Cash Flow Turned and the Cost Program Got a Number: Upgrading Rio Tinto to Outperform

Results vs. Consensus

Rio Tinto reports semi-annually, so the 29 July release covers the six months to 30 June 2026 and the Street's forecasting effort concentrates on half-year underlying EBITDA, underlying earnings, the interim dividend and closing net debt. This print beat on all four, and the beats were narrow on the earnings lines and wide on the balance sheet.

The more important point about the comparison is that guidance was not a variable. Every 2026 production, sales, unit-cost and capital-investment range had already been reaffirmed in the second-quarter operations review on 15 July, and copper C1 net unit cost guidance had been cut there to 30 to 50 US cents per pound from 65 to 75. Consensus had two weeks to reset before the financials landed. What 29 July actually delivered that 15 July had not was the cash statement, the dividend, the productivity run rate and the tax rate.

H1 2026 Scorecard

One caveat on the revenue line. It and the underlying EBITDA line are the two comparisons in the table supported by a single estimate rather than by multiple independent ones, and the basis is not stated. Rio publishes three different top lines: consolidated sales revenue of US$31,028M; reportable-segment revenue of US$32,618M, which adds the Group's proportionate share of equity accounted unit sales, principally Escondida; and the US$34,291M base used to compute the underlying EBITDA margin. A contributor modelling the segmental line lands close to the quoted estimate. We report the comparison and then set it aside, because for a diversified miner revenue is a weak signal in any case: realised prices, provisional pricing and US$1.3bn of Pilbara freight revenue all sit inside it.

Income statement, half on half

Cash flow and balance sheet

Revenue

Consolidated sales revenue of US$31,028M rose 15.5%, with almost all of the increase attributable to price. Copper's realised price rose 35% to 591 US cents per pound against an LME average up 39%, aluminium's realised price including premiums rose 39% to US$4,343 per tonne against an LME average up 33%, and lithium carbonate equivalent realised US$18,960 per tonne, up 22% against a spot index up 125%. Only the Pilbara was quiet: the realised iron ore price rose 2.3% to US$92.6 per dry tonne, and the Platts 61% index averaged US$92 per dry tonne FOB Western Australia in both halves.

By product, gold is the line that moved most in dollar terms relative to its size. Gold revenue rose to US$1,546M from US$693M, more than doubling on a 30% increase in mined gold volume and a benchmark price up 53%. Iron ore revenue rose to US$13,804M from US$13,184M, copper to US$4,052M from US$3,091M, aluminium, alumina and bauxite to US$8,912M from US$7,411M, and lithium to US$568M from US$308M. Industrial minerals fell to US$1,071M from US$1,200M. Greater China took 57.9% of consolidated sales revenue against 55.2% a year ago, and the United States 17.1% against 17.9%.

Two items inside revenue are worth flagging because they are not operating income in the ordinary sense. Provisional pricing adjustments added US$235M to copper revenue against US$266M a year ago, so this is a small year-on-year drag rather than a tailwind, and provisionally priced receivables closed at US$938M against US$1,431M at 31 December. And Pilbara segmental revenue includes US$1.3bn of freight revenue against US$0.8bn, a US$0.5bn increase that carries an offsetting cost and tells you nothing about the ore business.

Margins and EBITDA

Underlying EBITDA of US$14,826M rose 28.4% and the company-basis margin rose four points to 43%. The mix shift underneath is the most important number in the release. Copper's segment EBITDA margin rose to 66.3% from 50.0%, aluminium and lithium to 33.2% from 29.7%, while iron ore fell to 48.3% from 50.9%. Iron ore is no longer the highest-margin business in this portfolio and, for the first time, is not the largest contributor to reportable-segment EBITDA: it supplied 42.9% against 55.5% a year ago, and copper plus aluminium and lithium together supplied 57.1% against 44.5%.

The waterfall is again the most useful disclosure in the release, and this time it says something different from last year. Prices added US$3.6bn, led by copper up 39%, gold up 53% and aluminium up 33%, with iron ore resilient at a 2% higher realised price and bauxite the one offset. Exchange rates cost US$0.7bn as the US dollar weakened 11% against the Australian dollar and 2% against the Canadian dollar. General inflation cost US$0.4bn, including a US$0.2bn charge from updating the inflation assumption on closure provisions for closed or fully impaired sites, and energy cost US$0.3bn on supply disruption. Volumes and mix added US$1.2bn, of which US$0.3bn came from the productivity program and US$0.9bn from growth projects. Operating cash unit costs added US$0.3bn net, being US$0.5bn of productivity benefit against US$0.3bn of cost inefficiency on lower volumes at Kennecott, Escondida, Iron Ore Company of Canada and bauxite. Exploration cost US$0.1bn and non-cash items US$0.2bn.

Assessment: The controllable US$1.2bn is the number that matters for the thesis, and it is roughly the amount that would have been needed to hold EBITDA flat against the macro headwinds alone. That is not a criticism. It is the honest read on a half in which the cost program worked hard enough to neutralise an 11% adverse currency move, US$0.7bn of inflation and energy, and a diesel price that went from about US$85 to about US$140 a barrel. The problem with management's framing is not the direction, which is right, but the emphasis: on this half's numbers commodity prices did not merely "remain important," they supplied more than the entire increase.

Earnings and the gap between underlying and reported

Underlying earnings of US$6,851M rose 42.5%, faster than EBITDA at 28.4%, and net earnings of US$6,664M rose 47.2%, faster still. Both accelerations come from below the EBITDA line, and neither is an operating outcome. Depreciation and amortisation rose US$0.6bn pre-tax as AP60, the Oyu Tolgoi underground and Western Range commissioned. Interest and finance items improved by US$0.1bn. Tax on underlying earnings cost US$0.2bn more in absolute terms despite a much larger pre-tax base, because the effective rate on underlying earnings fell to 25.2% from 34.5%, reflecting the utilisation of previously unrecognised deferred tax assets and a change in the geographic spread of earnings. Non-controlling interests took US$0.5bn more as Oyu Tolgoi ramped, and the attributable-to-minorities line went to US$530M from US$8M.

The reconciliation between reported and underlying is unusually clean. Total items excluded from underlying earnings were US$187M against US$279M, and the company again reports nothing in its judgemental category in either period. The largest exclusion is US$271M of after-tax exchange losses on external net debt, intragroup balances and derivatives, itself a US$286M post-tax loss on intragroup balances offset by a US$15M gain on external net debt as the Australian dollar strengthened. There were no impairment charges, against US$86M a year ago, and a US$15M impairment reversal came through the share of equity accounted units at Porto Trombetas.

Assessment: Basic EPS at 409.9 cents is now only 2.7% below underlying EPS at 421.4 cents, against an 8.3% gap for FY25. The two measures have converged because the noise that separated them, principally impairments and closure-estimate revisions, did not repeat. That is worth noting but not worth extrapolating: a group carrying US$21.4bn of borrowings, an US$18.9bn closure provision and operations in Mongolia, Guinea, Argentina and Canada will keep generating currency and provision noise. The line to watch instead is the effective tax rate. At 25.2% it sits roughly five points below the approximately 30% full-year rate guided in December and reaffirmed in February, and no revised full-year figure was published with these results. If the second half reverts to the guided rate on a similar pre-tax base, roughly US$0.4bn of underlying earnings goes with it.

