The Thesis Worked and the Stock Took the Payment: Downgrading BBVA to Hold

This report grades BBVA against the standing thesis established in our first-quarter 2026 initiation and carried in our internal thesis of record. Where a pillar's status changed this quarter, the change is stated explicitly in the Thesis Scorecard.

Results vs. Consensus

Q2 2026 Scorecard

BBVA is a euro-reporting IFRS bank and the consensus that matters is the euro-denominated sell-side poll on group net attributable profit and net interest income. Two independent polls were struck for this quarter and they agree closely: €2,960M and €7,550M on one, €2,950M and €7,510M on the other. We use the first and treat the second as the range check. The bank reports before the European open; the print landed on the morning of July 30 and the call followed the same morning.

Two rows need a note. The ADR earnings line is this same quarter translated into dollars, so it corroborates rather than adds to the euro beat; we show it because it is the figure most readers outside Europe will see quoted. And the cost-of-risk row is flagged green against the prior quarter but that is a cumulative half-year figure improving against a cumulative first-quarter figure, which is a different comparison from the year-ago 132bp. On a year-on-year basis the cost of risk is still 11bp higher.

Year-Over-Year Comparison

The right-hand column is management's constant-currency presentation, which is how the bank frames every operating comparison on the call. The reported column is what actually hit the P&L. The gap runs about five points at the quarterly net interest income line this quarter because the Mexican peso and the Latin American currencies moved sharply in BBVA's favour. The polls above are struck on reported euros, so the surprises in the scorecard are reported-basis.

Constant-currency rates are shown only for the four revenue lines and gross income, which are the comparisons management quoted on the call. The bank publishes constant-currency detail on the cumulative half-year statement rather than the standalone quarter, and we have not manufactured the missing cells.

Quarter-Over-Quarter Comparison

This is the table the headline hides. Net attributable profit rose 2.4% sequentially, but it did so with gross income falling.

Segment Performance

BBVA reports five geographic areas plus a Corporate Center, with a pro-forma cross-cutting view of Corporate & Investment Banking laid over the top. The 2025 comparatives were restated at the first quarter for an internal resegmentation that moved certain commercial customers into CIB; group consolidated figures were unaffected. Area figures below are cumulative half-year, which is the only basis on which the bank publishes a full area income statement.

The single most important number in that table is the four-point gap between the areas at +15.6% and the group at +11.1%. Every euro of that gap is the Corporate Center, and it is not a rounding item.

Quarterly Contribution and Area Ratios

The bank discloses standalone quarterly profit by area in the call narrative rather than in a table. Ratio columns in the filing's area blocks compare against December 31, 2025 rather than against the prior quarter, so the sequential moves below are taken from the filing's own narrative.

Spain

Spain earned €2,172M in the half, up 2.3%, on gross income up 3.1% and operating income down 0.1%. Read on its own that is a business standing still. Read against the rate cycle it is a business that has just turned. Net interest income grew 4.1% on loan balances up 7.4%, and the reason for the gap is the average customer spread over the two half-years rather than anything happening now. Within the quarter, the customer spread widened for the first time in the cycle.

"Customer spread improved also in the quarter by 3 basis points, reflecting effective price management in the context of higher rates while maintaining the cost of deposits contained." — Luisa Gómez Bravo, Group CFO

Lending grew 3.3% sequentially, flattered by a 19.5% jump in public-sector balances from the advance payment of an extra pension, but corporate lending grew 2.9%, mid-sized companies 2.2% and consumer 2.6% on their own merits. Asset quality reached a record: the NPL ratio fell to 2.86%, a historical low for the franchise, with coverage up 192bp to 71%. Part of that came from selling a mortgage-backed portfolio, which also holds down the reported cost of risk of 31bp.

The cost line is the blemish. Operating expenses grew 10.3% in the half against gross income of 3.1%, which is deeply negative jaws, though the comparison carries both the first-quarter voluntary redundancies and a value-added-tax re-estimation booked in both periods. Underlying cost growth was given as 5% year on year.

Assessment: Spain is the cleanest evidence for our thesis and the weakest-looking line in the table, and both statements are true at once. A 3bp spread widening is small in isolation; it matters because it is the first one, and because management has now committed to repeating it every quarter. If Spanish net interest income growth converges toward the 7.4% loan growth through 2027, this area alone re-rates the group. If it does not, the whole rate-floor argument was wrong.

Mexico

Mexico earned €2,979M in the half, up 15.8% reported and 8.2% at constant FX, on loan growth of 9.9% year on year and an efficiency ratio of 30.8% that remains the best in the group. Every guidance line for the country moved up: loan growth to about 10%, net interest income growth to high single digit, cost of risk to below 335bp from around 340bp. The country is 44.1% of business-area profit, essentially unchanged from the first quarter.

"Based on this performance, we are upgrading our full year guidance. We now expect loan growth of around 10%, net interest income growth at high single digit and cost of risk, as I mentioned before, to end below 335 basis points." — Luisa Gómez Bravo, Group CFO

Underneath the upgrade sit two things that did not improve. The customer spread is still compressing, offset by volume and by a larger ALCO portfolio now at €19.1bn with duration extended to 3.2 years from 2.6 a year ago at an 8.8% yield. And BBVA Research cut its 2026 Mexican GDP forecast to 1.2% from 1.8%, the largest downgrade in the scenario table, on weaker investment and a softer labour market. Credit metrics moved slightly the wrong way: the NPL ratio rose 19bp sequentially on retail and wholesale inflows, and coverage fell 11 percentage points to 118%.

Assessment: The guidance raise is real and activity-driven, which is the kind we credit. But this is now a franchise being upgraded into a macro downgrade, with the spread half of the rate-floor argument still unproven. Management's case is that Banxico has bottomed at 6.5% and that a public infrastructure programme generates corporate demand from 2027. Both are plausible; neither is in the numbers yet.

