Record Profit, the Biggest Buyback of the Three, and an NPL Book Up 53%: Initiating Sumitomo Mitsui at Hold

Results vs. Consensus

Three conventions matter before any number below. First, Sumitomo Mitsui reports under Japanese GAAP on a cumulative basis: it files interim, nine-month and full-year results and never publishes a standalone fourth quarter. Every March-quarter figure in this note is derived by subtracting the nine-month filing from the full-year filing, and is labelled as such. Second, the print and the call are separate events. Results were released after the Tokyo close on May 13; the Investors Meeting at which the chief executive presented them and took questions was held five days later, on May 18. This note is written from both. Third, the company's investor deck states segment year-over-year changes on a managerial-accounting basis after adjusting for exchange-rate movements, while the filed segment note does not adjust. Where the two differ, the basis is stated.

The result is judged against the company's own forecast rather than a sell-side poll. That is not a softer test in this case: the forecast was raised once during the year, from ¥1,300bn in May to ¥1,500bn in November, and reaffirmed unchanged at the nine-month stage in January. The print cleared it by 5.5%.

Full year ended March 31, 2026, against the company's standing forecast

The shape of that table is the quarter in miniature. Every revenue and profit line beat, and the one line that missed is credit. Net business profit exceeded its target by more than three times the margin by which net income did, because ¥88.4bn of the operating outperformance was consumed by provisions the company chose to take.

Year-over-year comparison, full year (¥bn)

EPS grew 2.2 points faster than net income because the average share count fell 1.6%, from 3,906.5mn to 3,842.4mn, on ¥250bn of buybacks executed across the year in two tranches. That is the mechanism the company intends to keep running, and it is the cleanest part of the story.

The March quarter, derived (¥bn)

Subtracting the nine-month filing from the full-year filing isolates January through March. Residual differences of ¥1–2bn against the arithmetic are the issuer's practice of rounding amounts below one million yen down.

The 348% year-over-year jump is arithmetic off a broken base, not a signal. The March 2025 quarter carried a negative ¥246.9bn in net other operating income and the Vietnamese affiliate impairment that made equity in affiliates a negative ¥76.0bn, and it produced ¥42.0bn of net profit against ¥100.4bn of ordinary profit. Comparing the March 2026 quarter to itself sequentially is the more useful test: net profit of ¥188.2bn is down 59.2% from ¥461.3bn in the December quarter, and the March quarter contributed 11.9% of full-year profit on 25.8% of full-year gross profit. Fourth quarters at this bank absorb the clean-up, and this one absorbed a lot of it.

Revenue and net interest income

Consolidated gross profit of ¥4,844.7bn grew 17.4%, and the composition is the strongest part of the print. Net interest income rose ¥381.4bn, driven at the bank level by a domestic loan-to-deposit spread that widened 18 basis points to 1.14%, with the second half at 1.17% against 1.10% in the first. Interest earned on domestic loans rose 32bp to 1.34% while interest paid on deposits rose only 14bp to 0.20%, so 18bp of the 32bp increase in loan yield was retained as spread rather than passed to depositors. Average domestic loans grew ¥4.4tn to ¥66.7tn. Net fees and commissions added a further ¥261.4bn, up 16.8%, on wealth management, payments and domestic wholesale structuring. The one soft line is trading: net trading income fell 48% to ¥199.4bn, and in the March quarter alone it collapsed to ¥16.0bn from ¥161.5bn.

Margins and costs

The overhead ratio improved 3.5 points to 54.7%, which is the cleanest evidence of operating leverage in the result: gross profit grew 17.4% against a 10.4% increase in general and administrative expenses. But 10.4% cost growth is not a rounding error, and management attributes it to inflation and to variable marketing costs that rise with credit-card volume. Those are structural, not cyclical. The bank has told investors to expect close to ¥200bn of further upward cost pressure over the next three years, to be offset by ¥200bn of planned reductions. In other words, the three-year plan holds base expenses flat rather than cutting them, and it does so only after redefining base expenses to exclude IT investment.

EPS and below the line

EPS of ¥411.97 grew 36.6%, ahead of the 34.4% profit growth, on the 1.6% reduction in average shares. The effective tax rate was 29.6%, essentially flat. Below the operating line, extraordinary losses widened to ¥51.6bn from ¥19.5bn, of which ¥46.1bn is the loss on the sale of part of a United States banking subsidiary's business, a charge that landed almost entirely in the March quarter and that management sizes at ¥34bn after tax. The forecast for the coming year assumes those extraordinary losses do not repeat, which is most of the reason net profit is guided to grow ¥117bn on only ¥87bn of ordinary-profit growth.

Segment Performance

Sumitomo Mitsui runs four business units plus a head-office account. The filed segment note carries current-year and prior-year figures without exchange-rate adjustment; the investor deck restates the year-over-year change on a managerial-accounting basis after adjusting for currency. Both are shown, because on the Global unit in particular the two tell different stories.

Full year by business unit, as filed (¥bn)

*Share of the four business units' combined net business profit of ¥2,589.3bn. The head-office account carries a ¥258.4bn charge that reconciles the four units to the ¥2,330.9bn group total.

The March quarter by business unit, derived (¥bn)

Unit economics from the investor deck (¥bn, managerial accounting, currency-adjusted)

Wholesale: the engine, and the widest spread between revenue growth and asset growth

Domestic wholesale delivered the largest absolute contribution of any unit: gross profit up ¥230.2bn, net business profit up ¥213.5bn to ¥997.1bn, and a return on allocated CET1 of 21.4%. Excluding gains on sales of equity holdings, that return still rose 3.3 points to 16.3%, which is the number that matters because the equity gains are finite. Income on deposits rose ¥143.2bn as the policy rate fed through corporate balances, income on loans ¥24.1bn, and structured finance fees ¥48.8bn. The overhead ratio of 32.5% is the best in the group.

"In Wholesale, we captured strong funding demand and robust corporate activities among domestic clients, enhancing both earnings and our competitive position."
— Toru Nakashima, President and Group CEO

The three-year record supports that: the corporate loan balance grew 27% between March 2023 and March 2026 against 20% and 15% at the two peers named in the company's own comparison, and corporate deposits grew 12% to ¥73.2tn. Credit cost in the unit was a net reversal of ¥4.6bn.

