has been weighed down by a combination of higher financing costs, political pressure on data centres and regulatory concerns after the election. During the sell-off, Goldman Sachs cautioned that approximately 45% of planned US data centre capacity expansion could face moratoriums following the Democrats’ “blue tsunami”.

On this reading, investors were pricing in “a longer timeline for AI infrastructure construction” on November 5, not “the disappearance of AI demand”. Where valuations depend on future cash flows, delays alone justify lower valuations. That is, however, a fundamentally different disruption from demand dropping to zero.

To explain the concerns behind the selling, Goldman Sachs organised them into four categories: too little capital to finance construction on the required scale; too little revenue to justify the investment; inadequate power and permitting capacity to enable computing expansion; and an overall spending path that cannot be sustained.

Of these, power and permitting constraints, the third category, currently attract the most market attention, according to Goldman Sachs. Yet it regards this as the least significant of the four. Supply restrictions do not, in its reasoning, imply that demand has vanished. What ultimately matters is whether revenues justify the capital committed.

The pricing focus therefore moves away from “political noise” and returns to “revenue validation”. While political shocks make timelines uncertain, the eventual delivery of revenues will decide whether the comparison with 1998 remains valid.

What Could Invalidate the 1998 Comparison?

Beyond diagnosing the sell-off, Goldman Sachs set out three requirements for a change in the market narrative and two distinct developments that would overturn its assessment. Rebuilding the narrative would require a positive political feedback loop from AI investment, greater stability in financing markets, and tangible evidence that underlying spending and demand are holding up.

The 1998 parallel would fail under either of two developments. A material delay in AI capital expenditure, combined with revenues that undermine the investment thesis, would make 2000 the more relevant precedent. Alternatively, if refinancing failures surface first in funding markets, the downturn would take on a fundamentally different character.

On the trading side, early dip-buying was already evident to Goldman Sachs’ desk during the sell-off. It confirmed interest in buying pullbacks in Nvidia, TSMC, Microsoft and CoreWeave. However, the desk also stressed that “it is difficult to know when selling pressure will ease, and the rebound when it does could be sharp”.

Rather than the “AI bubble” explanation, Goldman Sachs’ review advances a different framework for the November 5 plunge. It identifies the principal driver as position liquidation caused jointly by forced leveraged ETF selling, negative gamma feedback loops and a lack of demand from long-only institutions, not an abrupt collapse in underlying AI demand. Uncertainty about data centre approvals after the Democrats’ congressional sweep supplied the political catalyst. The resulting disruption lengthened construction timelines without removing demand itself.

This framework helps reconcile two facts that appear at odds. Falls of 12% in semiconductors and 6.4% in the Nasdaq 100 were indeed crisis-level moves. However, Goldman Sachs’ cash equities desk recorded cumulative net technology selling of $27 billion, while positioning showed that “everyone was on the same side”. Together, these observations suggest the mechanics of a liquidity event rather than a fundamental shock.

There is nevertheless an explicit limit to the framework. The comparison with 1998 implicitly depends on the idea that “a backstop will arrive”; at that time, Federal Reserve rate cuts provided it. Goldman Sachs recognises that no such backstop is available today. Even a correct diagnosis would therefore not mean that the pace and course of recovery from this liquidity shock simply follow the 1998 playbook. The requirements Goldman Sachs has laid out for recovery remain clear: positive political feedback, more stable financing markets and tangible revenue evidence.