Türkiye

Türkiye earned €532M in the half, up 74.3% at constant FX, with the second quarter contributing €269M. The efficiency ratio improved to 40.6% from 44.4% at the end of 2025. That is the good news and it is genuine. The bad news is that this is the one area where a guidance line moved backwards.

"However, the normalization of the retail portfolio is taking longer than previously expected due to the current macro environment. As a result, we are updating our full year cost of risk guidance to around 220 basis points with an expected better second half of the year, but still above our previous guidance of 200 basis points." — Luisa Gómez Bravo, Group CFO

Net interest income declined sequentially on a tighter lira customer spread as funding costs stayed elevated, offset by strong payment-systems fees. The macro assumptions moved against the bank: BBVA Research cut 2026 Turkish GDP growth to 3.0% from 4.0%, raised the year-end inflation forecast to around 30% from 28% to 29%, and now sees the policy rate at 36% rather than 35%. The 2028 exit from hyperinflation accounting was described as "very much at risk."

Assessment: Türkiye delivered better profit than expected and worse credit than guided, and management chose to fix the guide rather than defend it. We credit the honesty. But the full-year guide of about €1bn against €532M delivered implies a second half no better than the first, in an area where cost of risk is running 16bp above the freshly-raised target and the currency is still depreciating. This is where the group's variance lives.

South America

South America earned €556M in the half, up 41.6% at constant FX, with Colombia the standout: €175M against €73M a year earlier, an increase of 123.5% at constant FX. Peru contributed €194M and Argentina €73M. The efficiency ratio improved to 41.5% from 44.4% at the end of 2025, and gross-income guidance was raised to high teens from high single digit. Cost of risk improved to 2.69% from 2.76%.

"Looking ahead, we expect cost of risk to continue improving and converge toward full year guidance of below 250 basis points, supported by solid underlying trends in Peru and Colombia as well as a gradual improvement in asset quality metrics in Argentina following the tightening of our risk appetite since late 2025." — Luisa Gómez Bravo, Group CFO

Argentina is the problem inside the good number. Its cost of risk is 7.70%, its NPL ratio rose 10bp to 6.4%, coverage fell 3 percentage points to 76%, and the hyperinflation adjustment cost €279M in the half against €211M a year earlier. Argentine profit fell 19.2% year on year in reported terms despite the country's banking system growing lending 45%.

Assessment: This was the single most exposed commitment we flagged at the first quarter, and it was half-met. Seven basis points of improvement against a 19bp gap to the guide, with the rest deferred to the second half again, is the same answer given one quarter later. Colombia and Peru are doing the work; Argentina is the drag. The area gets a pass this quarter and a hard grade next.

Rest of Business

Rest of Business earned €508M in the half, up 60.0% at constant FX, on loan growth of 52.5% year on year. It is now running above a €1bn annualized profit rate from a standing start, and it did so with a RoRWA of 2.1%. The area is roughly half of the group's CIB franchise by lending, and management disclosed a return on capital of about 24% for CIB excluding Argentina and Türkiye, against a stated cost-of-equity hurdle.

"So year-over-year growth in loans is now in the rest of business is 52%. But it is happening at a very profitable level as well. You might see it in the RoRWA. We are providing RoRWAs now, as you can see on this page and for CIB. The RoRWA for the rest of business is 2.1%." — Onur Genç, CEO

The credit line deserves attention that the disclosure does not give it. Non-performing loans in this area rose to €392M from €153M at the end of 2025, an increase of 157.1%. Management attributed it to "some specific clients migrating to Stage 3, which had already been largely provisioned in the first quarter," and the cost of risk of 14bp is genuinely below the 20bp steady-state guide. The base is small and the explanation is credible. It is also the fastest-growing loan book in the group entering its first stress.

Assessment: This is the most interesting business in the group and the least seasoned. Growth at 52% with a 24% return on capital is exactly what a bank with surplus liquidity should be doing with a steep curve, and the pre-emptive disclosure on AI-adjacent exposures was well handled. But a book compounding at this rate has not been through a cycle, and a 157% rise in non-performing loans off a small base is the kind of number that is noise until it is not.

Corporate Center

The Corporate Center lost €696M in the half against €390M a year earlier. Net trading income swung to a €232M loss from €85M on exchange-rate hedges, and personnel expenses rose to €501M from €408M, the latter carrying the holding company's share of the voluntary redundancies.

Assessment: The hedging loss is the mirror image of the currency gain flowing through the areas, so it is not a new economic cost. The personnel line is. A €306M widening in the Corporate Center loss consumed a third of what the operating businesses produced, and unlike the areas it carries no revenue to grow into. This line is where a "record earnings" quarter quietly gives back four points of growth.

Key KPIs

Key Topics & Management Commentary

Overall Management Tone: Confident and unusually specific, with the confidence resting on delivered numbers rather than on framing. Where the news was bad it was published rather than defended: the Türkiye cost-of-risk guide was raised, the 2028 hyperinflation exit was called at risk, and the portfolio sale that flattered the group cost of risk was disclosed before anyone asked. The one place management was less forthcoming was quantification of the future, where the answer on AI's effect on headcount was a flat "we don't know" and the four-year profit target was left unrevised despite an explicit invitation to raise it.

1. Rates Did Not Just Stop Falling, They Rose

Our initiation rested on a single mechanism: policy rates bottoming in Spain and Mexico, ending the spread compression that had been eating loan growth. The macro delivered more than that. BBVA Research's scenario in this report has the European Central Bank hiking in June to 2.25%, against a prior scenario that assumed cuts, and the Federal Reserve holding at 3.75% rather than cutting twice. The proximate cause is the energy shock from the conflict between the United States and Iran, which pushed eurozone inflation above 2.5% and Spanish June inflation to 3.2%.