Assessment: this is the highest-quality earnings stream in the group and the one most exposed to a Japanese rate cycle that has further to run. The caveat is inside the loan-spread data rather than the profit line: average spread on large-corporate loans fell 5bp to 0.54% even as the balance grew ¥3.6tn. Growth is coming at a price, and management's own explanation is bridge financing for merger activity, which is short-dated and repriced on refinancing.

Retail: the swing factor, on a base that was broken last year

Retail net income of ¥217.8bn is up ¥227.4bn, which means the unit lost money last year. The prior-year base carried the FE Credit impairment and a comprehensive allowance on consumer-finance interest repayments; their absence, not new earnings, explains most of the swing. Strip to net business profit and the growth is ¥139.4bn to ¥427.7bn on gross profit up ¥200.2bn, with income on deposits alone contributing ¥126.8bn of that. Wealth management added ¥49.6bn, payments ¥28.2bn and consumer finance ¥17.7bn. Income on loans went backwards by ¥10.1bn.

Olive, the integrated retail account, reached 7.5 million accounts and management raised the FY3/29 target to 15 million. The associated profit target was raised too.

"Olive related net business profit is expected to increased by JPY 30bn to JPY 110bn in FY3/29."
— Toru Nakashima, President and Group CEO

Assessment: the customer-acquisition record is genuinely strong and the cross-sell and active-user ratios of 70% and 90% are unusual for a bank app. But look at what the ¥110bn is made of: ¥80bn deposits, ¥10bn wealth management, ¥10bn-plus payments. Roughly three-quarters of the Olive profit target is deposit spread, which is a rate story wearing a platform story's clothes. If the policy rate stalls at 0.75%, the platform narrative and the profit target decouple. The overhead ratio of 72.9% also remains the highest in the group, 40 points above Wholesale.

Global: the unit management has stopped growing on purpose

Global is the only unit whose net income fell, down ¥37.6bn to ¥321.0bn, and it is a quarter of the four business units' combined net business profit. Gross profit grew ¥110.1bn but expenses grew ¥107.4bn, so the overhead ratio worsened 2.5 points to 68.6% and net business profit added only ¥16.3bn. Total credit cost in the unit rose ¥90.9bn to ¥257.9bn, which is two-thirds of the entire group credit charge. Loan balances fell: USD 289bn at March 2026 against USD 295bn a year earlier, down 3% excluding currency, with Asia down 5% and Europe, the Middle East and Africa down 6%. Spreads did improve, to 1.43% from 1.13% over three years, which is the intended trade.

"The Global Business Unit is in an “intentional pause,” prioritizing improvement of profitability over balance sheet expansion. As a result, revenue growth has been moderate, but we are steadily improving loan spreads and ROE."
— Toru Nakashima, President and Group CEO

Assessment: the pause is a defensible strategy and the spread improvement is real. The problem is that the credit deterioration is not obviously part of the plan. A unit that is shrinking its book by 3% while its credit charge rises 54% is not simply reallocating; it is also discovering things. Management's own framing of the next three years is "structural reform," and the ROTE gate it set is explicit: it will not re-accelerate asset growth until the unit clears 10% ROTE with visibility to 15%. That is a multi-year hold on a quarter of the group's earnings base.

Global Markets: an exceptional year that management has already told you not to extrapolate

Global Markets produced ¥508.7bn of net business profit, up ¥39.0bn, and ¥356.2bn of net income at a 21.7% return on allocated CET1. The deck figures exclude the bond-portfolio rebalancing, which is the honest presentation given the rebalancing was a deliberate balance-sheet action rather than a trading result. On the group bridge, an exceptional Global Markets year is ¥100bn of the ¥224bn of after-tax one-off profits, and it is explicitly listed as non-recurring.

Assessment: credit for the disclosure. Very few banks label their best-performing markets unit as a one-off in the same presentation that reports its record. The March quarter shows why the label is right: consolidated net trading income fell to ¥16.0bn from ¥161.5bn a year earlier, and at the bank level the bond result swung to a ¥120.9bn fourth-quarter loss from a ¥24.4bn nine-month gain. Whatever the unit earned in the first three quarters, the last one gave a large piece of it back.

Key Topics & Management Commentary

Overall Management Tone: Confident on the franchise, deliberately unconfident on the environment, and unusually willing to disaggregate its own record. The presentation opened by separating core growth from rates, currency and one-off items before anyone asked, and it labelled the best-performing markets unit as non-recurring in the same breath as reporting it. Where the posture softened was overseas: on the Global unit's profitability and on the ROE gap with peers, the answers were framework-level rather than quantified, and the timelines attached to them run past the end of the plan being presented.

1. The record, and how much of it management says is repeatable

The chief executive led with the result and immediately reframed it. The starting position was that the year began badly, that the feared damage did not arrive, and that the outcome should be read through a bridge rather than through the headline.

"FY3/26 began amid rising uncertainty over the global economic outlook, triggered by U.S. tariffs. However, the negative impact did not materialize to the extent expected. Supported by favorable business environment, including policy rate hikes, net income exceeded the revised target announced in November and reached a new record high."
— Toru Nakashima, President and Group CEO

The bridge behind that statement puts core-business growth at ¥212bn after tax, rates and currency at ¥94bn, and the non-repeat of the prior year's forward-looking provisions at ¥63bn. Against ¥224bn of one-off profits, management spent ¥188bn on measures for the future.

"Meanwhile, we leveraged one-off profits such as gains of stocks to implement future measures, including the disposal of low-return assets and the bond portfolio rebalancing. We also recorded forward-looking provisions considering the ongoing tensions in the Middle East."
— Toru Nakashima, President and Group CEO

Assessment: this is the right disclosure and it deserves the credit it will not get in the headlines. The number an investor should carry forward is roughly ¥212bn of underlying growth on a ¥1,178bn base, or about 18%, not 34%. That framing also explains why the guide looks conservative: management is building next year off a base it has already told you is inflated.

