Kaspi.kz: A Deep-Value Investment Study

There are plenty of dependencies, but they are broadly understandable. The company does not rely on a single large customer; the real dependencies are: first, the stability of Kazakhstan's domestic digital-payments and e-commerce ecosystem; second, the continued dominance of the Kaspi brand and app entry point; third, regulatory licenses and the bank capital framework; fourth, the product-execution culture of management. In 2025 Turkey became a new dependency, because Hepsiburada is already an important growth source; in 1Q2026 management even said Türkiye already accounted for roughly half of the company's e-Commerce GMV.

If the stock market were switched off for five years, I would be willing to hold this business itself, but only on the condition that I accept and understand its legal, sovereign, currency, and regulatory environment. If I had no read on those conditions and held it simply because "the P/E is low," I would in fact feel uneasy. Business comprehensibility score: 4/5.

Industry and Competitive Landscape

In terms of industry stage, Kaspi.kz sits not in a single mature track but at the intersection of three long rising curves: the electronification of payments, the penetration of online retail, and the spread of digital finance. Internet penetration in Kazakhstan is already very high: data from the World Bank / FRED indicate that in 2024 internet users were about 93.4% of the population. Kazakhstan's e-commerce penetration reached 15.3% in 2024 and rose further to 17.1% in the first half of 2025; at the same time, national statistics and payments data show cashless payment volumes are still growing. In February 2025, cashless transaction volumes on Kazakhstan-issued bank cards grew 21.4% year over year, reaching KZT 13.4 trillion. In other words, the "digital soil" Kaspi grows in is still thickening, not peaking.

Long-term demand in this industry is broadly stable. Residents always need to pay, shop, borrow, transfer money, and handle government affairs; as long as a platform holds the "entry point," demand fluctuation shows up more as the size of GMV/TPV/TFV than as a question of whether demand exists at all. Kaspi's most important feature is that it has compressed all of these needs into a single high-frequency entry point. Official materials show that in 2025 the number of Kazakh users accessing Government Services through the Kaspi Super App reached 12.2 million, showing it serves not only consumption and finance but is also deeply embedded in everyday administrative scenarios. That kind of usage frequency and mindshare is usually more stable than a single bank's.

The real threat is not "demand disappearing" but the platform entry point being displaced, regulatory rules changing, and cross-border expansion failing. Within Kazakhstan, the integrated competitor most worth watching is Halyk Bank. Halyk officially describes itself as Kazakhstan's largest financial group, and its app explicitly calls itself a "super app"; on Reuters/LSEG figures, Halyk currently trades at an adjusted P/E of about 3.89x, P/B of about 1.17x, and P/CF of about 3.81x. In other words, Kaspi's rival is not weak, and it does not operate in a vacuum free of challengers. It is just that, on brand, user engagement, and the cross-scenario integration of its e-commerce and merchant networks, Kaspi remains clearly stronger for now.

In e-commerce and local services, Kaspi's edge is a local closed loop rather than global scale. In 2025 its Marketplace Active Consumers reached 8.8 million, the Marketplace Take Rate rose from 9.2% in 2023 to 10.5% in 2025, and the e-Commerce Take Rate rose from 11.0% to 12.7%; this shows the company is not propping up GMV through subsidies but is deepening monetization while it grows. But once it enters Turkey, the competitive landscape becomes markedly more complex and regulation more cumbersome. In its annual report the company specifically disclosed Turkish e-commerce regulatory risk, and in 4Q/FY2025 it said it intended to keep Hepsiburada near "Adjusted EBITDA breakeven" while continuing to add investment in delivery, advertising, BNPL, and user experience. In other words, the industry environment in the international-expansion phase is clearly weaker than in domestic Kazakhstan.

