Spot gold retreated 1.2% to $4,066 per ounce and gold futures dropped 1.5% to $4,078.10, as an AI-driven equity selloff forced broad position liquidation, tightened liquidity conditions, and pushed the US dollar higher — the classic cocktail that pressures gold even when the underlying macro case for the metal remains intact. The trigger was the market disruption set off by DeepSeek, the Chinese AI model whose capabilities rattled assumptions about US technological dominance and sent Nvidia and other AI-adjacent equities sharply lower.

Gold's correlation with equities in stress episodes is well understood by practitioners: when portfolios bleed mark-to-market losses fast enough, liquid assets — including gold — get sold to meet margin calls or rebalance risk. That dynamic played out here. The dollar's concurrent bid compounded the move; gold is priced in dollars, so dollar strength mechanically reduces the metal's purchasing power for holders in other currencies, softening demand at the margin.

The DeepSeek episode sits inside a longer bull run that context requires acknowledging. Gold gained 12% in 2023 and a further 29% in 2024, per CME Group, with Asian physical demand and a growing derivatives market amplifying the move. At spot prices above $4,000 per ounce, the metal was already pricing in a substantial risk premium — geopolitical tension, central bank accumulation, and sticky-inflation hedging demand among them. A 1.2-1.5% single-session correction from those levels is notable but, in percentage terms, modest relative to what the equity side absorbed.

The AI Disruption Channel

The mechanism linking DeepSeek to gold prices runs through equities, not any direct commodity fundamental. BullionVault documented the market tensions as Chinese AI competition challenged US tech dominance, with Nvidia bearing the sharpest immediate losses. That repriced AI capex assumptions across the sector: if frontier models can be trained at a fraction of the cost implied by current GPU demand, the investment thesis underpinning the US hyperscaler trade weakens.

The spillover into gold via the liquidity-squeeze and dollar-strength channel, rather than any flight-to-safety bid, is the more interesting tell. In a genuine risk-off episode driven by macro fear, gold typically catches a bid as equities fall. Here it didn't — which signals that the selling pressure was liquidation-driven rather than sentiment-driven. Investors were closing longs, not rotating into cash or Treasuries with gold as a safe haven destination.

What Capital Economics Makes of the AI Trade's Durability

Capital Economics published its view in December 2025 that the AI equity rally was not finished — that it expected the trade to extend through 2026. That call has to be read alongside a separate warning the firm issued a month earlier: the growing use of debt and special purpose vehicles to fund AI projects represents a leverage risk worth monitoring closely, given that excess leverage has a well-documented tendency to become disorderly when sentiment turns. The two positions are not contradictory — bubbles can inflate further before the leverage that sustains them becomes the problem — but the gap between them is where the risk lives.

For gold, the implication is straightforward. If the AI equity rally resumes and the dollar holds its recent gains, gold faces a near-term headwind. The metal performed well during 2023-24 partly because real yields, while rising, were offset by central bank buying and geopolitical risk premia. If AI-driven productivity gains eventually translate into disinflationary growth — a path some economists are mapping — the inflation-hedge component of gold's valuation comes under pressure too.

None of that changes what happened on the day. Long positions were closed, the dollar rose, and gold fell 1.2-1.5% depending on whether you track spot or futures. The DeepSeek disruption was the proximate trigger. The structural question — whether gold at $4,000-plus has fully priced the macro risks that drove it there — is the one worth watching as AI's economic footprint becomes clearer over the next twelve months.