2. Domestic net interest income and the loan-to-deposit spread

The single largest driver of the year was domestic spread. Interest earned on domestic loans rose 32 basis points to 1.34% while interest paid on deposits rose 14bp to 0.20%, widening the loan-to-deposit spread 18bp to 1.14%. The second half ran at 1.17% against 1.10% in the first, so the exit rate is above the average and the repricing is not finished. Average domestic loans grew ¥4.4tn to ¥66.7tn, with large corporates up ¥3.6tn and mid-sized corporates and small businesses up ¥1.5tn.

Deposit growth kept pace: consolidated deposits rose ¥12.7tn to ¥201.3tn and the loan-to-deposit ratio sits at 58.4%, which is the structural reason a Japanese megabank levers so hard to the policy rate. Retail deposits reached ¥62.7tn, and the company's own three-year comparison has its retail deposit balance up 8% against 4% and 3% at the two peers it benchmarks against.

Assessment: the deposit franchise is the asset, and it is compounding faster than the peer group's. The watch item is on the other side of the balance sheet. Large-corporate loan spreads fell 5bp to 0.54% while that book grew ¥3.6tn, so the group is buying share at the margin and management attributes the compression to short-dated bridge financing for merger activity. That is fee-generative but it is not durable net interest income.

3. Rate sensitivity was raised, and the guidance does not use it

The bank increased its disclosed rate sensitivity during the presentation, and extended it out five years for the first time.

"We previously explained that a 25bps policy rate hike would increase net interest income by JPY 100bn. We have now updated this estimate to reflect recent changes in our balance sheet. Under the revised estimate, the positive impact in the first year is expected to be JPY 110bn, increasing by JPY 10bn from the previous assumption. Furthermore, the repricing of fixed-rate loans, which was not previously included, is expected to contribute gradually, raising the fifth year impact to around JPY 150bn."
— Toru Nakashima, President and Group CEO

The FY3/27 plan assumes a Japanese policy rate of 0.75%, which is where it already sits after the December 2025 move. The three-year plan assumes 1.25% by FY3/29, so half a point of tightening is embedded across three years and none at all in the guided year.

Assessment: this is the most important arithmetic in the release. At a 29.6% tax rate, ¥110bn of net interest income is roughly ¥77bn of net profit, or 4.6% of the ¥1,700bn guide. The gap between guidance and the ¥1,757.75bn consensus is ¥57.75bn. Three-quarters of one 25bp hike closes it. That is not a subtle beat-and-raise setup; it is the explicit design of the guide, and the Street has priced the design rather than the guide.

4. Fee income and the domestic wholesale engine

Net fees and commissions grew ¥261.4bn, or 16.8%, to ¥1,820.6bn, with the March quarter alone up 33.7%. At the bank level, domestic fees rose ¥46.7bn and overseas ¥27.4bn; structured finance was the standout at ¥88.9bn, up ¥48.8bn. The securities arm contributed ¥586.4bn of gross profit and ¥128.3bn of net income, and the collaboration with Jefferies contributed ¥22bn of profit against a ¥50bn target for FY3/31, with a Japanese equities joint venture due to start in January 2027.

League-table position moved in both directions over three years: equity underwriting improved from sixth to fourth and bonds from fifth to second, while mergers slipped from second to fourth.

Assessment: fee income is the part of the story least dependent on the policy rate and it is growing at a similar clip to net interest income, which is the healthiest possible split. The merger-advisory slip is the blemish, and it sits awkwardly next to the bridge-financing explanation for compressed loan spreads: the group is lending into transactions it is not always advising on.

5. Credit costs rose, and the underlying number rose more than the headline

Total credit cost of ¥388.4bn was ¥88.4bn worse than the ¥300bn plan and ¥43.9bn worse than last year. Management's presentation of it is careful and, on its own terms, accurate: excluding ¥65bn of forward-looking Middle East provisions and ¥31bn from the disposal of non-performing loans at the Indonesian consumer-finance affiliates, the charge was ¥292.4bn against a ¥300bn plan.

"In FY3/26, we increased forward-looking provisions to JPY 100bn, mainly to address risks related to the Middle East situation and inflation. In addition, our assumptions for foreign exchange rates and interest rates in FY3/27 are somewhat conservative. Therefore, even if the current tension persists for some time, I believe that we have capacity to absorb a certain level of downside risk."
— Toru Nakashima, President and Group CEO

What the same-basis comparison shows is less comfortable. The prior year's ¥344.5bn also contained items management excluded at the time: ¥90bn of forward-looking provisions for United States tariffs and ¥74bn for large borrowers in Brazil. On the same clean basis, the underlying charge went from roughly ¥180bn to ¥292bn, an increase of about 60%. At the bank level the picture is the reverse and much better: SMBC's own total credit cost fell to ¥86.0bn, or 7 basis points of claims, from ¥150.8bn and 12bp. The deterioration is entirely in the subsidiaries and overseas: the card and consumer-finance arm at ¥126bn, overseas banking subsidiaries at ¥111bn, and the Indian consumer-finance arm at ¥43bn.

Assessment: the domestic bank's credit book is in excellent shape and the group's is not. Guidance of ¥340bn for FY3/27 sits below this year's ¥388.4bn but 16% above the underlying ¥292.4bn, so management is not actually forecasting improvement in the run rate. It is forecasting fewer new surprises.

6. Asset quality: the Americas book tripled

This is the disclosure that did not make the headlines and should have. Consolidated non-performing loans rose from ¥881.7bn to ¥1,349.3bn, up 53%, and the NPL ratio went from 0.67% to 0.97%. The bank-level ratio rose from 0.43% to 0.71%. The regional split shows where it came from: the Americas balance went from ¥117.5bn to ¥367.3bn, more than tripling; Asia rose from ¥174.9bn to ¥246.9bn; the domestic book rose from ¥455.4bn to ¥584.4bn; Europe, the Middle East and Africa from ¥133.9bn to ¥150.7bn.

Assessment: a 30 basis point move in the NPL ratio at a 0.97% starting level is not a solvency question for a bank with a 12.41% regulatory CET1 ratio. It is a credibility question about the earnings trajectory. Management sized the Middle East exposure at ¥3.5tn and characterised it as 70% financials and sovereign with more than 80% investment grade, and it says the private-credit book carries no losses to date. None of that describes the Americas, which is where the deterioration actually happened and about which the presentation says nothing specific.