That is a mixed blessing rather than a pure gift. The same scenario cuts eurozone 2026 GDP growth to 0.7% from 1.1% and Mexican growth to 1.2% from 1.8%. A bank with 4% net interest income sensitivity per 100bp in Spain gets paid on the rate leg and taxed on the volume leg.

Assessment: The rate leg is worth more than the volume leg for this franchise, because volume growth is running at 17.7% at constant FX and is not the binding constraint. The thesis mechanism is confirmed and then some. It is also now substantially priced.

2. Spain's Customer Spread Turned, and the Net Interest Income Gap Is a Timing Problem

Spanish net interest income grew 4.1% in the half while loan balances grew 7.4%, and a question on the call put the gap directly to management. The answer was arithmetic rather than excuse.

"So 4.1% is the growth in net interest income when the loan balances, they have grown 7.4%. So why is it not at the same level as the activity growth? It goes back to the average spreads. So last year first half, this year first half, when you look into the average spreads, obviously, it's much lower in this first half. And that thing will disappear over time." — Onur Genç, CEO

The forward commitment attached to it was specific: spreads improving "every single quarter from now on," with the year-on-year average still lower in the second half and the effect disappearing when 2027 begins. Spain's disclosed sensitivity is around 4% of net interest income per 100bp, and management added a detail that has not been prominent before, which is that the sensitivity is asymmetric and now more exposed to the one- and three-month part of the curve than to twelve-month Euribor, because commercial lending is where the growth is.

Assessment: This is the most gradeable commitment management made. A 3bp sequential improvement is the first data point; three more quarters of it converts Spain from a 2% grower into a high-single-digit grower without any change in activity. The asymmetric short-end exposure is a favourable disclosure in a world where the ECB has just hiked.

3. The Cost-of-Risk Ratio Improved and the Credit Flow Deteriorated

Group cost of risk fell to 143bp from 154bp, and every area's cumulative ratio improved sequentially. That is the headline. Underneath it, gross non-performing loan entries were €4,094M in the quarter against €3,359M in the first and €3,219M a year ago, the highest of the five quarters disclosed. Net of recoveries the NPL balance grew €2,094M before write-offs, also a five-quarter high. Coverage fell 157bp to 85%.

Management was straightforward about what carried the ratio.

"Starting with the cost of risk on the bottom left, it stood at 143 basis points for the first half of the year, improving from 154 basis points in the last quarter. This improvement, it was supported partially by a portfolio sale that we did in Spain. But overall, underlying provisioning requirements, they remained broadly stable, even better than expectations in most geographies, except for retail portfolios in Turkey and in Argentina." — Onur Genç, CEO

The portfolio sale was not sized. Neither was the geographic split of the €4,094M of entries, though the filing attributes the sequential increase to Mexico, Rest of Business and Türkiye. Cumulative impairments still grew 24.2% at constant FX against 16.9% gross-income growth, so the line is still compounding faster than the business, just at 1.4 times rather than the 1.9 times of the first quarter.

Assessment: A ratio computed on a denominator growing 6.4% a quarter will improve even when the numerator is stable, and a portfolio sale improves it again. Both are real; neither is underlying improvement. The flow numbers are the honest read and they went the other way. We are not calling this a credit event, because a 2.62% NPL ratio and 85% coverage are good absolute numbers. We are saying the ratio understates what happened.

4. Türkiye Is the Only Guide That Moved Backwards

Cost of risk guidance for the year went to around 220bp from around 200bp, on retail normalization "taking longer than previously expected." Against €532M delivered in the half and a full-year guide still at about €1bn with a slight downward bias, the second half is being guided flat to lower. Management would not put a number on it.

The lira spread is the swing factor and it is entirely a function of the central bank.

"Our spreads, our margins are completely dependent on the macro interest rates. If interest rates come down, you would see a better number. If interest rates do not come down, you would see more or less very meager numbers as you see today." — Onur Genç, CEO

Management said the spread bottomed in June and is now slightly improving. The path sketched has the effective policy rate falling from 40% to the official 37% around September or October, and to roughly 36% by year end. That is a small move and management said so.

Assessment: Türkiye contributed 7.9% of business-area profit and a disproportionate share of the group's forecast variance. The guide was raised on credit, cut in substance on profit, and made dependent on a central bank that has just been forced to keep rates high by an oil shock. Everything management said here was honest and none of it was reassuring.

5. Rest of Business Is Now a Billion-Euro Franchise, and Nobody Has Seen It Stressed

Loans in Rest of Business grew 52.5% year on year and the area annualizes above €1bn of profit. The dominant line of questioning on the call was whether that growth is being underwritten properly, specifically whether it is exposed to artificial-intelligence infrastructure. Management pre-empted with numbers rather than assurances: data centres at 0.7% of exposure at default, technology at 0.5%, direct exposures to financial sponsors below 0.8%, and direct lending to software and IT services of around €700M to €800M.

"In the bank, we have developed this metric or the framework now on AI transition indicator. So we are looking into every single client of BBVA and identifying the vulnerability that they might have with the transition that's happening with the disruption that is happening with AI. And we don't see a major risk profile for BBVA in these subchapters." — Onur Genç, CEO

The strategic framing is that CIB is a cross-border corporate bank rather than a trading business: 40% of client revenues are cross-border, non-client revenues are described as small, and the target is €10bn of CIB revenue by the end of the plan period against €4,251M of gross income in this half.

Assessment: This is a well-run disclosure of a genuinely fast-growing book, and the return metrics support the growth. The unanswered question is not AI exposure, which management quantified, but seasoning. Non-performing loans in the area went from €153M to €392M in six months. Management's explanation, that specific clients migrated to Stage 3 having already been provisioned, is consistent with a 14bp cost of risk. It is also the explanation a bank gives the first time.

