Two headlines collided Wednesday. Goldman Sachs published a note reaffirming its year-end gold target of $4,900 per ounce, citing relentless sovereign demand and a People’s Bank of China it believes is buying far more bullion than Beijing admits. Within hours, the flash S&P Global US Composite PMI jumped to 58.4 in September from 56.0 in August. Spot gold fell to around $4,290.
The question is not which headline is right. Both are. The question is whether the yield spike that followed the PMI is a trend-ender or a dip.
What Goldman Actually Said
Goldman’s model puts the People’s Bank of China’s July purchases at 35 tonnes, roughly 75% more than the amount officially disclosed. The gap comes from how Goldman tracks the market: official reserve data often understates what central banks are buying, so the bank’s nowcast model also follows gold moving through London’s over-the-counter market into domestic vaults and third-party custodians.
Goldman’s fair-value forecast of $4,900/oz by end-2026 assumes continued strong central bank demand, with average purchases of 50 tonnes per month in 2026 and 40 tonnes per month in 2027, alongside a recovery in private investor ETF demand as the Fed remains on hold. Goldman continues to see net upside risk to its $4,900 target, but warned of greater two-sided volatility along the way.
The Real-Yield Problem
Wednesday’s PMI was not a minor beat. The September flash PMI showed the U.S. private sector expanding at the fastest pace in more than five years, with input-cost inflation accelerating to the quickest rate in nearly four years. The 10-year Treasury yield jumped to 5.058%, its highest level since July 2007, and the 2-year climbed to 4.874%.
For an asset that pays no interest, a 5% risk-free alternative is the most expensive competition it can face. The textbook playbook says to sell gold into real-yield surges. For most of the last two decades, that playbook worked.
It has not worked cleanly this cycle. By the pre-2022 model, real yields this high should cap gold. The fact that they have not is the regime question. The entry of sovereign buyers who are structurally indifferent to price creates a floor dynamic that is genuinely new in modern gold market history.
The data backs that framing. Central banks bought a record 289 tonnes of gold in the second quarter of 2026, a 62% jump from a year earlier. China extended its buying streak to 21 consecutive months through July. That is not tactical flow. That is reserve policy.
Where the Level Is
Gold recovered from its June low below $4,000 and spent most of August and early September trading between $4,350 and $4,500. Wednesday’s PMI-driven flush pulled it back to $4,280 to $4,310, roughly 200 points below the range midpoint.
The $4,280 to $4,300 zone held as support through the June recovery and represents the level at which sovereign-bid arguments are being actively tested. A close below $4,250 on a sustained basis would weaken the thesis meaningfully. Above $4,400 and the range reasserts itself.
Miners Over Metal Here
At $4,290 spot against an expected AISC of $1,680 at Newmont, the margin per ounce is wider than at any point in the last gold cycle. A gold miner’s earnings move faster than the metal itself, since a rise in bullion widens the gap between what it costs to produce an ounce and what that ounce sells for, and that widening spread lands directly on the bottom line.
Agnico Eagle’s latest quarterly results showed record free cash flow, with costs staying largely in check despite broader inflationary pressure across the industry. AEM ended the quarter with a net cash position of $3.267 billion as of June 30, 2026. For traders who want gold exposure without the full PMI-driven volatility in spot, AEM and NEM at current levels offer leveraged participation in any recovery toward $4,500 or above.
Trader’s Action Plan
The $4,280 to $4,300 zone is the line. Sovereign buying has defended every comparable flush since late 2025, and nothing in Wednesday’s data changes the structural demand picture Goldman quantified. A dip entry near $4,290 in GLD, with risk defined below $4,240, has a favorable structure against Goldman’s $4,900 target. Miners (NEM, AEM, GOLD) offer higher-conviction leverage on a recovery. The risk to respect: if the 10-year holds above 5% into October and the Fed signals another hike before year-end, the opportunity-cost pressure on gold ETF flows will intensify, and the $4,000 level comes back into play.