7. The Global Business Unit's declared pause

Management describes the group's global platform as generating around 40% of total profits, and it has explicitly stopped growing the overseas lending book. The stated gate for restarting is a return threshold, not a date.

"I am not satisfied with current profitability. We will accelerate our initiatives, at first by strengthening the acquisition of stable, sticky deposits. We have been investing in transaction banking infrastructure for the past two years and are making progress in talent acquisition. Second, we will shift our business model from loan-centric to fee-based businesses, including sales and trading and investment banking businesses. Third, we will expand asset management business, including private assets."
— Toru Nakashima, President and Group CEO

The plan reallocates risk-weighted assets away from overseas lending: total RWA moves from ¥113tn to ¥130tn while the overseas-loan share falls 12 points, with investment banking and sales and trading up 5 points and the multi-franchise strategy in Asia up 7. The unit exited the United States digital banking business and freight-car leasing during the year, and the ¥46.1bn extraordinary loss on the sale of part of a United States banking subsidiary's business is the accounting expression of that exit.

Assessment: the strategy is coherent and the spread evidence supports it. The risk is duration. A quarter of the group's business-unit earnings is being held flat while its credit costs rise, and the re-acceleration trigger management named requires clearing 10% return on tangible equity with visibility to 15%, which the deck implies is beyond the current three-year plan for this unit. Investors are being asked to fund a rebuild without a stated finish line.

8. The fourth quarter absorbed the clean-up

Three charges concentrated in January through March, and together they explain why a quarter with 104% net-business-profit growth produced only 11.9% of full-year net income. At the bank level the bond result swung to a ¥120.9bn fourth-quarter loss from a ¥24.4bn gain through December, as the deliberate portfolio rebalancing was executed. Extraordinary losses of ¥46.8bn in the quarter are essentially the whole ¥46.1bn United States subsidiary charge. Consolidated net trading income fell to ¥16.0bn from ¥161.5bn a year earlier.

Management's after-tax accounting of the year's ¥188bn of forward-looking measures lists forward-looking provisions at ¥46bn, the bond rebalancing at ¥42bn, the United States restructuring at ¥34bn, a comprehensive allowance for dormant deposits at ¥24bn, sales of low-return assets at ¥21bn and the Indonesian non-performing loan disposal at ¥21bn.

Assessment: taking the pain in the fourth quarter of a record year is textbook capital management and there is nothing to object to in the decision. The thing to hold onto is that the shape of the year is now unusually clean going forward: the bond book has been repositioned, the United States exit is expensed, and the dormant-deposit and interest-repayment issues are provided. The FY3/27 guide of ¥1,700bn assumes essentially none of it repeats, which is why net profit is guided to grow ¥117bn on ¥87bn of ordinary-profit growth.

9. Equity holdings: ahead of schedule, and also the earnings engine

Gains on stocks of ¥446.1bn are 19.4% of ordinary profit. They fell ¥63.8bn year over year despite ¥94bn from the sale of an Indian banking stake, because underlying disposal gains fell ¥99bn to ¥386bn and a Hong Kong bank stake was sold at a ¥28bn loss. Reduction progress against the ¥600bn five-year plan reached ¥309bn, or 52% at the two-year mark against a 40% standard pace, with ¥185bn in FY3/25 and ¥124bn in FY3/26 plus ¥69bn of consented sales. Market value of equity holdings as a share of consolidated net assets was 27.5% at March 2026, down from 32.9% at March 2024 but up from 27.3% a year earlier, against a sub-20% target for March 2029.

"We plan to reduce equity holding by JPY 600 billion over five years, of which JPY 300 billion has already been achieved over the past two years. For FY3/27, we aim to reduce at least JPY 120 billion, in line with the standard pace. While the ratio of market value to net assets has been slower to decline due to rising stock prices, we will increasingly focus on reducing the market value itself going forward."
— Toru Nakashima, President and Group CEO

Assessment: the pace is genuinely ahead of plan and the disclosure is unusually specific. The tension is structural rather than managerial: unrealized gains of ¥2,497.2bn on domestic listed equities are simultaneously the source of a fifth of ordinary profit, a drag on return on tangible equity, and a capital cushion the rating agencies count. Management is now explicit that they are all three, and the answer it has chosen is to keep selling at a fixed pace rather than accelerate. That keeps the earnings contribution alive for years and keeps the return gap open for just as long.

10. Capital return stepped up, and the internal argument was disclosed

The dividend forecast is ¥180 per share for FY3/27 against ¥157 paid for FY3/26, an increase of ¥23 or 14.6%, at a 40% payout on the ¥1,700bn target. The progressive policy was upgraded from a maintenance commitment to an explicit annual-increase commitment. The buyback authorisation is up to ¥180bn, covering up to 40 million shares or 1.0% of shares outstanding, running from May 14 to July 31, with all repurchased shares scheduled for cancellation on August 20.

"If earnings exceed expectations, we will determine whether to allocate the upside to share buybacks or divided increases depending on the circumstances. For the JPY 80 billion upside in FY3/26, we prioritized share buybacks. While the current environment entails downside risks and there were internal views calling for caution on share buybacks exceeding last year's level, we decided on JPY 180 billion in order to clearly demonstrate our strong commitment to shareholder returns."
— Toru Nakashima, President and Group CEO

Two comparisons matter and they point in opposite directions. Against the opening tranche a year ago, ¥180bn is up 80% from ¥100bn, and it is 80% larger than the opening buyback at either of the other two megabanks. Against what was actually executed, ¥250bn in each of the last two years, ¥180bn is a step down, and the shareholder-returns slide labels the FY3/27 line "180 +α" precisely because more is expected to follow.

Assessment: the disclosure of an internal argument about buyback size is unusual and useful, and it tells you the ¥180bn is a floor set under an uncertain environment rather than a ceiling. The total payout ratio of 54% in FY3/26 is the reference point; a "51% +α" guide for FY3/27 implies the top-up is expected. Investors who want the second tranche will be waiting for the interim results in November, which is when both of the last two top-ups arrived.

