Five market signals on colored stones, in five minutes.

On September 25, Gemfields told the market to expect a net loss of $73.5 million for the first half of 2026. On September 30, its reviewed interim report put the number at $98.5 million. Same company, same six months, five days apart, and a $25 million gap that none of the coverage we read explains.

That is the kind of detail this newsletter exists to catch. A profit warning is management's forecast. The report is the accounts. Where they differ, the accounts win.

Two corrections to last week. First, last week's edition carried $73.5 million as the loss. The final figure is $98.5 million, with a pretax loss from continuing operations of $102.0 million. Second, last week's edition called the September 30 release "audited." Gemfields titles it a reviewed interim report. In a newsletter about paperwork, the word matters.

Key Takeaways

→ The loss landed $25 million wider than the warning. Net loss was $98.5 million, against $73.5 million flagged on September 25 and $24.6 million a year earlier. The $125.2 million non-cash impairment against Montepuez is unchanged, and the restated 2025 impairment stands at $65 million, up from $35 million (Gemfields; National Jeweler; Mining Weekly).

→ The business beneath the write down is better than the headline, and thinner than it looks. EBITDA was $40.7 million against a $4.9 million loss a year ago. Free cash flow before working capital was $17.4 million. Net debt was $44.2 million, before $33.3 million of auction receivables collected after June 30. Most of the revenue gain traces to one deferred sale: Montepuez's $76.1 million was $53 million from the February auction, moved from December, plus $23 million from the first Trade Select auction in June (Gemfields; National Jeweler; AJ Bell).

→ Montepuez mined far more and recovered fewer premium rubies. Ore mined rose 133% in the half, while premium output came in 13% below the first half of 2025. Premium grade is stabilizing near 0.025 carats per tonne at Mugloto and around 0.04 at Maninge Nice. Management said it is too early to call this a sustainable improvement (TipRanks call summary; AJ Bell).

→ A new target points at volume. On October 5, Gemfields set Montepuez a goal of processing 11.39 million tonnes of ore over three years to June 2029, against 1.39 million tonnes in 2025 and 1.21 million in the first half of 2026. Twenty percent of executive share awards are tied to it, among other metrics (BusinessDay).

The index tracks the 8 reference stones every week. Each row holds until a new dated transaction moves it. A reviewed report is not a per carat price, so every row holds. The ruby row stays at the June Trade Select average of $66. The $23 million June figure in the report is the revenue behind that average, not a new print.

Auction prices realized, not retail. Sources: Christie's, Sotheby's, Gemfields, Bonhams, GemGuide, Pala International.

Notable this week: the sapphire row has not moved since June 9, four months ago. That is a fact about the calendar, not about demand. Rows move on dated prints, and none has landed.

What the two numbers say together. Ore mined up 133%. Premium output down 13%. Put them side by side and the implied premium yield per tonne fell by roughly 60 percent (0.87 divided by 2.33 is about 0.37). That is our arithmetic, and it assumes ore mined is the right denominator, so treat it as a direction, not a measurement. The mine's own reported grades are the better guide.

Why a busy mine can be a poor one. More tonnes at a lower premium grade means each premium carat carries more cost. The second processing plant is largely operational, with full commissioning expected in the fourth quarter. Throughput helps only if grade holds. A plant that processes more rock at the same grade makes more rubies. A plant that processes more rock at a lower grade makes a bigger bill.

Read the new target against the old pace. Eleven point three nine million tonnes over three years averages about 3.8 million tonnes a year. The first half of 2026 ran at 1.21 million, which is roughly 2.4 million annualized. The target sits about 57 percent above that pace, and about 2.7 times 2025's volume. Those are our calculations from reported figures.

What we cannot see. The coverage we found says tonnes is one of several metrics behind the share awards. It does not say whether premium grade is another. On the call, management said grades above 0.03 premium carats per tonne would materially improve the financial model (Investing.com transcript). Mugloto's 0.025 sits below that line. Maninge Nice's 0.04 sits above it. Which pit feeds the plant may matter more to the accounts than how many tonnes go through it.

1. A warning is not the accounts. The September 25 figure was $25 million short of the September 30 figure. Use trading updates as direction and wait for the report before quoting a number.

2. A volume target and a grade target are different promises. Tonnes are easy to count and easy to hit. Premium carats per tonne is the number the impairment says the mine is short of. Ask which one a company is measuring itself on.

3. Cash can be healthy while the asset is written down. Positive EBITDA and a $125.2 million impairment are both true. The first describes what the mine earned. The second describes what management now expects it to earn.

→ Gemfields' two remaining ruby auctions. Management plans two ruby auctions and one high quality emerald auction before year end. We found no dates in the coverage. The ruby sales are the next test of whether the grade story shows up in a basket.

→ Final commissioning of the second plant. Expected in the fourth quarter. Watch premium carats per tonne, not throughput.

→ Kagem. About $30 million of revenue was already secured from the September commercial quality emerald auction, per the call summary. The high quality sale is the one that moves the emerald row.

Open any producer's latest report and find two numbers: tonnes processed and carats recovered. Divide one by the other, then do the same for the previous period. If tonnes rose and the ratio fell, the company is working harder for less. This week, that was the whole story in one division.

— Carat Brief

Carat Brief is editorial. Nothing here is investment, legal, or tax advice. Gemstones are illiquid and require certified verification before purchase. Work with credentialed labs: GIA, AGL, SSEF, Gübelin, Lotus Gemology.

Sources: Gemfields reviewed interim report for the six months to 30 June 2026 (published 30 September 2026); National Jeweler; Mining Weekly; AJ Bell; TipRanks; Investing.com earnings call transcript; BusinessDay (5 October 2026). Derived figures are marked as our arithmetic.