6. Capital: Another €2bn, and the 12% Commitment Restated Without Qualification

CET1 rose 7bp to 12.90%, with 75bp of earnings generation offsetting 40bp of dividend accrual and AT1 coupons and 41bp of risk-weighted asset growth net of risk transfers. The €3,960M framework programme completes August 3. A new €2,000M programme was announced the same morning, with the first €1,000M tranche starting August 5 and running no later than October 9.

"We don't like to work with excess capital. Our target is our target, 11.5% to 12%. We take the upper end of that range as the key target, 12%. So we have excess capital. When we have excess capital above 12%, we will distribute it back to our shareholders." — Onur Genç, CEO

Pro-forma CET1 after the new programme is 12.41%, so the excess above the 12% ceiling is roughly 41bp, or about €1.7bn of risk-weighted-asset-equivalent capital, before any second-half generation. Significant risk transfers contributed 6bp in the quarter and 18bp in the half against a 30bp to 40bp annual guide, with management expecting the high end and having executed the first Mexican transaction in July.

Assessment: This is the cleanest pillar of the thesis and it strengthened. The bank funded 41bp of asset growth, accrued 40bp of dividend, and still added capital. The commitment to return everything above 12% is now three quarters old and has been honoured twice. The extension of the risk-transfer programme into Mexico and Türkiye, where the regulatory capital charge is furthest above the market's loss expectation, is the highest-return capital action available to this bank and it is being scaled.

7. Costs: Reported Jaws Went Negative, Underlying Jaws Widened

Half-year operating expenses grew 17.9% against gross income of 16.9% at constant FX, which is 100bp of negative jaws. At the first quarter the same comparison was 80bp positive. That looks like deterioration and it is not.

"Excluding these effects, you see it in the bubble, cost growth rate would have been 14.5%, again, maintaining our positive jaws, which is important to us." — Onur Genç, CEO

The two effects are the first-quarter voluntary redundancies, concentrated in Spain and the holding company, and a value-added-tax re-estimation booked in the second quarter of 2025 and again in this one. Excluding both, cost growth of 14.5% against revenue of 16.9% is 240bp of positive jaws, three times the first quarter's. Sequentially, expenses fell 2.4%. Spain's underlying cost growth was 5% and management reaffirmed a below-35% efficiency target for Spain with a 2028 goal of "low 30s or circa 30%."

Assessment: The bear point we carried on cost discipline is weaker after this quarter, not stronger. Reported jaws are the wrong lens when both periods carry a distortion and one carries two. The underlying widening from 80bp to 240bp is the number that matters, and it happened while the bank kept investing.

8. The Corporate Center Quietly Took a Third of the Growth

The five operating areas grew net attributable profit by €911M year on year. The group grew by €604M. The gap is the Corporate Center, whose loss widened to €696M from €390M. About half of the €306M widening is net trading income, which swung to a €232M loss from €85M on exchange-rate hedges. Another €93M is personnel expense, up to €501M from €408M.

Assessment: The hedging loss is not an economic cost. It is the price of the currency gain that shows up as +22.9% reported net interest income against +17.8% at constant FX, and a reader who counts the gain but not the hedge is double-counting. The personnel line is different and it was not addressed on the call. A holding company whose staff cost grew 22.8% year on year while the group's headcount grew 0.9% is a legitimate question that nobody asked.

9. The 2025-2028 Plan Was Not Revised, Despite an Invitation

An analyst did the arithmetic publicly: annualizing the second-quarter run rate gets close to the €48bn cumulative profit target with two and a half years still to run. Management declined to revise.

"The only thing I can tell you is for the first 2 years that we had in the plan versus what we have already realized in the 18 months, we are doing better than in the EUR 48 billion number. We are doing better than what we originally planned." — Onur Genç, CEO

The strategic plan horizon was also restated as 2025-2029 in this report while the financial targets remain the 2025-2028 set, and the next update is scheduled for the strategic talks on October 6.

Assessment: Refusing to raise a target you are beating is either discipline or sandbagging, and the distinction matters for how much credit the multiple should give the plan. The October event is the forcing function. If the €48bn is not raised there, with ten of sixteen quarters left before the target window closes, the honest reading is that the plan was set low.

10. The CFO Is Leaving, and the Handover Was Framed as Organizational

The call opened with the chief executive noting that this was the Group CFO's last results presentation, with Luisa Gómez Bravo moving to board seats at subsidiaries and Gonzalo Rodríguez taking the role. The first question of the session asked whether anything follows from it.

"On the first one, should we expect any changes in our strategic thinking or financial management principles? Obviously, no. It's a natural transition. So no changes you should expect." — Onur Genç, CEO

No start date was given for the successor, and no rationale was offered for the timing beyond the CFO's own remark that she "didn't anticipate stepping off the train at this particular station."

Assessment: The outgoing CFO built the disclosure discipline that makes this bank gradeable quarter to quarter: published assumptions, published sensitivities, guides that move when assumptions move. That is a real asset and it is person-dependent until proven otherwise. We take management at its word on continuity of strategy. Continuity of disclosure quality is the thing to watch, and the first test is the October strategic talks.

11. Artificial Intelligence: A Framework, 100,000 Users, and Still No Euro Figure

BBVA launched "The Frame," described as a common framework for governance, architecture, security and performance measurement of AI agents, and created an AI Transformation unit represented at the highest level of the organization. More than 100,000 employees are using AI tools. Asked directly when the bank would have a view on what this does to the ideal size of the workforce, the answer was that it does not yet have one.

"We will give some more update on this in October when we have the strategic talks, but the real impact, quantitative impact, it's too early to put on the table." — Onur Genç, CEO

The reply opened with "the answer is we don't know," which is the correct answer and a rare one.