11. The CET1 target went up, which absorbs capital before it can be returned

The operating CET1 target for FY3/29 was raised 50 basis points to around 10.5%, with a stated range of 10.0% to 11.0%. The rationale is explicit and worth reading carefully.

"Rating agencies consider unrealized gains on securities when assessing capital adequacy. Given that these gains are expected to decline in line with the reduction of equity holdings, we set our CET1 ratio target at 10.5%. Supported by strong earnings, our CET1 ratio reached 10.3% at the end of March 2026, limited gap to our target. As the additional capital required to reach 10.5% is not substantial, we will gradually increase the ratio toward this level."
— Toru Nakashima, President and Group CEO

Note the two CET1 figures in circulation are not the same measure. The 12.41% in the regulatory disclosure is the preliminary Basel ratio; the 10.3% management refers to is the finalised-Basel-III figure excluding net unrealised gains on other securities, and it is the one the 10.5% target applies to. On the regulatory basis the ratio fell 3 basis points over the year as risk-weighted assets grew 8.6% to ¥101.1tn against 8.3% growth in common equity Tier 1 capital.

Assessment: raising a capital target while raising a buyback is a defensible pairing only if earnings are growing faster than risk-weighted assets, and this year they did not by the measure that counts. The three-year plan also commits ¥1tn of technology investment with an approximate 20 basis point CET1 impact. The capital story is therefore tighter than the headline buyback suggests, and it is one of the reasons the "+α" is a hope rather than a plan.

12. The new three-year plan sets a hard number for the first time

The plan that began in April targets 13% return on tangible equity and roughly ¥2tn of net income by FY3/29, against 11.4% and ¥1,583bn today, with a longer-horizon ambition of 15% ROTE requiring net income in the mid-¥2tn range. Management put its own return next to a named peer group of global banks on the same slide, which is not a comfortable comparison and was clearly not meant to be.

"Leading global players typically target ROTE of 15–20%, based on returns excluding goodwill and other intangible assets. As we establish ourselves as a leading global player, we are committed to achieving a ROTE of 15% over the medium- to long-term. This will require raising net income to the mid-JPY 2tn level."
— Toru Nakashima, President and Group CEO

The mechanics are a 0.5 point improvement in return on risk-weighted assets, an overhead ratio held in the low 50s, and a portfolio shift out of overseas lending into investment banking, the Asian multi-franchise strategy, and asset-light transaction banking and asset management. Seven strategic areas are named, of which three are domestic.

Assessment: adopting return on tangible equity as the headline metric is the right call for comparability and a slightly awkward one for accountability, since goodwill from the very acquisitions under scrutiny sits outside the denominator. Management pre-empted that point by committing to keep disclosing return on equity alongside it. The bigger question the plan does not answer is what happens if the policy rate stalls: the FY3/29 target embeds a 1.25% policy rate, and roughly ¥110bn of pre-tax income per 25 basis points means half a point of the gap between 11.4% and 13% is rates rather than execution.

Guidance & Outlook

Macro assumptions: a Japanese policy rate of 0.75%, a United States policy rate of 3.5%, and USD/JPY at 150. The Japanese assumption is the level already in force since December 2025, so the guided year contains no domestic tightening at all. The United States assumption embeds easing, which is a headwind to overseas net interest income. The currency assumption of 150 sits below the 159.90 rate at the March 2026 balance-sheet date and below the 151.06 FY3/26 average, so a weaker yen than assumed would be an additional tailwind on translation.

Implied bridge: net business profit is guided up only ¥69.1bn while net profit is guided up ¥117.0bn. Working down: net business profit less credit cost improves ¥117.5bn, gains on stocks and other items are implied to fall about ¥31bn to roughly ¥330bn, and ordinary profit therefore grows ¥86.6bn. Converting ¥2,390bn of ordinary profit at this year's 29.6% tax rate with no extraordinary losses gives roughly ¥1,683bn, close enough to ¥1,700bn that the guide is best read as ordinary profit up 3.8% plus the non-repeat of ¥51.6bn of extraordinary charges.

Street at: one published consensus for FY3/27 net profit stood at ¥1,757.75bn, and an independent poll of twelve analysts had ¥1,759bn. The ¥1,700bn guide is therefore 3.3% light. That places Sumitomo Mitsui in the middle of the three megabanks on guided growth, at 7.4% against 12.5% at the largest and about 4% at the smallest, all three of which reported within three days of each other.

Guidance style: the pattern is now legible. FY3/26 opened at ¥1,300bn in May, was raised to ¥1,500bn in November, was reaffirmed in January and landed at ¥1,583bn. That is a 21.8% total upward revision from first guide to actual across a single year. Management explicitly builds a business slowdown into the plan, treats the Middle East as a live downside, and describes its rate and currency assumptions as conservative. On the evidence of the last two years, ¥1,700bn is a floor set to be raised at the interim stage rather than a forecast.

Analyst Q&A Highlights

Sumitomo Mitsui publishes its own record of the analyst session and does not identify the questioners. The attributions below reflect that convention. All answers are the chief executive's.

How much of the three-year record was the company and how much was the environment

The opening question asked management to position both the result and the coming year's target against the 15% return-on-tangible-equity ambition. The answer produced the most useful single disclosure of the session: an explicit split of three years of profit growth into self-generated and externally supplied.

Q: "How do you evaluate the latest financial results? How are the FY3/26 results and FY3/27 targets positioned within the medium- to long-term ROTE target of 15%?"
— Analyst question, FY3/2026 Investors Meeting, May 18, 2026

A: "I believe that the previous Medium-Term Management Plan progressed steadily and delivered solid results overall. Over three-year period, net income increased by JPY 800 billion. Of this, we estimate that JPY 300 billion came from our own growth through business expansion and strengthened earnings capacity, while the remaining JPY 500 billion was supported by external factors such as rising interest rates and foreign exchange movements."
— Toru Nakashima, President and Group CEO

Assessment: management put a number on the question every investor in Japanese banks is asking, and the number is not flattering. Five-eighths of three years of profit growth came from rates and currency. That is not a criticism of execution, it is a description of the sector, but it should discipline the multiple an investor is willing to pay for the trend. The same answer described FY3/27's targeted ¥120bn increase with the line that replicating last year's growth "will not be easy," which is as close to a formal expectations reset as this format allows.