Segment Performance

Rio reports three product groups following the restructure announced on 27 August 2025, with Simandou, Rio Tinto Iron & Titanium, Borates, Diamonds and closed sites sitting outside them. Comparatives below are the company's restated figures, which also reflect the Gove refinery moving from central closure management into Aluminium & Lithium.

Product group summary

Reportable-segment revenue exceeds consolidated sales revenue because it includes Rio's proportionate share of equity accounted unit sales, principally Escondida. The reconciling item is US$3,263M for the half against US$2,528M a year ago. "Central and other items" above aggregates central pension, share-based payments, insurance and derivatives of positive US$214M, restructuring, project and one-off costs of US$368M, central costs of US$425M, central exploration and evaluation of US$110M and US$1M of inter-segment eliminations.

Production and price KPIs

Two lines in that table deserve to be read together with the guidance ranges rather than on their own. Pilbara unit cash costs of US$25.0 per wet tonne are already at the top of the full-year guided range of US$23.5 to US$25.0, and the range assumes an Australian dollar of 0.67 in a half where the currency moved 11% the wrong way and diesel rose about 65%. And copper C1 net unit costs were negative 24.7 cents per pound in the half, so the 30 to 50 cent full-year guide requires a very different second half. Production-weighting the two halves at the guidance midpoints implies roughly 113 cents per pound in H2, with a range of about 86 to 142 cents across the guidance corners. The guide was cut on 15 July, after the late-June Kennecott furnace breach, so it should already embed the loss of by-product credits and payable volume.

Copper

Copper delivered a record first half and did it on price, gold and Oyu Tolgoi rather than on copper tonnes. Production rose 1% to 442 kt on a consolidated basis, with a 31% increase at Oyu Tolgoi almost entirely offsetting declines at Kennecott and Escondida. Mined gold rose 30% to 250 koz. Segmental revenue rose 38.9% to US$8,622M and EBITDA rose 84.0% to US$5,713M, of which the company attributes about US$2bn to stronger copper, gold and silver prices. Operating cash flow rose 147.9% to US$3,910M, capital investment fell 9.0% to US$756M now that the Oyu Tolgoi underground project is complete, and free cash flow rose 324.4% to US$3,149M. That is despite the US$443M Mongolian tax payment made in March.

The productivity contribution here is specific and checkable. Drawbell development rates at Oyu Tolgoi rose 15%, which the CEO put at roughly US$80M of benefit, and molybdenum volumes rose at Kennecott.

"At OT, we have redesigned the way we approach underground development, harnessing data and speeding drawbell construction. This has accelerated production, helping to generate around $80 million in productivity improvements."
— Simon Trott, Chief Executive

Against that, both US assets had a difficult half. Kennecott lost production to a safety stand-down after a fatality in the first quarter, to geotechnical remediation after a rockfall, and then to a flash converting furnace breach in late June. Escondida delivered lower concentrate production on expected lower grade from mine sequencing, partly offset by better leaching performance on the refined side.

"Mining performance at Kennecott is expected to recover in the second half as geotechnical management activities conclude and access to planned mining areas is restored. However, following the furnace breach in late June, some metal sales and associated cash flows will shift into 2027, while remediation work is completed."
— Peter Cunningham, Chief Financial Officer

Assessment: This is the segment that justifies the equity story and it is now delivering at the cash line as well as the earnings line, with free cash flow more than quadrupling and capital intensity falling as Oyu Tolgoi shifts from build to ramp. The qualification is unchanged from February and is now visible in the guidance arithmetic rather than only in the framing: the full-year copper guide of 800 to 870 kt against 442 kt in the first half implies 358 to 428 kt in the second, 11% below H1 at the midpoint. Copper's contribution to 2026 earnings is front-loaded, and the second half will show it in both tonnes and unit costs.

Iron Ore

The Pilbara had an excellent operating half and the segment as a whole had a flat one, which is the clearest illustration in this release of how much the group has changed. Pilbara production rose 6% to 162.3 Mt on a 100% basis, the highest first half since the 2018 record, and shipments rose 4.7% to 157.7 Mt. Segmental revenue rose 4.1% to US$14,027M and Pilbara underlying EBITDA rose 5% to US$7.0bn. Yet product-group EBITDA fell 1.3% to US$6,769M, because Iron Ore Company of Canada production fell 22% on pit and asset health, and evaluation expenditure rose. Operating cash flow rose 8.6% to US$5,186M with Pilbara cash conversion improving to 77% from 70%, and free cash flow fell 4.6% to US$2,980M because capital investment rose 34.1% to US$2,139M.

The product-strategy reset that began in 2025 is now essentially complete. SP10 sales volumes fell 65% and represent 10% of total iron ore sales against 29% a year ago. Portside sales in China fell to 5.5 Mt from 16.3 Mt, and portside inventory closed at 6.1 Mt including 3.4 Mt of Pilbara product.

On costs, the segment absorbed a great deal and still landed inside the guided range. Pilbara unit cash costs rose US$0.7 per tonne to US$25.0, comprising a US$2.1 per tonne headwind from the Australian dollar and a US$0.8 per tonne impact from diesel, against which productivity and volume benefits recovered most of the gap.

"In the Pilbara, we've generated around $55 million in annual benefits by removing redundant capacity through stronger system resilience, building on the changes we made to product strategy."
— Simon Trott, Chief Executive

Assessment: The operating performance here is better than the segment result suggests, and the segment result is better than it looks once you separate the Pilbara from Iron Ore Company of Canada. But the unit cost sits at the very top of the guided range with half the year gone, and the range assumes a currency that has already moved against it. Management is guiding to full-year costs inside the range, which means the second half has to come in below the first. The other half of the story is that the Pilbara now consumes a 34% larger capital budget for three replacement mines that preserve capacity rather than adding it, with first ore in 2027 and Rhodes Ridge's feasibility study not concluding until 2029. Iron ore is a cash engine being maintained, not grown, until the end of the decade.

Aluminium

Aluminium had the cleanest half of any business in the portfolio. Segmental revenue rose 21.3% to US$9,401M and EBITDA rose 31.3% to US$3,093M on a realised price of US$4,343 per tonne, up 39% against an LME average up 33%, with the difference coming from regional premiums. Production was flat at 1.676 Mt as Kitimat, New Zealand Aluminium Smelters and the new AP60 line offset the closure of Arvida's final potlines in June. Bauxite recovered strongly in the second quarter after first-quarter weather, though first-half production was still 7% lower at 28.5 Mt, and alumina rose 8% with QAL now consolidated at 100%. The productivity program contributed US$130M here through contractor management at the smelters, cast house mix optimisation and throughput improvements.

The tariff table is the disclosure to read closely, and it now covers three halves.

Two derived facts sit inside that table. US-destination shipments have fallen 19.1% year on year while the total tariff bill has risen 140.8%, so the commercial response has been to re-route metal rather than to absorb the cost. And the Midwest premium duty paid, at US$2,406 per tonne, is now well above the realised tariff cost of US$1,322 per tonne, which is what "fully compensating" looks like in practice. The premium applies to approximately 40% of total volumes in the half, against 55% a year ago. Value-added product sales fell to 41% of primary metal sold from 46%, at a premium of US$355 per tonne against US$292.

Assessment: This franchise is doing exactly what it should in a policy-distorted market, and it is being paid for it. The caution is unchanged: a margin underwritten by a regional premium that has tripled in twelve months is a policy-contingent margin, and the 19% decline in US-destination shipments shows the physical flows adjusting underneath. Add the Yarwun curtailment of 40% of alumina production from October 2026 and the ramp of AP60 to full capacity by year end, and the second half aluminium result has more moving parts than the first.