Assessment: This is the second consecutive quarter of AI narrative without a euro attached. The candour is preferable to a fabricated number, and the October date is a commitment rather than a deflection. But headcount is up 0.9% year on year and operating expenses are growing 17.9%, so whatever the technology is doing, it is not yet visible in the cost line. Until October this remains a story, not a driver.

Guidance & Outlook

Four items were raised, one was cut, and the rest were reiterated. The raises are activity-driven and concentrated in Latin America; the cut is credit-driven and sits in Türkiye.

Implied second-half path: group ROTE of around 21% for the full year against 22.2% delivered in the half implies a second half below the first. Management confirmed the direction without quantifying it, attributing the step-down mainly to Türkiye, and noted that the denominator moves too as buybacks shrink equity. On the areas, Türkiye's roughly €1bn full-year guide against €532M delivered is the explicit flat-to-lower commitment; Mexico, South America and Rest of Business were all guided up; Spain was guided to sequentially improving spreads.

Street at: the two polls were at €2,960M and €2,950M for the quarter and the print came in at €3,062M. On the full year, an around-21% ROTE guide is the only group-level number the Street has to anchor to, and it now sits above the 2025 outturn of 19.3% and above what was guided three months ago.

Guidance style: unchanged and still the most legible in the European sector. Management publishes the assumptions, publishes the sensitivities, and moves the guide when an assumption moves, including when it moves the wrong way. The Türkiye cost-of-risk cut is the proof: a bank that manages its guide rather than its numbers does not volunteer a 20bp deterioration in the middle of a record quarter. The persistent omission is a group revenue or net interest income path, now four quarters old.

Analyst Q&A Highlights

Whether the CFO Transition Signals Anything

The first question of the session went to the management change announced at the top of the call, and asked specifically whether financial strategy, capital-return priorities or guidance philosophy should be expected to move with it. The answer was a flat denial of any read-through, framed as a routine succession.

Q: "My first question is on the management changes. Should we infer any change in financial strategy, capital return priorities or guidance philosophy from the CFO transition? Or is the handover to Gonzalo Rodriguez purely organizational?"
— Marta Sánchez Romero, JPMorgan

A: "On the first one, should we expect any changes in our strategic thinking or financial management principles? Obviously, no. It's a natural transition. So no changes you should expect."
— Onur Genç, CEO

Assessment: The answer is what it had to be and it settles nothing. A CFO transition at a bank whose disclosure quality is a genuine differentiator is a real, if unquantifiable, risk to the reporting franchise rather than to the operating one. The absence of a stated start date for the successor and of any rationale for the timing leaves the question open. The October strategic talks are the first observable test.

Whether the Commitment to Distribute Above 12% Survives the New Programme

With pro-forma CET1 landing at 12.41% after the newly announced buyback, a question pressed whether the standing pledge to return everything above 12% remains live or whether 12% is a 2028 destination and the year's distribution is now complete. The response reaffirmed the pledge in unusually plain language.

Q: "But I wonder if the commitment to distribute any excess above 12% is still valid and if we should expect more buybacks by the end of the year or the 12% is a target for '28 if we should be done with the EUR 6 billion for '26."
— Francisco Riquel, Alantra

A: "We don't like to work with excess capital. Our target is our target, 11.5% to 12%. We take the upper end of that range as the key target, 12%. So we have excess capital. When we have excess capital above 12%, we will distribute it back to our shareholders."
— Onur Genç, CEO

Assessment: The most valuable exchange on the call. The answer converts 12% from a target into a ceiling and makes the distribution mechanical rather than discretionary: capital generated above the ceiling comes back, and the only variable is timing. With 41bp of pro-forma excess still standing and 30bp to 40bp of annual organic generation guided, the arithmetic points to further announcements rather than a pause.

Whether the Four-Year Profit Target Has Become Too Easy

A question put the plan arithmetic on the record: annualizing the second-quarter run rate lands close to the €48bn cumulative target with two and a half years still to run, which invites either an upgrade or an explanation. Management gave neither, confirming only that execution is running ahead of the original plan.

Q: "Could you maybe just discuss how we should think about the upside risk to your EUR 48 billion target because that seems very, very easy for you to reach."
— Sofie Peterzens, Goldman Sachs

A: "The only thing I can tell you is for the first 2 years that we had in the plan versus what we have already realized in the 18 months, we are doing better than in the EUR 48 billion number. We are doing better than what we originally planned."
— Onur Genç, CEO

Assessment: A deferral, not a dodge, and the October date makes it checkable. The relevant question for the multiple is whether the plan gets raised there. A target confirmed as beatable but not revised is worth less than a target raised, and the market is currently paying for the latter.

How the Fastest-Growing Loan Book Is Exposed to Artificial Intelligence

The dominant line of questioning on growth quality concerned the Rest of Business and CIB loan book, which expanded by roughly €11bn in the quarter, and specifically whether any of it sits behind AI infrastructure at a moment when the market has grown wary of how that lending is underwritten. Management answered with a framework and with disclosed exposure percentages rather than with reassurance.

Q: "And second, the market is getting twitchy about how AI capabilities are being underwritten. So give us some color on your exposures. How much of the book is AI related?"
— Marta Sánchez Romero, JPMorgan

A: "In the bank, we have developed this metric or the framework now on AI transition indicator. So we are looking into every single client of BBVA and identifying the vulnerability that they might have with the transition that's happening with the disruption that is happening with AI. And we don't see a major risk profile for BBVA in these subchapters."
— Onur Genç, CEO

Assessment: A good answer to the question asked. Data centres at 0.7% of exposure at default, technology at 0.5% and financial sponsors below 0.8% are small enough that the AI-credit question is genuinely closed for this bank. What the exchange does not close is the broader seasoning question on a book growing 52% year on year, which is a different risk from the one the market was asking about.

Whether Spanish Net Interest Income Catches Up With Loan Growth

A question returned to the gap between 4.1% net interest income growth and 7.4% loan growth in Spain and asked whether the two converge as spreads bottom, both in the coming quarters and into 2027. The answer located the entire gap in the year-on-year average spread rather than in current pricing.