Why the return gap with peers is not closing

A recurring line of questioning pressed on the persistent shortfall in return on equity relative to the domestic peer group. Management's answer identified the balance-sheet architecture rather than the earnings engine, and conceded the rest.

Q: "What is preventing the ROE gap with peers from narrowing?"
— Analyst question, FY3/2026 Investors Meeting, May 18, 2026

A: "I expect that one factor is the impact of foreign exchange. Our capital structure is designed to reduce the sensitivity of our CET1 ratio to foreign exchange movements. This allows us to maintain CET1 stability even in the current weak yen environment, but it also increases capital in yen terms, creating a headwind to ROE. That said, I acknowledge that there is still room to improve capital efficiency."
— Toru Nakashima, President and Group CEO

Assessment: the currency explanation is real and it is a deliberate trade, capital-ratio stability bought at the cost of a larger yen-denominated denominator. It is also only part of the answer, and management said so. The rest is the ¥2.5tn of unrealized equity gains sitting in book value and earning nothing, which the company elsewhere concedes is weighing on returns. Naming currency first and capital efficiency second is a choice about emphasis, not about facts.

What is actually driving the retail platform's engagement metrics

The domestic digital account has become the centrepiece of the retail case, and the question went at whether its unusually high usage rates are structural or a function of promotional acquisition.

Q: "What drives Olive's high active user ratio, and what impact does Olive have on the overall Retail Business?"
— Analyst question, FY3/2026 Investors Meeting, May 18, 2026

A: "Olive has a structurally high utilization rate, in line with our original expectations. This is because customers with strong service engagement voluntarily open accounts online, rather than through traditional pushed-base, product-driven sales."
— Toru Nakashima, President and Group CEO

Assessment: self-selection is a genuine and durable explanation for a 90% active-user ratio, and it is a better answer than an incentive-spend story would have been. The unaddressed part is what happens to that self-selection as the account base doubles from 7.5 million to 15 million, since the marginal customer in the second tranche is by definition less engaged than the first. The profit target does not appear to assume otherwise.

How much capital the Asian franchise strategy is entitled to

With the overseas lending book shrinking and risk-weighted assets earmarked for reallocation to Asia, the question was what discipline governs the reallocation. The answer set an explicit return gate and an explicit sunset.

Q: "What is your approach to allocating risk assets to the Multi-Franchise Strategy?"
— Analyst question, FY3/2026 Investors Meeting, May 18, 2026

A: "We do not assume large-scale inorganic investments in the new Plan. We will allocate assets to each investee in line with local economic growth, implying a certain level of deployment. However, we will carefully assess whether investments can achieve around 13% ROTE during the plan period and 15% over the longer term. If returns are not expected to meet these thresholds, we will review our approach in a disciplined and flexible manner."
— Toru Nakashima, President and Group CEO

Assessment: a stated hurdle rate on inorganic capital is exactly what this strategy has lacked. The credibility test is that the same programme has already produced impairments in Vietnam and credit costs in Indonesia, and both are inside the base against which the hurdle is now being applied. "Review our approach" is not the same as "exit," and no divestment trigger was named.

Whether the Indian build-out is a platform or a collection of stakes

India is the one market management describes as top priority, and it now holds a branch network, a consumer-finance arm and a stake in a listed private-sector bank. The question was how those pieces become one business and how the risk is contained.

Q: "What is your growth strategy and risk management framework in India?"
— Analyst question, FY3/2026 Investors Meeting, May 18, 2026

A: "In India, SMBC operates five branches, primarily focusing on corporate banking and project finance. SMICC provides consumer finance services, including SME and retail lending. In addition, we invested in YES BANK in FY3/26, completing the key components required for growth. Under the new Plan, our focus is on enhancing coordination among these entities. If successfully executed, we can establish a leading position among foreign banks in India. However, we recognize that India is a complex market, particularly in terms of regulation and taxation."
— Toru Nakashima, President and Group CEO

Assessment: the honesty about regulatory complexity is welcome and the dedicated India headquarters is a concrete response. What the answer does not address is that the Indian consumer-finance arm contributed ¥43bn of credit cost this year, up ¥12bn, which is the second-largest increase among the named group companies. Coordination between three entities is a synergy argument; the credit line is a cost that is already running.

The three exposures the market was most worried about

One question bundled private credit, the Middle East and artificial-intelligence lending, and the answer quantified all three. It is the most data-dense response in the session and the closest management came to addressing the credit trend directly.

Q: "How do you assess the risk for exposure to private credit, the Middle East, and AI?"
— Analyst question, FY3/2026 Investors Meeting, May 18, 2026

A: "For private credit, we have JPY 1.2 trillion of exposure to BDCs, primarily to high-quality borrowers. With a high proportion of investment-grade exposure, low LTV, and a well-diversified portfolio, we do not expect any significant impact on earnings. No losses have been realized to date. Our Middle East exposure is concentrated in Qatar, Saudi Arabia, and the UAE, mainly to financial institutions and sovereign. Credit quality remains high and we recorded forward-looking provisions of JPY 65 billion in FY3/26. While we consider the current position manageable, downside risks remain if geopolitical tensions escalate further."
— Toru Nakashima, President and Group CEO

Assessment: three named exposures, three sizes, and a clear statement of what has and has not been provided against. That is better disclosure than most banks give on private credit. It is also conspicuously not an answer about the Americas, where the non-performing balance tripled this year, and the question as published did not ask about it. A ¥250bn move in a regional non-performing book went undiscussed in the only forum where it could have been discussed.

Whether cost discipline survives an inflationary three-year plan

The final substantive exchange on the operating model tested the credibility of holding expenses flat while inflation, card volume and technology spending all push the other way.