Lithium

Lithium finally produced a number worth discussing. Segmental revenue rose 84.4% to US$568M and EBITDA rose 419% to US$218M, at a 38.4% margin against 13.6%, on a 53% volume increase and a 22% higher realised price. Part of the volume increase is arithmetic, because H1 2025 consolidated Arcadium only from March, but the operating news is real: first production was achieved at both Fénix 1B and Sal de Vida in the second quarter, ahead of plan, and the Rincon starter plant continued to ramp with a focus on reactor stability.

The cash position has not changed. Capital investment rose 45.1% to US$627M and free cash flow was negative US$551M against negative US$538M, essentially flat. Rincon's full-scale 60 ktpa expansion still has US$1.9bn to spend against a US$2.5bn total, with first production in 2028 and a three-year ramp. Nemaska's first production is planned for 2028. The two Chilean joint ventures with Codelco and ENAMI, which in February were expected to close in the first half of 2026, are now expected to close in late 2026 or early 2027.

Assessment: The trajectory is right and the delivery is ahead of schedule on two of four in-flight projects, which is more than we expected when we scored this pillar in February. The economics have not changed. This is still a business that consumed US$627M of capital in six months to generate US$218M of EBITDA, still burns roughly US$0.6bn of free cash flow a half, and still does not reach its 200 ktpa capacity target until 2028. The status improves from a standing start; it does not clear.

Outside the reportable segments

Other operations, which contains Rio Tinto Iron & Titanium, Borates, Diamonds and closed sites, swung to a US$229M EBITDA loss from US$78M of profit, a US$0.3bn deterioration driven principally by changes to closure assumptions including the inflation rate applied, a weaker US dollar, soft titanium dioxide feedstock demand and the Diavik wind-down, partly offset by US$0.1bn of productivity benefit. Restructuring, project and one-off costs rose to US$368M from US$320M, covering the operating-model change, a major ERP upgrade and other corporate projects. Central costs were flat at US$425M, with US$0.1bn of savings from streamlining central functions about half offset by currency. Simandou, still reported outside the segments, recorded a US$48M EBITDA loss on US$68M of first revenue.

Assessment: The below-the-segments block cost US$967M of EBITDA this half against US$817M, so the drag continues to widen even as the segments improve. Most of the increase is closure-provision arithmetic rather than operating deterioration, and the closure provision itself rose US$1.05bn to US$18.9bn, of which US$0.5bn came from the inflation assumption change. That provision is now equal to 29% of equity attributable to owners, and it compounds: the amortisation of the discount on provisions cost US$444M pre-tax in this half alone. It is not a thesis point yet. It is the largest slow-moving liability in this balance sheet and it deserves a line in anyone's model.

Key Topics & Management Commentary

Overall Management Tone: Management arrived with evidence rather than framework, which is a change from February. Where the last call answered questions about the cost program with methodology, this one answered with a banked figure, a run rate and a year-end target, and volunteered a divestment deadline that February had refused to give. The confidence was earned on the operating lines and thinnest on the second-half shape, where almost every guidance range implies a step down from the first half and none of that was addressed unprompted. Tone was the most assured of the three calls we have now read from this team, and the least forthcoming on what the second half actually looks like.

1. The cost program stopped being a promise

In February the company would only say that 2026 cash delivery would be "materially above" a US$650M run rate, and would not size the multi-year phase. That was the reason we scored the productivity pillar Neutral rather than Confirmed at initiation. Six months later there is a number, and it is roughly three times the December figure.

"When I launched our program to build a stronger, sharper, simpler Rio Tinto at Capital Markets Day last December, I told you I'd deliver $650 million in productivity benefits, and we've delivered. We've already banked $870 million to the end of June. I also told you that we would continue to grow the program. Again, we've delivered. Today, I can announce we are targeting a year-end run rate of $1.8 billion, almost triple where we were just 7 months ago."
— Simon Trott, Chief Executive

The US$870M banked splits into roughly US$0.5bn of unit cost improvement and US$0.3bn of volume uplift, and both halves are traceable into the EBITDA waterfall rather than sitting on a separate slide. The annualised run rate reached US$1.3bn inside the half. Management was explicit that this is not a budget exercise.

"This is not a top-down exercise where we simply squeeze budgets. It's a structural change with more than 80 large initiatives running at every level of the business."
— Simon Trott, Chief Executive

Assessment: This is the change that moves our rating. A program that beat its own target by 34% at the banked level and nearly tripled its exit ambition in seven months has earned the benefit of the doubt on the second phase, and the fact that the benefits reconcile into the waterfall rather than living on a bridge slide is what separates this from the usual cost-out announcement. The remaining gap is that a US$1.8bn year-end run rate is an exit figure, not a 2026 cash number, and the second-half increment from US$1.3bn to US$1.8bn is the part still to be delivered.

2. The cash statement turns, and turns a year early

Our loudest objection in February was that the dividend was being funded from the balance sheet through the capital-spend peak. In FY25, free cash flow of US$4,025M supported US$6,145M of cash dividends, a 153% cash payout, and net debt rose 162%. This half the company funded US$5,037M of capital investment and paid US$4,211M of dividends, and net debt still fell US$301M to US$14,061M. Gearing fell to 16% from 18%.

"The earnings uplift has translated directly into cash with free cash flow rising by 75%. And even with our increased capital investment, we were able to reduce net debt during the period."
— Peter Cunningham, Chief Financial Officer

Two qualifications belong with that. The 75% figure uses a free cash flow definition Rio adopted this period, substituting its share of capital investment for gross purchases of property, plant and equipment. The change is defensible, because it credits Rio only with its economic share of Simandou and Nemaska capital after partner funding, but it flatters the current half by US$703M against US$223M for the restated comparative. On the prior basis, free cash flow rose 59.6% to US$3,131M and the cash payout was 134% rather than 110%. Second, the half also absorbed a US$1.6bn working capital outflow that included the US$443M Mongolian tax payment, so the underlying cash generation is better than the headline rather than worse.

Assessment: On either definition this is the largest improvement in the release, and it happened while capital investment rose 11.8% and the dividend rose 43%. The bear point does not close, because cash dividends still exceed free cash flow on both bases and will do so again in the second half when the interim is paid. But the shape has changed from a structural gap into a timing one, and it changed in the first year of a two-year investment peak rather than after it.

3. The dividend: a 50% interim by convention, up 43% by arithmetic

The interim ordinary dividend of 211.0 US cents, US$3.4bn in aggregate, is a 50% payout on underlying EPS of 421.4 cents. That is the same payout ratio as last year's interim, so the entire 43% increase is earnings. The full-year payout has been 60% for ten consecutive years.

"In line with our usual practice at the interims, we're declaring a 50% payout for the dividend, delivering a 43% increase to our shareholders."
— Peter Cunningham, Chief Financial Officer

"We paid out 60% of underlying earnings for 10 years. This provides you with cash flow today while keeping us disciplined with how we deploy residual capital."
— Peter Cunningham, Chief Financial Officer

The arithmetic that follows is mechanical rather than predictive. If the second half repeats the first half's underlying EPS, full-year underlying EPS would be about 843 cents, a 60% full-year payout would be about 506 cents, and the implied final dividend would be about 295 cents against the 211 cents interim. On the 29 July close that is a full-year yield of roughly 5.4%, on an assumption that the second half matches the first, which the volume guidance says it will not.