Q: "So we're seeing NII up 4% year-on-year in the first half. So loans are growing around 7%. You did mention you expect customer spreads to have touched bottom, if I understood correctly. So should we expect NII pace of growth in the coming quarters, the year-on-year growth on the quarter to catch up with the volumes growth?"
— Carlos Peixoto, CaixaBank

A: "The second half obviously would be much better. It might be even better in the average spread for the second half only. But year-over-year, still it's going to be lower. So when next year we start, that average spread notion will disappear if rates develop as we forecast at the moment."
— Onur Genç, CEO

Assessment: This is the thesis stated as a schedule. The convergence is a 2027 event, not a 2026 one, because the year-on-year average spread stays lower through the second half even as the spot spread improves. That is an honest answer and it is also a reason not to pay today for a re-acceleration that shows up in fifteen months.

Whether the Buyback Is the Right Use of Capital at This Return

One exchange framed the buyback as arithmetically dilutive to a group return on tangible equity of 22.2% and asked whether that implies an unusually high hurdle for external growth. The answer was a return threshold rather than a preference, and it was stated as applying uniformly across geographies and segments.

Q: "The second is on your share buyback. It is a welcome news. At the same time, it is dilutive to group RoTE, which is very high. Do you consider the hurdle for external growth very high at this stage?"
— Andrea Filtri, Mediobanca

A: "the hurdle rate for growth is cost of equity because we are in very different geographies, in very different segments. We want to make sure that we use the cost of equity as the benchmark."
— Onur Genç, CEO

Assessment: The question is sharper than the answer. Buying stock at roughly 2.3 times tangible book with a bank earning 22% on that book is value-accretive to per-share metrics and dilutive to the ratio, and management is right that the ratio is not the objective. But a cost-of-equity hurdle is a floor, not a ranking rule, and it does not explain why buyback beats redeployment at a 21.7% marginal return. The pledge to distribute above 12% answers it in practice: the bank has decided the ceiling matters more than the ranking.

Whether the Turkish Lira Spread and Net Interest Income Recover

A question sought the shape of Turkish net interest income after a first half management itself described as bumpy, with the lira customer spread compressed by elevated funding costs. The answer disclaimed any independent control over the outcome.

Q: "And the second one is on Turkey. Just trying to get a bit of a sense on how should we think about the Turkish lira spread and the evolution of NII in the coming quarters after a bit of a bumpy first half?"
— Ignacio Ulargui, BNP Paribas

A: "Our spreads, our margins are completely dependent on the macro interest rates. If interest rates come down, you would see a better number. If interest rates do not come down, you would see more or less very meager numbers as you see today."
— Onur Genç, CEO

Assessment: An unusually clean statement of where the risk sits. Turkish earnings are a rate call, not a management call, and the rate call is currently hostage to an oil shock that has pushed year-end inflation forecasts up rather than down. Sizing the position for Türkiye means sizing it for a central bank the bank does not influence.

Whether the Second Half Steps Down From a Record First

A question worked the guidance backwards: a trailing return on tangible equity of 22% in the half against an around-21% full-year guide implies roughly €5.5bn of second-half profit, below the €6.05bn just delivered. Management confirmed the direction and located the cause.

Q: "if I take the RoTE 12 months trailing, which was 22% in the first half, the 21% guidance implies around EUR 5.5 billion of profit in the second half, slightly down on the first half. If that is correct, could you just comment briefly on some of the main drivers here?"
— Britta Schmidt, Autonomous Research

A: "It doesn't imply that the second half would be much lower than the first. It might be slightly lower because of Turkey mainly. But overall, we expect still a very good second half."
— Onur Genç, CEO

Assessment: The confirmation matters more than the hedge around it. A bank that just printed a record quarter and raised four guidance lines is nonetheless guiding the second half flat to lower, and the stock closed at an all-time high on the same day. Those two facts have to be held together, and holding them together is most of the reason for the rating change in this note.

What They're NOT Saying

1. Still no group revenue or net interest income guidance. Return on tangible equity remains the only group-level commitment for 2026. Four quarters into a thesis whose entire mechanism is the net interest income line, the absence of a group net interest income path is the most conspicuous omission in the disclosure set, and it is the number the market has to build for itself.
2. The Spanish portfolio sale was not sized. It was credited unprompted with "partially" supporting the improvement from 154bp to 143bp of group cost of risk, and separately with helping Spain's non-performing loan ratio to a record low. No euro amount, no capital gain, no cost-of-risk contribution in basis points. The single most-cited improvement of the quarter is therefore unauditable from the outside.
3. The €4,094M of non-performing loan entries was not broken down. The filing attributes the sequential increase to Mexico, Rest of Business and Türkiye in that order, with no figures. Entries rose 21.9% sequentially and 27.2% year on year, so the split is material to judging where the book is actually turning.
4. Digital bank economics remain undisclosed. Asked directly about deposit plans, management gave €11.9bn of deposits across Italy and Germany, described a universal-bank strategy for both, and acknowledged that a portion of German deposits is promotional and will run off. No revenue, no loss, no capital consumed, no break-even date.
5. Second-half Türkiye profit was implied but never stated. A roughly €1bn full-year guide against €532M delivered puts the second half at or below the first, and management confirmed the direction on a group basis while declining to put a Turkish number on the table.
6. USMCA has not been resolved and was not discussed. At the first quarter, management framed a July decision point on the trade agreement and assigned near-zero probability to cancellation. July has passed. The filing now says Mexican activity recovers "as uncertainty surrounding the renewal" decreases, which means it has not, and no analyst raised it.
7. The €48bn plan was not raised despite an explicit invitation. Management confirmed performance is ahead of the original plan and declined to quantify by how much, deferring to October.
8. No euro figure for artificial intelligence, again. A governance framework, an organizational unit and more than 100,000 users, with the impact on workforce size answered as "we don't know" and deferred to October. Headcount is up 0.9% year on year.
9. The Corporate Center's personnel line went unexamined. Staff costs there grew 22.8% year on year while group headcount grew 0.9%, contributing €93M to a €306M widening of the holding company's loss. It was not raised on the call.
10. Romania still has no quantified gain or capital impact. The disposal to Raiffeisen is now reflected on the balance sheet as assets held for sale, with a fourth-quarter close expected. Carried over unanswered from the first quarter.

