Gold just had its best month since the January record, and it barely dented the most bullish call on Wall Street. Spot gold fell to about $4,336.30 an ounce by 2 September 2026, a four-week low that erased most of its 2026 gain, so JPMorgan's $6,000 year-end target now needs a 38% rise in under four months, while Goldman Sachs' $4,900 needs about 13%.
The gap between those two numbers is about $1,100, roughly a quarter of the entire price of the metal. Forecasts that far apart usually mean the outcome hangs on one variable. This time there are two, and they are pulling in opposite directions.
What are the banks actually forecasting?
The two houses have different targets because they use different machinery. JPMorgan's framework treats quarterly demand tonnage as the primary price driver, and the bank estimates that relationship explains roughly 70% of the quarter-on-quarter change in the gold price. On that model, sustained central bank and investor demand does most of the work to reach $6,000, with $6,300 following by end-2027.
Goldman's number runs through rates instead. Goldman Sachs does not expect the Federal Reserve to cut rates at all in 2026 and has pushed its first expected cut to June 2027, which removes the tailwind behind most of gold's historic moves. Gold pays no yield, so it competes with cash, and cash paying above 4% is a real opponent.
Did central bank buying come back?
Emphatically, and this is the single biggest change since the last time we looked at this. Central banks bought a net 288.9 tonnes of gold in the second quarter of 2026, a 62% jump from 177.9 tonnes in the same quarter of 2025, after a first quarter revised down to just 57 tonnes. The pillar the bull case rests on did not break. It went quiet for one quarter and then came back louder.
The buyers are the usual names with one addition at the top. The National Bank of Poland was the largest buyer at 51 tonnes, taking its reserves to 632 tonnes by end-June, while the People's Bank of China added 33 tonnes and the Bank of Russia sold the most at 22 tonnes. The World Gold Council's own survey found 89% of central bankers expect global gold reserves to rise over the next 12 months.
One caveat keeps this honest. First-half 2026 central bank demand of 345 tonnes is still the lowest for any first half since 2022, so the Q2 rebound restores the trend without yet restoring the 1,000-tonne-a-year pace of 2022 to 2024. The structural case our central banks buying gold piece lays out is intact, just running at a slower cruising speed.
One bank needs gold to rise 38% in four months. The other needs 13%. Both are describing the same metal in the same year.
What is actually driving this rally?
Three separate fears, arriving in the same week. The US national debt crossed a record $40 trillion, five months after passing $39 trillion, which is the kind of number that sends institutional money toward assets with no counterparty. The US Treasury then announced it would at least double its long-dated debt buybacks to contain borrowing costs, which knocked yields and the dollar sharply lower and lifted gold more than 4% in one session.
Then the whole thing reversed. Fed chair Kevin Warsh sharpened his inflation warning at Jackson Hole on 28 August 2026, renewed US strikes on Iran pushed crude toward $97, and September rate-hike odds jumped from about 35% to roughly 66%, lifting real yields and the dollar and knocking gold down about 7% in nine days. The mechanism is set out in our why gold fell when the war restarted piece.
Why is the Fed still the problem for gold?
Because the debate is no longer about how fast rates fall. The Federal Reserve under chair Kevin Warsh is weighing whether to raise rates at its 16 September meeting, with CME FedWatch pricing roughly a 68% chance of a hold (as of 20 August 2026) and the rest on a hike. Odds of a September hike had run near 50% in early August before soft jobs, CPI and producer price readings knocked them back toward 31%.
That is a very different world from the one most gold forecasts were written in. The August jobs report published on 4 September 2026 showed US payrolls rising 162,000 against expectations of 56,000, with July revised from a 23,000 fall to a 21,000 gain and unemployment steady at 4.1%, which removed the weak-labour-market argument that gold bulls had been leaning on and pushed September rate-hike odds to roughly 63 to 65%.
Warsh speaks at the Jackson Hole symposium on 28 August, his first as chair, 19 days before that decision. He has said his remarks will focus on long-term structural questions rather than near-term guidance, which will not stop every gold desk from trading each sentence.
Why this matters for Indian buyers
Indians do not buy gold in dollars, which changes the arithmetic. The Indian price stacks import duty, GST and the rupee on top of the international rate, and the rupee slipped to about 95.76 per dollar (as of 20 August 2026), a three-week low on higher oil and elevated US yields. A flat dollar gold price with a falling rupee still means a more expensive wedding.
That has a behavioural effect every year at this point in the calendar, with the festive and wedding season approaching, covered in our buying gold this festive season piece. High prices push demand toward lighter jewellery, old-gold exchange and financial forms of gold, which our how to invest in gold in India guide and the gold rate in India page both cover. Anyone weighing metal against the other 2026 hedge should read our gold vs bitcoin comparison, since the two have diverged sharply this year.
What to watch between now and December
The September FOMC decision is the variable that decides which bank is right. A hike would validate Goldman's rates-first framework and cap the metal; a hold followed by softening data revives the JPMorgan path.
Quarterly central bank tonnage is the second signal, and it has now turned back up. Another quarter near 289 tonnes would put annualised demand back in the range that carried gold through 2022 to 2024.
Exchange-traded fund flows show whether investors are still adding. Rallies that run without ETF inflows tend to be thinner and reverse faster than those with them, and August's move needs confirmation on this front.
Risks to monitor
The second risk is positioning. After a month like this one, gold carries a crowded trade and a lot of recent buyers with short holding periods, which makes any sharp reversal faster than the fundamentals justify. The 20 August session already showed this, with prices dropping 0.9% on nothing more than profit-taking, before the metal added 2% the very next day.
The third is that geopolitical premiums decay. Much of the current price includes a risk premium for the Middle East, and premiums built on headlines unwind when the headlines stop. This is general information, not investment advice.
The honest read is that both banks are describing scenarios, not forecasts anyone should trade against. What makes the $1,100 gap interesting is not who wins it, but the reminder that on the most-watched asset of 2026, the best-resourced research desks in the world cannot agree on a number within a quarter of each other.