Assessment: The February debate about whether 60% is a floor or a ceiling has not moved, and management again declined to reopen the band. What has moved is the cash the policy is drawn against. A 60% payout on earnings 42% higher is a far less uncomfortable claim on the balance sheet than the same policy applied in FY25, and this is the first half in three where the dividend comes close to being covered by the cash the business actually generated.

4. The waterfall says price, management says self-help

The most interesting tension on the call is between the CFO's characterisation and the CFO's own chart. He opened by acknowledging the price contribution and then immediately qualified it.

"We've delivered a step change in our financial performance this half, supported by stronger commodity markets, particularly copper and aluminum, which now represents almost 60% of our EBITDA. However, this was not just a price story. As Simon mentioned, our productivity program is delivering."
— Peter Cunningham, Chief Financial Officer

"The broader point is that while commodity prices remain important, creating value for shareholders is increasingly within our control. It will be driven by improving operational performance, delivering our growth projects successfully and maintaining disciplined capital allocation."
— Peter Cunningham, Chief Financial Officer

Both statements are true as statements of direction. Neither describes this half's arithmetic, in which prices contributed US$3.6bn against a total EBITDA increase of US$3.3bn, and everything else netted to negative US$0.2bn. On the company's own preferred aggregation, external factors supplied US$2.1bn and controllables US$1.2bn, so roughly two-thirds of the increase came from outside the business.

The CFO did draw a useful distinction inside the headwinds, separating what should reverse from what should not.

"It's important to distinguish between those that are persistent, such as general price inflation and those that are more temporary in nature, such as higher diesel and raw material prices following Middle East supply disruptions. We would expect the latter to reverse over time and therefore, class them as temporary and one-off."
— Peter Cunningham, Chief Financial Officer

Assessment: The useful way to read the waterfall is not as a scorecard on management but as a measure of how much self-help it took to stand still against the macro. US$1.2bn of controllable benefit was roughly the amount required to neutralise an 11% adverse Australian dollar move, US$0.4bn of inflation and a diesel price that rose about 65% during the half. That is a good outcome. It is also a reminder that the productivity program is currently buying offset rather than upside, and that the upside only shows up when the macro stops taking it back.

5. Simandou converts from a capital commitment into a ramp

Simandou achieved first high-grade iron ore sales in April, and the release records 0.4 Mt sold at an average grade of 65.8% iron against 2.2 Mt shipped to China in the half. The gap is inventory: 7.6 Mt of uncrushed ore sat at the mine at the end of June, part of 9.6 Mt across the whole system including the Guinea port, ships at sea and Chinese ports, reflecting the two to three month lag between mine gate and sale. The SimFer mine is about 77% complete and the port about 85%, commissioning of the common rail infrastructure completed in the first quarter, first ore through the primary crusher is expected in the fourth quarter of 2026, port commissioning in the first quarter of 2027, and the ramp toward full production rates during the second half of 2028. The Rio share of remaining capital is US$1.7bn against a US$6.2bn total. The workforce stands at 19,460 with 76% Guinean participation.

"Simandou is now more than 3/4 complete. OT continues to ramp up and is achieving record production. Lithium in-flight projects are advancing and the Rhodes Ridge study is progressing on track."
— Simon Trott, Chief Executive

Assessment: In February the concern was that guidance of 5 to 10 Mt of 2026 sales had been reaffirmed 48 hours after a fatality stopped all site works, and that it was the number most exposed to the safety review. Half the year has gone and 0.4 Mt has been sold, leaving 4.6 to 9.6 Mt to land in the second half. That is not implausible with 9.6 Mt already sitting in the system and a two to three month sales lag, but it means the guide converts almost entirely on second-half shipments, and it makes the 2026 Simandou number the most binary line in the volume guidance. The construction schedule, by contrast, is now the least concerning part of this project.

6. Kennecott had two accidents and one of them moves cash into 2027

Kennecott lost a colleague in the first quarter, triggering a safety stand-down, then suffered a rockfall requiring geotechnical remediation, then a flash converting furnace breach in late June. The smelter continues to produce marketable copper matte, so full-year total production including copper contained in matte is unchanged, but refined copper and gold production will be lower in the second half and some metal sales and the cash behind them shift into 2027 while remediation proceeds.

"In terms of KUC, and I want to start by talking about safety. Obviously, fatality there earlier in the year, significant impact on the business and the team and really a moment in that business to reflect on where we were and what we needed to do to make sure that, that business operates safely."
— Simon Trott, Chief Executive

The forward story at Kennecott is more constructive. The Apex life extension would take the operation into the 2040s from a current end date of 2032, and the decision is described as close.

"So progressing at pace. We're well into that work. And so that decision will be coming in the not-too-distant future and extends it out into the 2040s. And so looking very promising."
— Simon Trott, Chief Executive

Assessment: Kennecott has the least predictable operating record in a copper portfolio the equity story now depends on, and it has had three separate interruptions inside twelve months. The furnace breach is the one with a cash consequence, and it lands in a second half already guided to lower copper volumes. Against that, an Apex approval extending the mine into the 2040s, attached to one of only two copper smelters in the United States, would be a strategic addition rather than a maintenance decision. That final investment decision is the thing to watch across the next two prints.

7. Capital release finally has a deadline, and still has no signature

The US$5bn to US$10bn capital release program from the December Capital Markets Day was reaffirmed, with the incremental disclosure that around US$5bn of it is being progressed for delivery by the end of 2026. On the call the CEO went further and put a commitment on announcements rather than on completions.

"No, progressing. I mean you've seen really strong cash generation today. The balance sheet is in good place and probably refer to my comments around capital discipline and efficiency. The divestment program, $5 billion to $10 billion tracking. We'll make decisions about that and make sure that we get full value. And so targeting $5 billion of announcements this year as part of that broader program."
— Simon Trott, Chief Executive

Nothing has been signed. The strategic reviews of Rio Tinto Iron & Titanium and Borates that were at market testing in February remain unresolved, production guidance for both is still withdrawn, and the release describes the mechanisms generically as portfolio management, infrastructure and other options.

Assessment: In February the framing was patient, systematic and explicitly not under pressure, which we read as removing the timetable altogether. A stated target of US$5bn of announcements before year end is a real change, because it is a date a management team can be held to and because the CEO volunteered it in Q&A rather than burying it in the deck. It is still not a transaction. We continue to assign the capital release no value in the base case, but the probability of a signed deal inside twelve months is materially higher than it was, and the balance sheet needs the proceeds less than it did.

8. Safety got worse in a half where everything else got better

Rio lost two colleagues in the first half, at Simandou and at Kennecott. The all-injury frequency rate was 0.40 against 0.37 for FY25. The CEO opened the presentation on it, ahead of any financial content, and the group launched a new Management Operating System on 1 July covering safety, risk and standards, people and leadership, and planning and performance.

"And this half, we lost 2 of our colleagues, and I'll carry that with me. Nothing we report today means anything if our people do not go home safely. Safety is my first priority. It is Rio's first priority, and it will always be."
— Simon Trott, Chief Executive

Assessment: Two fatalities in a half is a deteriorating record on the measure that matters most and the one carrying the largest tail risk in this industry. Both events also carried direct operational cost: the Kennecott stand-down cost production, and the Simandou stoppage cost construction days. The new operating system launched on 1 July, so it contributed nothing to this half and cannot be assessed until the next one. This is the only pillar of the thesis where the direction of travel this half was negative.