Market Reaction

• Pre-print setup: The Madrid line closed at €22.82 on July 29, up 13.8% year to date against the IBEX 35 up 12.2%, up 68.3% over trailing twelve months and up 4.3% over the trailing thirty days. The 52-week closing range entering the print was €14.36 to €23.25, so the stock came in 1.8% below its own closing high. The New York ADR closed at $25.66, up 10.1% year to date against the S&P 500 up 6.9% and up 65.1% over twelve months. This is the mirror image of the first-quarter setup, when the stock entered the print down 8.7% year to date against a rising market.
• Reaction session (July 30, before-the-open print): The Madrid line opened at €23.10, a 1.2% gap, traded €22.94 to €24.02, and closed at €23.96, up 5.0% or €1.14. Volume was 13.4M against a 9.8M thirty-day average, or 1.4x. The close was a new 52-week closing high.
• New York listing: The ADR opened at $27.28, a 6.3% gap, traded $27.17 to $27.60, and closed at $27.60, up 7.6%. Volume was 3.2M against a 1.3M thirty-day average, or 2.4x. It too closed at a new 52-week high.
• Peer reaction: The whole European bank complex was higher on the day. Banco Sabadell rose 3.1%, Banco Santander 2.5%, HSBC 2.5%, UniCredit 2.4%, CaixaBank 1.8%, BNP Paribas 1.8% and Intesa Sanpaolo 0.9%. The IBEX 35 rose 1.8% and the S&P 500 1.7%. BBVA outperformed every listed peer.
• Intraday path: The Madrid line was up about 2% in early European trade and built through the session, finishing near its high. The gap between the two listings, 5.0% against 7.6%, is euro strength against the dollar plus New York's later close on a day when the broad US market rose 1.7%. The Madrid close is the cleaner read on how the print was received.

Three points of relative move over the home index on 1.4 times normal volume is a strong endorsement, and the composition of it is more informative than the size. This was not a beat-driven move: at 3.4% against the poll it was the narrower of the two beats we have graded. What was new on the page was the €2bn buyback. Management said as much during the capital discussion, noting that some listeners had not been expecting the announcement, and three separate questioners opened by welcoming it.

The guidance action was the second driver and it was broad: four upgrades against one downgrade, with the upgrades in the two areas the market had been most sceptical about and the downgrade in the area it already discounts. A return on tangible equity guide moving to around 21% from above 20% is a change to the earnings power the stock capitalizes, not a quarter's result.

The setup deserves emphasis because it is the inverse of last quarter's. In April this was a stock that had de-rated 8.7% year to date into a rising market, where a good quarter produced a clean 3.9% move on below-average volume because the marginal holder had already de-risked. In July it entered the print 1.8% off its high, having risen 21.3% in Madrid over the quarter, and it still gained 5.0%. That is a stock where the marginal holder is now paying up, and it is why the same operating performance supports a different rating.

Street Perspective

Debate: Is Around 21% the Real Number, or the Start of a Rising Sequence?

Bull view: The bull case being made on the Street is that a guide raised twice in six months by a management team that publishes its assumptions is a floor. Twenty-two point two percent was delivered in the half, the four-year plan is described as running ahead of schedule, and the October strategic event is the obvious venue for a formal upgrade to the 2028 targets.

Bear view: The bear camp contends the guide implies exactly what it says: a second half below the first, on management's own confirmation, with Türkiye named as the cause. On that path the full year lands near 21% and the trailing 22.2% is the peak rather than the run rate, which means the stock is capitalizing a number the company has already told you it will not repeat in the near term.

Our take: The bears have the better of the near term and the bulls of the medium term. The second-half step-down is confirmed, small and concentrated in the one area with an explicit downgrade, so it is not a thesis problem. But it is a timing problem for a stock that has just re-rated 27% in a quarter. The bull case requires the October event to deliver a raised plan; if it delivers only reaffirmation, the multiple has nothing to work with until 2027, when the Spanish spread convergence starts showing up.

Debate: Is 143 Basis Points a Real Improvement or a Denominator Effect?

Bull view: The optimistic reading on the Street is that every area's cost of risk improved sequentially, the non-performing loan ratio fell to 2.62% with coverage at 85%, Mexico's cost-of-risk guide was cut, and provisioning growth decelerated from roughly twice revenue growth to about 1.4 times. Credit is normalizing on schedule.

Bear view: The sceptical reading is that the ratio improved on a credit-risk denominator that grew 6.4% in the quarter, was helped by an unsized portfolio sale, and sits above a gross non-performing loan inflow of €4,094M that is the highest in five quarters. Coverage fell 157bp. The one cost-of-risk guide that changed direction went the wrong way.

Our take: The bears are describing the more informative dataset. Ratios lag flows, and the flows turned this quarter while the ratios improved. That said, the flow deterioration is concentrated in exactly the places management identified, the absolute levels remain healthy, and a book compounding at 18.8% at constant FX mechanically generates entries. Our read is that this is the growth cost of a growth strategy rather than a credit event, and that the right response is to size the position for it rather than to sell it. The bear point in our standing thesis stays where it is.

Debate: Does 2.4 Times Tangible Book Still Have a Margin of Safety?