Q: "What is your outlook for cost increases and potential cost reduction?"
— Analyst question, FY3/2026 Investors Meeting, May 18, 2026

A: "We expect inflationary pressures both in Japan and overseas. In addition, we anticipate increases in performance-linked costs driven by the strong credit card business, as well as IT-related costs associated with continued investments. Each of these factors is expected to push costs higher by close to JPY 200 billion. That said, we will maintain strict cost discipline. We executed cost reductions of JPY 160 billion under the previous Plan and JPY 130 billion under the Plan before that. Under the new Plan, we aim to achieve a further JPY 200 billion in cost reductions."
— Toru Nakashima, President and Group CEO

Assessment: the track record supports the claim, ¥130bn then ¥160bn delivered, and ¥200bn against ¥200bn is a plausible next step. The qualifier at the end of the answer is the part that matters: the definition of base expenses has been changed to exclude technology investment, which is the fastest-growing line and the one carrying a ¥1tn three-year commitment. Flat base expenses under the new definition is a different promise from flat expenses.

What They're NOT Saying

1. The Americas non-performing balance. It tripled from ¥117.5bn to ¥367.3bn and drove the bulk of the 30 basis point rise in the group ratio. It appears once, as a bar on an asset-quality slide, and is not mentioned in the prepared remarks or in any published question. The three exposures management chose to quantify were private credit, the Middle East and artificial intelligence. The one that actually deteriorated was not among them.
2. What is inside the ¥46.1bn United States banking subsidiary charge. The loss is disclosed as a line in extraordinary items and as a ¥34bn after-tax entry on the bridge. Which business was sold, to whom, at what multiple, and what remains of the United States banking platform are not addressed. The plan separately lists an exit from the United States digital banking business, but does not connect the two.
3. A reserve-adequacy explanation. Reserves as a share of non-performing loans fell from 105.0% to 74.7%. The presentation shows the coverage and unsecured-reserve ratios, both of which improved, and not the reserve-to-NPL ratio, which fell 30 points. Both presentations are accurate; only one is in the deck.
4. Any quantification of the "+α" on the buyback. The shareholder-returns slide carries "180 +α" and the answer on returns says additional buybacks will be considered during the year. No size, no trigger, no timing. Both of the last two years' top-ups came at the November interim, but management did not say that.
5. The FY3/27 target by business unit. The group targets ¥1,700bn and the three-year plan sets a FY3/29 destination, but the coming year is not broken down by unit. Given that Global is in a declared pause and Global Markets has been labelled non-recurring, the composition of the ¥117bn of guided growth matters more than usual and was not provided.
6. What happens to the plan if the policy rate does not reach 1.25%. The FY3/29 targets assume it does. At roughly ¥110bn of net interest income per 25 basis points, the assumed half-point of tightening is a large share of the distance from 11.4% to 13% return on tangible equity. No sensitivity of the ROTE target to the rate path was shown.
7. Large-corporate loan pricing. The spread on that book fell 5bp to 0.54% while the balance grew ¥3.6tn. The prepared remarks describe the growth and the fee opportunity attached to it; the compression appears only in a footnote-scale table and is never discussed as a trend.
8. Whether the Olive profit target survives a flat rate. Roughly ¥80bn of the ¥110bn FY3/29 target is deposit income. The presentation frames the target as a platform outcome. The rate dependency inside it is disclosed on the same slide and not addressed in the commentary.

Market Reaction

This print produced two distinct reactions in two venues on four different sessions, and the sequence is the story.

• Pre-print setup: the Tokyo line closed at ¥5,852 on May 13, up 16.1% year to date and 64.5% over twelve months, and within 6% of its 52-week closing high of ¥6,204. The New York depositary shares closed at $22.11 on May 12, up 14.4% year to date and 51.2% over twelve months, against a 52-week closing range of $14.20 to $23.81. Both lines entered the print near the top of their ranges after a two-session run of roughly 3%.
• Results day, New York (May 13): the release landed after the Tokyo close, so New York traded it first. The depositary shares gapped up 1.3%, traded between $22.06 and $22.52, and closed at $22.33, up 1.0% on 3.4 million shares against a 2.1 million 30-day average, or 1.6 times normal. The S&P 500 rose 0.6% the same session.
• First Tokyo session with the print (May 14): the considered verdict was negative. The shares opened 2.0% lower, traded down to ¥5,644, and closed at ¥5,675, down 3.0% on 18.6 million shares against a 12.2 million average, or 1.5 times normal. The Nikkei 225 fell 1.0% the same day, so the relative move was about two points. The depositary shares followed, down 2.7% to $21.72 and giving back the whole of the previous session's gain.
• Into the Investors Meeting (May 15 to May 18): both lines drifted, with Tokyo at ¥5,702 and then ¥5,732 and New York at $21.84 and then $21.92, while the Nikkei fell 2.0% and 1.0% on those two sessions. Volume on the depositary shares on meeting day was 2.1 million against a 2.0 million average, exactly normal.
• First Tokyo session after the meeting (May 19): the shares rose 3.7% to ¥5,942 on 21.2 million shares against a 12.1 million 30-session average, or 1.8 times normal, on a day the Nikkei fell 0.4%. That is a four-point relative move and the largest single-session outperformance of the sequence.

The print was sold and the plan was bought. That is an unusual split and it is not hard to explain. What the market received on May 13 was a record it already expected, a guide 3.3% below where the Street sat, and a credit charge ¥88.4bn worse than the company's own plan. Both of the wire reports circulating that day led with the guidance shortfall and attributed it to conservative provisioning against Middle East risk. The Tokyo session on May 14 priced exactly that, and it did so on elevated volume, which argues for a considered institutional response rather than a headline reflex.

What changed on May 18 was the framing, not the numbers. The Investors Meeting delivered the three-year plan with a hard 13% return-on-tangible-equity milestone, the upgrade of the progressive dividend to an explicit annual-increase commitment, the ¥180bn buyback set against ¥100bn at each of the other two megabanks, the raised rate sensitivity, and a shareholder benefit programme attached to a 2-for-1 split. None of that was new information about the year just closed. All of it was new information about the distribution of outcomes ahead, and the market re-rated the shares on it within one session.

The depositary shares are not a clean read on this print. With one American share equal to 0.6 of a common share and a yen that moved from 157.85 to 159.08 across the sequence, the New York line carried both the fundamental move and about eight-tenths of a point of currency translation working against it. Investors sizing the reaction should use the Tokyo line, which is where the price of record is set.