9. Resolution clears its legal overhang and the drills start turning

The congressionally mandated land exchange for Resolution completed in March. That removes the largest legal obstacle that has sat over this asset for years, and the project moved immediately into physical work: initial underground development including expansion of the mining station at about 6,800 feet, surface drilling in newly accessible areas, and underground drilling scheduled to start in the third quarter. Exploration and evaluation expenditure rose partly because of it, with 57% of the US$480M year-to-date charge going to the copper product group.

"So the next step for Resolution is drilling out the ore body. And so we've got rigs on site. We should be intersecting the ore body shortly, and that's the next phase of that project is to really characterize the ore body that will allow us then to make decisions around what the development path for that looks like."
— Simon Trott, Chief Executive

Assessment: In February we listed the pending Ninth Circuit decision as one of three legs of the jurisdiction and execution risk pillar. It has resolved favourably, and the project has converted from a legal question into an orebody-characterisation question. That is real progress and it costs money before it makes any: the spend runs through the income statement as exploration and evaluation, up 38% year on year, and there is no development decision, no capital number and no timeline. Resolution improves the option value of this portfolio in the 2030s. It contributes nothing to the numbers this decade.

10. Mongolia: the money is paid, the dispute goes to arbitration

Oyu Tolgoi paid the MNT 1.6 trillion assessment, about US$443M in primary tax, interest and penalties covering the 2021 and 2022 financial years, in full in March, reserving its right to dispute. That payment sits inside the half's working capital outflow. The company describes the assessments as inconsistent with the Oyu Tolgoi Investment Agreement and with Mongolian legislation. Separately, the Government of Mongolia agreed during the half to commence work on the Entrée licence transfer for Panel 1, which had been unresolved at the February report.

"And Kate, clearly, on the tax, there's a formal arbitration process there to solve it. So that is moving forward through that formal process."
— Peter Cunningham, Chief Financial Officer

Assessment: Two developments in opposite directions. The licence transfer starting is a genuine improvement, because mine sequencing at Oyu Tolgoi has been carrying an avoidable constraint. Escalating a tax dispute to formal arbitration is not a deterioration in itself, but it converts a negotiation into a process with a multi-year clock, in the jurisdiction hosting the single asset most responsible for this company's copper growth. US$443M is immaterial against a US$152bn market capitalisation. The signal is worth more than the sum: this is the second consecutive report in which the Mongolian relationship generates a line item.

11. Capital allocation: the guide holds and the return claim gets specific

Capital investment guidance is unchanged at up to US$11bn in 2026 and 2027, stepping down to US$10bn from 2028 in real 2025 terms, with sustaining capital stable at about US$4bn a year and sustaining, replacement and decarbonisation together at roughly US$7bn to US$8bn. Growth capital is dominated by copper, with about US$1bn a year on lithium and Simandou completing by the end of 2027. The half's US$5,037M split into US$1.3bn growth, US$1.9bn replacement, US$1.8bn sustaining and US$0.04bn decarbonisation.

"Our CapEx guidance is unchanged, up to $11 billion in 2026 and 2027 before a reduction from 2028 to $10 billion in real '25 terms. Sustaining capital is stable at around $4 billion a year."
— Peter Cunningham, Chief Financial Officer

The new claim is a portfolio return figure, offered without a supporting bridge.

"Returns are high. We assess the current portfolio as delivering an average a 26% IRR."
— Peter Cunningham, Chief Financial Officer

Assessment: Holding the capital guide for a second consecutive report, through a half in which currency and energy both moved against the cost base, is a better signal than the number itself. The 26% internal rate of return claim is the opposite: it is unfalsifiable as presented, with no disclosure of which projects sit in the average, what price deck stands behind it, or whether it is a Rio-share or 100% figure. Take the capital discipline as evidenced and the return claim as marketing until it is broken out.

Guidance & Outlook

Rio does not guide revenue or earnings. It guides volume, unit cost, capital spend and tax rate, and the operative point about this release is that none of it changed on 29 July. Every 2026 range had already been set at the December Capital Markets Day, reaffirmed on 19 February and reaffirmed again on 15 July, when the one change of the year so far was made: copper C1 net unit cost guidance was cut to 30 to 50 US cents per pound from 65 to 75, attributed to higher than expected gold prices and to productivity improvements.

Implied second-half shape. The pattern across the table is consistent and it goes one way. Copper is guided 11% below the first half at the midpoint. Bauxite, alumina and aluminium all sit at or below their first-half run rates against the guided ranges. Only iron ore and lithium require a second half above the first: the Pilbara needs roughly 173 Mt at the midpoint against 157.7 Mt delivered, which management supports with the statement that around half of the 8 Mt of first-quarter weather impact is expected to be recovered during the year, and lithium needs 33.7 to 36.7 kt against 27.3 kt, which the Fénix 1B, Sal de Vida and Rincon ramps should supply. Nothing here is a stretch. The point is simply that a reader extrapolating the first half into the second will overstate the full year on most lines.

Unit costs are the tighter constraint. Pilbara unit cash costs of US$25.0 per wet tonne are already at the top of the US$23.5 to US$25.0 range, at a currency assumption of 0.67 against a half in which the US dollar weakened 11% against the Australian dollar, and with diesel having moved from about US$85 to about US$140 a barrel. The company estimates that a US$10 a barrel move in diesel is worth about US$0.15 per tonne on full-year Pilbara costs. Holding the range therefore needs the second half to come in below the first, which requires either the volume recovery to arrive or the diesel spike to unwind. Copper C1 runs the other way for a reason that is already disclosed: negative 24.7 cents in the first half against a 30 to 50 cent full-year guide implies roughly 113 cents in the second at the midpoints, because Kennecott's by-product credits and payable volume both fall after the furnace breach.

Street at. Because guidance was reset on 15 July, the Street's 2026 volume and unit-cost assumptions were already current when the financials landed. The consensus movement out of this print therefore comes from three places rather than from the guidance table: a first-half base 0.3% to 1.0% above forecast at the EBITDA and earnings lines, a net debt position US$1.19bn better than modelled, and an effective tax rate roughly five points below the full-year rate guided in December. The last of those is the one most likely to move published numbers, because no revised full-year tax figure was given.

Guidance style. Rio has now beaten its own copper unit-cost guide twice in a row, delivering 67 cents against an 80 to 100 cent range for FY25 and then cutting the 2026 range by 30 cents at the midpoint mid-year. It landed Pilbara unit costs exactly in line for FY25. Volume ranges have been honest midpoints rather than sandbagged floors. On that record, the second-half implied numbers above should be read as achievable rather than conservative, and the cost ranges retain a modest bias to the favourable end.

Analyst Q&A Highlights

Bridging the gap between the banked figure and the run-rate target

The first question of the call went straight at the one number that changed: how the US$870M already banked becomes a US$1.8bn annualised exit rate, and what the composition of the additional US$1bn is across operating cost, capital and volume. Management drew the distinction between banked benefits and a run rate, acknowledged transition costs against the increment, and then handed to the CFO for a composition that came back at the level of the waterfall split rather than at the level of the increment itself.