Bull view: The argument circulating on the bull side is that a bank compounding tangible book plus dividends at 21.8% before the effect of buying stock above book, earning 22.2% on tangible equity, and returning capital under a hard 12% ceiling deserves whatever multiple that combination implies. On any residual-income framework a sustainable 22% return supports something close to 2.4 times, so the stock is not expensive against its own return profile.

Bear view: The bear framing is arithmetic. The shares rose 27.4% in a quarter in which reported earnings grew 2.4% sequentially and gross income fell. The entire move is multiple, and it takes the stock from a 1.97 times tangible book that offered a discount to fair value to a 2.36 times that does not. Published fair-value estimates on the Street now bracket the price on both sides by wide margins, which is what happens when a valuation stops being obvious.

Our take: Both sides are computing the same thing from different return assumptions, and the resolution is that the assumption matters more than the framework. A residual-income model with an 11% cost of equity and 3% growth supports 2.12 times at a 20% sustainable return, 2.25 times at 21% and 2.38 times at 22%. Management guides to around 21% for this year and 22% as a four-year average. The stock trades at 2.36 times. That is not expensive; it is fully valued, with the answer entirely dependent on whether you underwrite the top of management's own range. We initiated at 1.97 times explicitly because the de-rating supplied the margin of safety. It does not any more.

Debate: Is the Rest of Business Growth a Franchise or a Cycle?

Bull view: The constructive view is that a 52.5% loan increase at a 2.1% return on risk-weighted assets for the area, and roughly 24% return on capital for CIB excluding Argentina and Türkiye, in a corporate bank following existing clients cross-border with 40% of client revenues generated outside the client's home market, is a franchise being harvested rather than a book being bought. The disclosed exposure percentages close the AI-credit question.

Bear view: The sceptical view is that a book growing at 50% in a period of abundant corporate credit demand is by construction unseasoned, that non-performing loans in the area rose 157% in six months, and that a 14bp cost of risk on a book this young tells you nothing about its through-cycle loss content.

Our take: The bulls are right about the economics as reported and the bears are right that reported is not the same as tested. What tips it constructive is the disclosure behaviour: management volunteered the return metrics, the exposure percentages and the concentration data before being pushed, which is not how a bank behaves when it is worried about a book. We would underwrite this as a real franchise with an unproven loss curve, which argues for crediting the earnings and discounting the multiple applied to them.

Model Framework & Valuation

The table below marks our first-quarter framework to the half-year outturn. Four assumptions move up, one moves down, and the rating change comes from the price rather than from any of them.

Valuation. At the July 30 Madrid close of €23.96 the shares trade at 2.36 times the €10.14 tangible book per share reported at June 30 and 2.24 times the €10.71 book value. Trailing four-quarter reported earnings per share of €1.88 puts the multiple at 12.7 times; annualizing the second quarter's €0.53 puts it at 11.3 times. Market capitalization on the 5,581M shares outstanding is approximately €134bn. Three months ago the same arithmetic read 1.97 times tangible book and 10.4 times trailing earnings.

The residual-income arithmetic is the honest way to frame the change. A bank earning a sustainable 21% on tangible equity, growing tangible book at roughly 3%, and carrying an 11% cost of equity supports a price to tangible book of about 2.25 times. BBVA trades at 2.36 times while currently earning 22.2% on a trailing half-year basis and guiding to around 21% for the year. The multiple is therefore fair to modestly full, against a first-quarter position of fair to modestly cheap, and the entire move happened in one quarter without a corresponding change in the earnings path.

Distribution. Cash distributions against 2025 results totalled €0.92 per share, a 3.8% yield at €23.96 against 4.9% at the price we initiated on. The two extraordinary buyback programmes together total €5.96bn, roughly 4.5% of market capitalization, though the larger of the two has been executing since December and completes August 3. The distribution is still substantial and still funded by capital the bank does not need. It is simply worth less per share at a 27% higher price.

What would take us back to Outperform. A pullback toward 2.1 times tangible book without a deterioration in the operating story. A formal upgrade to the 2025-2028 plan at the October strategic talks rather than a reaffirmation. Spanish net interest income growth converging toward loan growth faster than the 2027 schedule management described. South American cost of risk clearing the sub-250bp guide rather than approaching it. Or evidence that the second-quarter non-performing loan inflow was a single-quarter artifact rather than a turn.

What would take us to Underperform. Two consecutive quarters of accelerating non-performing loan entries with coverage continuing to fall. A Türkiye cost-of-risk guide raised a second time. Group cost growth staying near 15% while revenue growth decelerates below the mid-teens as the currency tailwind fades. Or a rate path that reverses again and puts Spanish and Mexican spread compression back on the table.

Thesis Scorecard

The pillars below are the ones established at our first-quarter initiation, graded against what this quarter's print and call actually showed. The status tags are the ones carried in our thesis of record; where a tag moved, the move is stated.

Overall: The thesis strengthened. All four bull pillars confirmed, one of them by more than we underwrote, and the cost-discipline bear point weakened materially. The credit bear point is unresolved rather than worse: the ratio says one thing and the flow says another, and we are holding it at emerging until the next quarter settles which is right. Nothing in the operating story argues for a lower rating.

Action: Hold. Trim into strength rather than exit. We initiated at Outperform three months ago at 1.97 times tangible book, on the explicit argument that a year-to-date de-rating against a rising market had supplied the entry. The stock has since risen 27.4% in Madrid to a record close and now trades at 2.36 times, which is at or slightly above what our own residual-income framework supports for a 21% sustainable return. The business did its part; the multiple has taken the payment in advance. A second-half that management has already guided flat to lower, against a first half that was a record, is not the setup in which to pay a full multiple. We remain constructive on the franchise and neutral on the shares, and we will revisit at the October strategic talks, which is where the 2025-2028 plan either gets raised or does not.