In terms of profit-pool concentration, Kaspi is a strong profit capturer at home. In 2025 all three of Kazakhstan's main domestic platforms still made money, with Payments and Fintech both maintaining extremely high profitability; but after the Heps consolidation, the consolidated net margin fell from 41.7% in 2024 to 26.4% in 2025, showing the company now holds a combination of a "high-quality parent + low-margin expansion vehicle." That change matters: it does not break the parent's moat, but over the next few years it will dilute how good the consolidated figures look.

My judgment is that Kaspi corresponds not to "an excellent company in a poor industry" but more to a strong platform in a good industry; only this "good industry" holds mainly within domestic Kazakhstan, and once it crosses into Turkey, the industry's attractiveness must be discounted. Industry attractiveness score: 4/5.

Moat and Management

Kaspi's moat — its strongest part — is not patents, nor a single cost edge, but a composite moat of brand + network effects + data loop + high usage frequency. At year-end 2025 the company had 15.7 million Average MAU, 10.7 million Average DAU, and 764,000 active merchants in Kazakhstan; Payments Active Consumers were 14.6 million, Marketplace Active Consumers were 8.8 million, and Fintech Active Consumers were 6.2 million (deposits) and 6.3 million (loans). A two-sided network of this scale makes both consumers and merchants more willing to stay on the platform, and makes new services easier to seed with initial penetration.

The brand advantage is real. In its 4Q/FY2025 presentation the company disclosed that the Kaspi brand ranks first across key categories such as "phone-installation rate, payments awareness, e-commerce awareness, low-price mindshare, travel awareness, and car-purchase-app awareness"; relative to the "nearest brand," its payments awareness even reaches 13.2x. While this is a survey the company commissioned from KResearch Central Asia rather than a fully independent third-party industry report, given its MAU, DAU, and transaction-frequency data, this brand advantage is not an empty slogan.

The network effect is also quite strong. For consumers, more merchants, more payment points, and more lending/BNPL/advertising/delivery/government services raise retention; for merchants, more consumers and higher payment-completion rates raise the willingness to onboard. At year-end 2025 the company stressed on both the consumer and merchant sides that "penetration is still low" — for example, advertising, delivery, merchant finance, and merchant term deposits are all still in early stages. A platform that is already large but still has untapped secondary-monetization room is usually worth more than one that is "already saturated with no room to thicken."

The data and risk-control advantage is a crucial layer of the moat. Fintech 2025 TFV was KZT 11.652 trillion, the Average Net Loan Portfolio was KZT 6.415 trillion, the Fintech Yield held steady at 24%, and the Cost of Risk was just 2.2%; the company attributes the low, stable loss rate to continually improving data-driven underwriting and collection capabilities. For a platform doing consumer credit, BNPL, and merchant finance, keeping the loss rate at this level amid high growth is a strong operating advantage.

Switching costs and channel advantages are not the traditional contract-lock-in kind, but they still exist. Users are not "legally unable to leave"; rather, leaving means losing a unified entry point that simultaneously holds payments, shopping, deals, lending, government services, travel, and classifieds. Merchants are not "only able to connect to Kaspi" either, but once they leave, they lose collections, merchant marketing, delivery, merchant financing, and consumer-traffic synergy. Such soft switching costs are often stronger than they appear on the surface.

License and regulatory barriers matter just as much. Kaspi Bank is bound by bank capital-adequacy requirements; on a Basel III basis both the total capital ratio and the Tier 1 ratio were 19.6% in 2025, clearly above the minimum; the company has also obtained investment-grade ratings of Fitch BBB- and Moody's Baa3, and in April 2026 issued $600 million of senior unsecured notes due 2031 with a 5.9% coupon. Building credit at both ends — local financial infrastructure and international capital markets — is part of the barrier.

As for the direction of the moat, I think it is widening, or at least holding steady, in domestic Kazakhstan; in Turkey it cannot yet be called "widening," and "to be proven" is the better description. If a competitor wanted to replicate Kaspi in Kazakhstan, it would need not just capital but a payments network, brand trust, merchant acceptance, government-cooperation interfaces, risk-control data, and high-frequency scenario integration — none of which can be copied in a year or two. By contrast, Turkey's competitive environment is more open and more crowded, so Heps does not automatically inherit all of Kaspi's advantages from Kazakhstan.