Q: "Can we just talk about the gap between $870 million you've exited at the end of June and the $1.8 billion, the increase there. There's obviously 3 buckets here. There's OpEx, there's CapEx, there's some productivity-related cost out as well. So of that $1 billion increase, how do we actually think about the breakdown of that $1 billion increase?"
— Paul Young, Goldman Sachs

A: "Yes. I mean, Paul, on the slide on our waterfall, we set out that breakdown between costs and volume there. Cost was about the $530 million and then the volume was the rest. I mean this year, I expect a very similar breakdown for the full year as we bring that through. But it is very dependent. I mean this is bottom up and being driven by the business. So it will change, but that broad profile will continue."
— Peter Cunningham, Chief Financial Officer

Assessment: The answer describes the first half rather than the increment, and it is worth being precise about what it does and does not commit to. Roughly a 60/40 split between cost and volume, held for the full year, is a useful modelling anchor, and the confirmation that the program is bottom-up rather than allocated is the reason to believe it is durable. What was not answered is how much of the second-half increment is capital rather than operating, which matters because a capital-side benefit shows up in free cash flow but not in EBITDA. That distinction was in the question and not in the response.

Where aluminium sits between simplification and growth

A pointed structural challenge followed, noting that the aluminium business runs fourteen smelters across two continents with two more under study, that some individual assets are now more valuable than others, and that this is the one product group never mentioned in the same breath as portfolio simplification even as the company invests to grow it. The response defended the asset quality and reframed the question as one of improving the existing base rather than reshaping it.

Q: "But the aluminum business is fragmented. It doesn't seem to come up along conversations around the focus on streamlining this business. So I'm just wondering where it fits in as far as simplification and that strategy, considering that it appears you're looking to grow the business."
— Paul Young, Goldman Sachs

A: "So as I moved into role, we had a bit of a step back and really looked across our full business and the commodities we want to be in, and you've seen us simplify the business down to the 3 product groups and the 4 commodities. And we chose those commodities because we see those as the strongest in terms of growth going forward and reflecting our own position in those assets. And so we've got the best aluminum assets in our view, in the industry."
— Simon Trott, Chief Executive

Assessment: This is a clear answer even though it is not the answer that was asked for. Aluminium is one of the four chosen commodities, so it is not a source of divestment proceeds, which narrows the capital-release perimeter to businesses already named. The unaddressed half of the question is asset-level rather than commodity-level: a portfolio of fourteen smelters in a business the company intends to grow, with a greenfield study running in Finland, is a portfolio with individual assets worth more to someone else. That question has now been asked and deflected once. It will be asked again.

Whether the divestment program has a date attached

The most valuable disclosure in the entire Q&A came from a one-line follow-up on the capital-release program, and it went further than anything in the written release. Management confirmed the US$5bn to US$10bn target is tracking and, unprompted, attached a calendar-year target for announcements rather than for completions.

Q: "And my follow-up question is just on the potential $5 billion to $10 billion of asset divestments. Any comment you can make on that, please?"
— James Redfern, RBC

A: "No, progressing. I mean you've seen really strong cash generation today. The balance sheet is in good place and probably refer to my comments around capital discipline and efficiency. The divestment program, $5 billion to $10 billion tracking. We'll make decisions about that and make sure that we get full value. And so targeting $5 billion of announcements this year as part of that broader program."
— Simon Trott, Chief Executive

Assessment: Six months ago the answer to this question was patient, systematic and explicitly untimetabled, and we wrote at the time that a patient seller of a cyclical asset with no pressure to transact is a seller who may not transact. A target of US$5bn of announcements before year end is a commitment that can be scored, and the caveat about full value is doing less work than it was because the balance sheet no longer needs the money. The tell is what surrounds it: the answer opens by pointing at the cash generation and the balance sheet, which is a management team telling you it is negotiating from strength rather than from need.

What the productivity program looks like after 2026

The sharpest exchange of the call pressed on whether a roughly US$1.2bn year-on-year improvement can be repeated in 2027, or whether the easy decisions are taken first and the program decays. Management would not put a number on 2027, deflected once into the operating-system narrative, was called on the deflection, and then gave the honest answer.

Q: "Nice side step, Simon. But $1.2 billion -- $1.2 billion this year versus last year is the target. And then -- so can that momentum be sustained like real momentum to Peter's words, another $1.2 billion the following year? Or does it start to get harder?"
— Glyn Lawcock, Barrenjoey

A: "Look, the program will mature. And so inevitably, you start with some of the decisions in front of you. I think for us, there's 2 bits. There is maintaining the momentum on the increase, but also making sure that we sustain and maintain it going forward because if you embed it in the way people work and you embed it in the culture, then I no doubt that our people, and we've got fantastic people across the business. They'll find better ways of doing things."
— Simon Trott, Chief Executive

Assessment: "The program will mature" is the most useful three words in the Q&A and the least flattering to the extrapolation the share price would like to make. It is a direct concession that the 2027 increment is smaller than the 2026 increment, delivered without hedging, and it is consistent with the December framing of a multi-year program whose second phase runs through 2027 and 2028. Investors should model the US$1.8bn exit rate as a level that persists rather than as an annual increment that repeats. That is still worth a great deal against a US$25bn to US$30bn EBITDA base; it is not a compounding cost curve.

The cumulative size of the Mongolian tax dispute

A questioner put a cumulative number on the Mongolian tax position across all periods and asked what the remaining decision points are. Management did not engage with the arithmetic, redirecting to the relationship with the government and to the quality of the asset, before the CFO added the one substantive fact: the matter is now in a formal arbitration process.

Q: "In the result, we had the new disclosure around the tax dispute in Mongolia from the prior years. And I think we're now up to about $900 million if we put everything together. How do we think about this moving forward? Obviously, optically, that's not a great place to be. Are there any more decision points to work through or anything you can talk to there?"
— Kate McCutcheon, Bank of America

A: "I think the thing to take away from it, we continue to engage closely with the Mongolian government, and we'll continue to resolve things that need to be resolved as part of that project. And so really happy with the way that project continues to ramp up. It's going to be a fantastic asset for us for many, many decades."
— Simon Trott, Chief Executive

Assessment: The company's own disclosure in this release covers one assessment, the MNT 1.6 trillion demand of about US$443M for the 2021 and 2022 financial years, paid in March under protest. The larger cumulative figure put to management was neither confirmed nor disputed, and no decision points were named. What did come out of the exchange is that this has moved from discussion to formal arbitration, which is a change in character rather than in amount. For a company whose copper growth to 2030 rests disproportionately on this asset, the useful frame is not the dollar value but the fact that the relationship keeps producing line items, one per report for two reports running.

Copper growth beyond 2030 and Kennecott's predictability

A two-part question asked how the company de-risks the next Kennecott development given the asset's recent record, and whether the pipeline beyond 2030 needs inorganic help. The answer leaned on the optionality of Tier 1 assets, opened on safety rather than on engineering, and confirmed the near-term path without addressing the acquisition half of the question.

Q: "So I guess my question is in 2 parts. Firstly, you've had a bit of unpredictability at the asset and Apex is the next one that comes up beyond 2030. So in that development, how can you derisk that to make sure that you have a much more predictable production profile? ... are there brownfield opportunities that the market doesn't see within the portfolio for copper or beyond Resolution ... or do you need to solidify that by doing inorganic moves to kind of have a clearer path beyond 2030 in terms of your growth?"
— Rahul Anand, Morgan Stanley

A: "So the great things about Tier 1 assets is the optionality they provide and it's true in copper. And hence, we've got the 1 million tonnes by 2030, really building off the ramp-up at OT, 40% to 50% production growth at KUC. ... It is a real strategic card for us having a smelter in the U.S., 1 of only 2 in the U.S. and so thinking about ways that we best monetize that."
— Simon Trott, Chief Executive

Assessment: The inorganic half went unanswered, which is consistent with February's position that this company will not be the winning bidder for high-return, high-risk assets and should not be modelled with transformational acquisitions in it. The more interesting phrase is the one about monetising the US smelter, which is the first time an aluminium-style "strategic card" framing has been applied to a copper asset, and which sits oddly next to a US$5bn capital-release program that has so far named only mineral sands and borates. The path to one million tonnes of copper by 2030 rests on Oyu Tolgoi and Kennecott, both of which are already built.