Street Perspective

Debate: Is guidance a forecast or a floor?

Bull view: the bull case treats ¥1,700bn as a number designed to be beaten. It embeds no domestic tightening, a currency assumption below both the year-end and average rates actually realised, and a credit-cost line 16% above the underlying run rate. Last year's guide opened at ¥1,300bn and finished at ¥1,583bn. On the Street's own arithmetic, three-quarters of one 25 basis point hike closes the entire gap to consensus.

Bear view: the bear camp reads the same conservatism as information. Management chose to raise forward-looking provisions to a ¥100bn balance rather than release them, chose to guide credit costs above the clean run rate, and chose to characterise downside risk as the dominant consideration. A company that has just watched its non-performing book rise 53% and is guiding cautiously may simply be right.

Our take: both are true and they resolve at different horizons. The guide is very likely a floor for the November interim, because the rate path and the provisioning both bias upward, and the last two years both delivered mid-year raises. Over the full three-year plan, the caution is better founded than the extrapolation, because the credit trend has been widening for four consecutive reporting periods and none of it has yet run through a recession.

Debate: How much of the record is repeatable?

Bull view: the bull argument is that the company has already done the work of separating signal from noise, and the signal is ¥212bn of after-tax core growth on a ¥1,178bn base. Add a rate cycle that is not finished, a fee engine growing at 16.8%, an overhead ratio down 3.5 points, and a domestic franchise taking share in both deposits and corporate lending, and 18% underlying growth is a better base rate than the sector average.

Bear view: the bear response is that ¥446.1bn of stock gains is 19.4% of ordinary profit and comes from a portfolio the company has publicly committed to shrinking to below 20% of net assets. Management's own three-year attribution puts ¥500bn of ¥800bn of profit growth down to rates and currency. Take out the finite gains and the exogenous drivers, and what is left is a good bank in a good rate environment, not a structurally re-rated one.

Our take: the bear framing is closer to right on the numbers and slightly unfair on the direction of travel. The equity-gains engine has at least three more years to run at the stated pace, and it funds the clean-up charges that would otherwise depress reported earnings, so the practical effect is smoothing rather than inflation. The number to underwrite is somewhere between the ¥212bn of core growth and the ¥405bn reported, and closer to the former.

Debate: Does the credit trend matter at a 0.97% non-performing ratio?

Bull view: the dismissive case is arithmetic. A 0.97% non-performing ratio is low by any global standard, the coverage ratio improved to 69.5%, the domestic bank's own credit cost fell to 7 basis points, and the largest single provision of the year was a forward-looking judgment against a geopolitical scenario rather than a realised loss. Reserves against unsecured exposure rose from 36.9% to 49.6%.

Bear view: the sceptical case is about direction and location. Non-performing loans rose 53% in a year with no recession anywhere in the group's footprint. The Americas balance tripled. Reserves grew 8.8% against a 53% increase in the asset they cover. Roughly ¥131bn of the increase arrived in the final quarter, so the trend was accelerating into the print, and the overseas unit carrying it is simultaneously being told not to grow.

Our take: the level is not the issue and the bulls are right that it does not threaten capital. The issue is that this is the second consecutive year in which a large, previously unflagged credit item has appeared in a different geography, Brazil last year and the Americas this year, and in neither case did the presentation address it directly. Until an investor can see the Americas line stop widening, the appropriate response is to underwrite the ¥340bn credit-cost guide rather than the ¥292bn underlying figure, and to treat the difference as the cost of the uncertainty.

Model Update & Valuation Framework

This is an initiation, so there is no prior model to revise. These are the assumptions we carry in, and where they sit against the company's own guide.

Valuation. At the ¥5,942 Tokyo close on May 19, the shares trade at 1.44 times the ¥4,135.71 net assets per share reported at March 2026, 14.4 times trailing earnings, and 13.3 times the company's guided FY3/27 earnings of ¥447.50 on a pre-split basis. The forecast dividend of ¥180 is a 3.0% yield. A residual-income frame using a 2% terminal growth rate gives a justified multiple of about 1.21 times book at a 10.5% sustainable return and a 9.0% cost of equity, and about 1.46 times at an 11.5% sustainable return, between today's 10.4% and the company's 12% FY3/29 target, at an 8.5% cost of equity. Applied to an estimated FY3/27 book value of roughly ¥4,390 per share, after retained earnings of about ¥268 and the small book dilution from buying back stock above book, that spans ¥5,300 to ¥6,400.

We centre the range near ¥5,900, which is 0.7% below the last close. Adding the 3.0% forecast dividend yield gives a low-single-digit expected total return over twelve months. For holders of the depositary shares, ¥5,900 converts at the May 19 rate of 159.08 yen per dollar and the 0.6 common-share ratio to roughly $22.25, about 1.5% above the $21.92 close on May 18, with the caveat that the dollar outcome moves with the currency and not with the business.

What would move us. To Outperform: two consecutive reporting periods in which the Americas non-performing balance stops widening, or a policy-rate move that the guide does not contain, or a second buyback tranche at the November interim that takes the FY3/27 total above the ¥250bn executed in each of the last two years. To Underperform: evidence that the ¥65bn of Middle East provisioning was a first instalment rather than a sizing, or a stall in the domestic spread with the loan-to-deposit spread failing to hold the 1.17% second-half exit rate.

Thesis Scorecard Post-Earnings (Initiating Coverage)

No prior coverage exists, so this quarter establishes the pillars rather than grading them. The status tags below are the starting positions we will carry into the next report.

Overall: a genuinely strong domestic franchise, the best capital-return policy of the three megabanks, and a credible three-year plan, set against a credit book that widened all year in a geography the presentation did not discuss and an overseas unit that has been told to stop growing. The operating story and the risk story are moving in opposite directions at the same time, which is precisely the configuration that argues for a neutral starting position rather than a directional one.

Action: initiate at Hold. Own the rate optionality and the capital return through the shares only after the credit line stops widening, or at a lower entry. The two dated catalysts are the June annual meeting, which carries the split authorisation and a shareholder proposal on buyback governance the board has opposed, and the November interim, which is when both of the last two buyback top-ups arrived and when the FY3/27 guide has been raised in each of the last two years.