On management, ownership alignment is a clear plus. As disclosed at year-end 2025, Mikheil Lomtadze holds 22.58% and Vyacheslav Kim holds 20.74%, management collectively holds 3.87%, and the two founders together already exceed 43%. This makes them less "salaried professional managers" and more like long-term owners.

On capital allocation, the company both pays dividends and buys back stock. Between 2022 and 2025 the company repurchased a cumulative 4.5965 million GDR/ADS for a total outlay of KZT 149.134 billion; by year-end 2025, shares outstanding moved from 189.3 million at year-end 2023 to 190.2 million at year-end 2025 — essentially flat — showing buybacks largely offset the dilution from equity incentives. In 1Q2026 the board recommended a dividend of KZT 850 per ADS, a 64% payout ratio; in the 4Q/FY2025 results management also said this level was "sustainable at least for the remainder of 2026." This reflects management's confidence in cash generation, but also shows it does not intend to plow all profits into high-risk international expansion.

The places to dock points are equally clear. First, while the board has six members, of whom four meet independence standards, the company explicitly states it uses the foreign-private-issuer exemption and that the board is not required to meet Nasdaq's "majority-independent" requirement. Second, related-party transactions are not zero: the company has business dealings with both Kolesa and Magnum, and the two core figures each have a shareholding or control relationship with those entities. Third, the company had a material weakness in internal control in 2023; although it was fully remediated in 2025, Hepsiburada still had a material weakness in internal control in 2024. For long-term investors, all of this means the governance discount should not be simply ignored.

On the whole, Kaspi's moat is strong at home, but the combination of "controlling-shareholder culture + emerging-market governance discount + overseas expansion" makes the "investability" of this moat lower than its "admirability." Moat strength score: 4/5. Management and capital allocation score: 3.5/5.

Financial Quality and Owner Earnings

Start with the key financials. The table below is mainly compiled from the company's 2023 and 2025 Form 20-F and the FY2025 results files; the "free cash flow" line is only an approximate observation, because for a bank/financial platform, operating cash flow is heavily affected by loan expansion, changes in customer deposits, and regulatory reserves, and cannot be equated directly with distributable cash the way an industrial company's can.

Sources: company Form 20-F 2023/2025, FY2025 results; ROE is a simplified estimate based on beginning and ending equity attributable to shareholders, not the company's own basis.

These data point to three things.

First, the growth is real, but the basis must be viewed separately. Over 2021–2025 revenue rose from KZT 936.8 billion to KZT 4.046 trillion, a very high compound rate; net income attributable to shareholders rose from KZT 431.9 billion to KZT 1.073 trillion, an equally strong rate. But the 2025 consolidated net margin stepped down clearly, not because the Kazakh parent suddenly worsened, but as the combined result of the Heps consolidation, the scaling of retail revenue, and higher funding costs and provisioning requirements. In both the FY2025 and 1Q2026 results management stressed that the core focus should be "the strength of the underlying business" rather than a single consolidated net margin.

Second, profit is broadly real profit, but CFO cannot be used mechanically. In Kaspi's operating cash-flow statement, changes in loans, customer accounts, mandatory reserves, and other operating liabilities can massively distort headline CFO. For example, in 2025 operating cash flow was KZT 673.6 billion, while loans rose by KZT 1.663 trillion, customer accounts rose by KZT 974.5 billion, and mandatory reserves rose by KZT 247.8 billion; these are uncommon for an industrial company but are the operating core for a financial institution. So taking "CFO – capex = FCF" straight into a valuation would overstate or understate the company's true distributable cash. A more reasonable approach is to approximate owner earnings as net income attributable to shareholders – maintenance spending – the regulatory capital needed to support growth.