Whether Pilbara unionisation threatens the productivity gains

One questioner stepped outside the numbers to raise the first material change in Pilbara labour arrangements in three decades and asked how it interacts with a productivity program built on frontline decision-making. The answer stayed at the level of principle and did not quantify anything.

Q: "But obviously, operating conditions in the Pilbara now are undergoing a little bit of change, first time in 30-odd years, we've got incremental unionization coming into more and more sites. How do you think about this, Simon, in terms of your -- the potential to impact operations? And how should we think about the ability for you to manage that and mitigate some of these forces on a sort of short, medium-term basis?"
— Lachlan Shaw

A: "In Australia and any jurisdiction we operate around the world, we obviously operate in accordance with the local terms and conditions. Our focus has been and continues to be how do we best work together with our employees to make sure that we have the conditions, safe, respectful workplaces, really listening to what people need in their day-to-day job to do those jobs better."
— Simon Trott, Chief Executive

Assessment: A non-answer, and probably the correct one to give in public on an active industrial-relations question. The substance is that the productivity program's stated mechanism is pushing accountability to the frontline so that people closest to the work change how it is done, and that mechanism is more sensitive to labour arrangements than a capital-driven cost program would be. Roughly US$55M of the banked benefits came from removing redundant Pilbara capacity and parking equipment, which is precisely the category of decision that becomes harder to take unilaterally. This is not a 2026 risk. It belongs on the list of things that could make the 2027 and 2028 phases harder than the 2026 phase.

What They're NOT Saying

1. No second-half framing anywhere in the presentation. Copper, bauxite, alumina and aluminium are all guided at or below their first-half run rates, and copper C1 requires a swing of roughly 138 cents per pound between the halves at the guidance midpoints. Management presented the first half as momentum without once addressing the shape of the second, and no analyst asked.
2. No updated full-year effective tax rate. The half came in at 25.2% against a full-year rate of roughly 30% guided in December and reaffirmed in February, on drivers that are explicitly non-operating: utilisation of previously unrecognised deferred tax assets and a change in the geographic spread of earnings. Neither the operations review nor the results carried a revised figure, which leaves roughly US$0.4bn of annualised earnings sitting on an unconfirmed assumption.
3. No 2027 number on the productivity program. The CEO conceded the program will mature and declined to size the second phase, exactly as in February. The difference is that there is now a 2026 figure to anchor on, so the omission costs less than it did.
4. No composition for the US$5bn of announcements. The target was given as an aggregate with no indication of whether it is asset sales, infrastructure monetisation, streaming, or partner sell-downs, and no named candidate beyond the Iron & Titanium and Borates reviews that have been running since 2025.
5. Nothing on why the Chilean lithium joint ventures slipped. The Codelco and ENAMI transactions were expected to close in the first half of 2026 at the February report. They are now expected in late 2026 or early 2027. The change appears in a project table with no explanation beyond regulatory approvals and customary closing conditions, and it was not mentioned on the call.
6. No cost or capital number for the Kennecott furnace remediation. The disclosure says refined copper and gold will be lower in the second half and that some sales and cash shift into 2027, but gives no repair cost, no duration, and no split between what shifts and what is lost.
7. No update on the size or resolution path of the wider Mongolian tax position. A cumulative figure roughly double the disclosed assessment was put to management in Q&A and neither confirmed nor rejected, and the release quantifies only the US$443M already paid.
8. No supporting detail for the 26% portfolio internal rate of return. No project list, no price deck, no ownership basis. It is presented as a single number in the prepared remarks and was not tested in Q&A.
9. Nothing on the Chinalco shareholding. The February call confirmed ongoing engagement with nothing to announce. This call did not raise it at all, in either direction.
10. No Yarwun detail beyond the curtailment date. Production drops 40% from October 2026 to extend the refinery's life to 2035, and the release again offers no annual care-and-maintenance cost and no long-term tailings solution.

Market Reaction

• Pre-print setup: RIO closed at $91.64 on 28 July, up 14.5% year to date against the S&P 500's +8.5%, up 47.2% over twelve months and down 2.8% over the trailing thirty days. The 52-week closing range into the print was $59.49 to $112.04, and the high was set on 13 May, so the stock entered the result 18.2% below its own recent peak.
• After-hours on the release evening: the results reached the wires on the evening of 28 July US time, after the New York close and before the Sydney presentation. The ADR added 0.85% in after-hours trade to $92.42, an initial reaction that was clearly muted relative to what followed.
• Reaction session: Rio releases before the US open. On 29 July the shares gapped up 2.0% to open at $93.46, traded a $91.98 to $95.10 range and closed at $93.66, up 2.2% or $2.02 on the day. The Australian line, which traded first on the news, rose 3.7%.
• Volume: 4.5M shares against a 2.8M thirty-day average, or 1.6x. This was a volume event, which the February print was not.
• Peers on the same session: BHP -0.8%, Vale -0.3%, Freeport-McMoRan -2.7%, Glencore ADR +3.2% and Alcoa -3.3%, against the S&P 500 at -1.5%.

A relative move, not just an absolute one. The headline +2.2% understates what happened. The S&P 500 fell 1.5% on the same session and four of the five diversified and base-metal comparators fell with it, two of them by more than 2.5%. Rio outperformed the index by roughly 370 basis points on a risk-off day, on 1.6 times normal volume, which is what a stock-specific re-rating looks like rather than a beta move. The February print produced the mirror image: a 2.6% decline while BHP and Vale rose.

The setup did most of the work. In February the problem was that the stock arrived at the print within 0.6% of its 52-week closing high after a 57.6% twelve-month run, so an in-line result with no new information was a sell. This time the stock arrived 18.2% below a high set eleven weeks earlier, having given back 2.8% over the trailing month while the copper price came off a record. In that position, a beat on earnings, a large beat on net debt, a 43% dividend increase and a tripled cost-out target is a buy, and the same information delivered into February's positioning would probably have produced a much smaller move.

What the market appears to have paid for. The pattern of the day argues that the balance sheet and the cost program did the work rather than the earnings beat, which was less than one percent on both EBITDA and underlying earnings and would not on its own move a US$150bn company two percent. Net debt came in US$1.19bn below where it was modelled, free cash flow rose 75%, and the productivity target went from US$650M to US$1.8bn. Those are the three items an equity holder could not have known from the 15 July operations review, and they are the three items that change what the next four quarters of cash look like.

Street Perspective

Debate: Does a price-led half deserve a re-rating?

Bull view: The bull case being made is that the composition question is less interesting than the cash question. Whatever drove EBITDA, the company funded a peak capital program, raised the dividend 43% and still reduced net debt, and it did so with the productivity program only halfway to its exit run rate. On this reading the multiple should expand because the balance sheet risk that justified a discount has receded, and because copper and aluminium now supply more than half of segment earnings at higher margins than iron ore.

Bear view: The bear camp contends that this is what a cyclical looks like at the top: copper reached a record in mid-May, aluminium is up a third, gold is up more than half, and a multiple applied to peak-price earnings is the classic value trap in this sector. Strip the price line and the waterfall nets to negative. On this view the correct response to a price-led beat is multiple compression, not expansion, and the second-half guidance for lower copper volumes at sharply higher unit costs is the first evidence of the reversal.

Our take: The bears are right about the earnings and wrong about the equity. Trailing twelve-month underlying EBITDA of US$28.6bn is the first half of 2026 plus the second half of 2025, so it already blends the current price environment with a period in which realised copper averaged well below the 591 cents achieved this half, and the stock trades at 5.8 times that number against 6.7 times FY25 EBITDA in February. The market is not paying a peak multiple for peak earnings. It is paying a lower multiple than it paid six months ago for earnings that are higher and cash flow that is much higher, which is the opposite of the value trap the bear case describes.

Debate: Is the productivity program a level or a curve?

Bull view: The optimists point to the trajectory. December's target was US$650M, June's banked figure is US$870M, the in-half run rate hit US$1.3bn and the year-end target is US$1.8bn, all inside seven months, with more than eighty initiatives running and a new management operating system launched on 1 July to institutionalise them. Against a cost base of this size and a stated four percent compound unit-cost reduction to 2030, the argument is that this is a multi-year curve worth several billion dollars of cumulative EBITDA.

Bear view: The skeptics note that the CEO himself conceded the program will mature, that the easy decisions come first, and that the second-phase figure for 2027 and 2028 remains unquantified for the second consecutive report. They also point out that this half's US$1.2bn of net controllable benefit was almost exactly consumed by currency, inflation and energy, so the shareholder has seen the effort and not the money.

Our take: Both are describing the same thing from different ends. Model the US$1.8bn as a level that persists rather than an increment that repeats, which is what "the program will mature" means in plain language. That is still worth roughly six percent of trailing EBITDA in perpetuity for no capital, and it is the highest-return dollar in this company. The bear point about the money being consumed by macro is fair for this half and is not a structural claim: currency and diesel do not move eleven percent and sixty-five percent every year in the same direction.

Debate: Does the capital-release deadline change anything?

Bull view: The constructive reading is that a target of US$5bn of announcements before year end, volunteered rather than extracted, is the first hard commitment the company has made on a program that has been an aspiration since December. With gearing at 16% and net debt falling, the proceeds now go to shareholder returns or growth rather than to repair, which is a much better use than the deleveraging case that was implied in February.

Bear view: The bear camp observes that half of 2026 has gone with nothing signed, that the same two businesses have been under strategic review since 2025 with guidance withdrawn, that mineral sands remain in a weak part of their cycle, and that the company has stated repeatedly it will not sell below its own view of value. A target on announcements rather than completions is a lower bar than it sounds, and a management team that walked away from a transformational merger on valuation is not a management team that will accept a soft price for a non-core division.

Our take: The bear reading of the mechanics is correct and the conclusion is too harsh. We continue to carry nothing for the capital release in the base case, so any signed transaction is upside rather than a support. What has changed is the asymmetry: in February a failure to transact left the balance sheet carrying a dividend it could not fund, and today it simply means the upside does not arrive. The deadline is worth watching because it is scoreable, not because the outcome is now essential.

Model & Valuation Framework

We do not carry a published financial model on Rio Tinto. The framework below updates the assumptions we set out at initiation in February, marked to what the first half delivered, and is anchored on the 29 July closing price of $93.66.

Where the equity trades

The comparison in that table is the arithmetic behind the upgrade. Over the five months since we initiated, the shares fell 3.5% while the S&P 500 rose 5.9%, and trailing underlying earnings per share rose 18.7%. The result is a multiple that has compressed from 14.4 times to 11.8 times on a business whose net debt is lower, whose gearing is lower, whose free cash flow yield is 80 basis points higher and whose cost program is now quantified. Nothing in the fundamentals deteriorated over that window; the price did the adjusting.

Valuation impact. Applying 6.0 to 6.75 times enterprise value to trailing underlying EBITDA of US$28.6bn, and deducting net debt held flat at US$14.1bn, gives an implied equity value of US$158bn to US$179bn, or roughly $97 to $110 per share against $93.66. The midpoint of about $104 implies approximately 11% upside before the dividend, and roughly 16% on a total-return basis assuming the trailing distribution rate holds. We deliberately apply a lower multiple range than the 6.5 to 7.0 times we used in February, because more of the current EBITDA is price-derived and copper reached a record in mid-May, and the range still clears the current price. Two variables widen the band from here: a signed disposal toward the upper end of the capital-release target, and a second-half productivity increment that carries the exit run rate above US$1.8bn.

What would change the rating. Downgrade triggers: copper delivery below the 800 kt low end of the guide, which would put both the volume and the unit-cost guide in question at once; a second-half free cash flow outcome that puts cash dividends back above 150% of free cash flow; the year-end productivity run rate landing materially short of US$1.8bn; or a Simandou sales outcome below the 5 Mt low end, which would reopen the schedule question we thought had closed. Upgrade beyond the current rating would need a signed disposal at a credible price plus a quantified 2027 and 2028 productivity target, which together would convert two option-value items into base-case cash.

Thesis Scorecard Post-Earnings

The pillars below are the ones established at initiation in February and carried in our standing thesis. They are scored against what this half's print and call revealed, not restated.

Overall: Thesis strengthened. Two of the four bear points that produced the February Hold have been challenged by the print rather than merely deferred, and the bull pillar we could not score for want of a number is now the best-evidenced item in the release. The offset is a new bear point about the quality of the earnings increase, which is real and which we have added rather than smoothed over.

Action: Upgrading to Outperform. Add here. We would trim into a copper-price-driven move back toward the May high without a corresponding step-up in the productivity run rate or a signed disposal, and we would revisit the rating if second-half copper delivery falls below the low end of guidance.

Bottom Line

In February we wrote that Rio Tinto was a good business, a credible plan and the right commodities, priced for a plan that did not pay until 2027 and 2028, and we initiated at Hold because the equity already assumed the delivery. Five months later the equity is 3.5% lower, the S&P 500 is 5.9% higher, and the delivery arrived early.

The half itself is more mixed than the headline suggests. Underlying EBITDA rose 28.4% and prices supplied more than the entire increase, so the operating story has to be read out of the controllable US$1.2bn rather than out of the US$3.3bn. The second half is guided lower on copper, bauxite, alumina and aluminium, Pilbara unit costs are already at the top of their range, Kennecott's furnace breach pushes metal and cash into 2027, and two colleagues died.

But the two objections that produced our Hold have both weakened. The productivity program went from an unquantified promise to US$870M banked, a US$1.3bn run rate and a US$1.8bn year-end target, with the benefits traceable into the same waterfall the company uses to explain its earnings. And the cash statement turned a year before the capital-spend schedule said it would: net debt fell while the company funded US$5.0bn of investment and paid a US$4.2bn dividend, and gearing dropped to 16%.

Meanwhile the price did what it was always going to have to do for this to become interesting. At $93.66 the shares trade at 11.8 times trailing underlying earnings against 14.4 times in February, at a 3.4% free cash flow yield against 2.6%, on 5.8 times enterprise value to trailing EBITDA against 6.7 times, 16.4% below a high set eleven weeks ago. We are upgrading to Outperform.