Third, the balance sheet remains sound for now. At year-end 2025 the company's customer accounts were KZT 7.531 trillion and loans were KZT 7.172 trillion, a loan/deposit ratio of about 95%; the Basel III total capital ratio and Tier 1 ratio were both 19.6%, clearly above the regulatory floor; NPL was 6%, with provisions covering about 80% of NPL. At the same time, in April 2026 the company successfully issued $600 million of 2031 notes, with the deal about 3.5x oversubscribed, showing the external debt market is still willing to extend it liquidity and credit.

On accounting and signs of manipulation, I see no direct evidence of fraud, but I would put two items on the red-flag list: first, goodwill on the 2025 balance sheet jumped from KZT 17.44 billion to KZT 447.13 billion, mainly from the Hepsiburada acquisition; second, Hepsiburada had a material weakness in internal control in 2024. The former means impairment may appear in the future; the latter means consolidation quality and audit integration still need time to verify. On the other hand, the Kaspi parent's own 2023 material weakness was deemed fully remediated in 2025, and external auditors recognized its 2025 internal control as effective. The conclusion is not "no risk" but "risk concentrated in overseas-expansion integration, not the hallmarks of domestic financial fraud."

Now to Owner Earnings. I use a conservative framework adapted to a financial platform:

• Fact: 2025 net income attributable to shareholders was KZT 1,073.18 billion.

• Fact: among 2025 non-cash items, share-based compensation was about KZT 15.48 billion and depreciation and amortization about KZT 78.25 billion.

• Assumption: maintenance capex is estimated at about KZT 80 billion. The reason is that the 2025 total capex of KZT 182.5 billion clearly includes the Heps consolidation and expansion components, rather than pure maintenance spending.

• Assumption: the minimum incremental capital that must be retained to support growth is about KZT 255 billion, based on 2025 NBK-basis risk-weighted assets rising from KZT 805.9 billion to KZT 997.9 billion and a Tier 1 capital ratio of about 12.7%.

On this basis, 2025 conservative-basis Owner Earnings can be roughly estimated at about KZT 830 billion, equivalent to about $1.65 billion. This figure is markedly below accounting net income, but I think it is closer to the cash capacity that is "distributable to owners over the long run without harming growth and capital adequacy." Measured against the current market cap of about $16.66 billion, that implies a P/Owner Earnings of about 10x. This is not expensive, but it is not cheap enough to ignore sovereign, regulatory, and integration risk either.

Valuation and Margin of Safety

As of the latest quote, KSPI trades at about $87.78 with a market cap of about $16.66 billion.

Based on 2025 net income attributable to shareholders, equity attributable to shareholders, and the conservative-basis Owner Earnings above, I arrive at the following set of derived values:

Sources: KSPI current market cap and price, FY2025 equity/income attributable to shareholders, 2025 recommended dividend, Adjusted EBITDA and balance-sheet data; EV/EBITDA is a simplified derivation that excludes customer deposits.

Owner-Earnings Discount Method

Here I do not want to offer a "precise to the last digit" pseudo-precise valuation, but rather three more honest ranges.

The key premise of these ranges is the conservative Owner Earnings base of about $1.65 billion noted earlier, together with differentiated assumptions for growth, capital retention, and discount rate over the next 10 years. The most fragile assumption is not the growth rate, but whether the discount rate and regional risk premium should stay elevated long-term. If you require a 14%–15% equity return, today's price is not very cheap; if you think 11%–12% is enough, today's price is attractive.

Relative Valuation Method

Putting Kaspi alongside several "comparable but not fully homogeneous" assets reveals a very interesting discount.

Sources: KSPI current price and market cap, Halyk Reuters/LSEG search results, Nu Reuters search results, MercadoLibre and Block current market data. The accounting bases, regional risk, and business structures of these companies are not consistent, so this table is for "directional comparison" only, not mechanical pricing.

What relative valuation tells me is not that "Kaspi should be as expensive as MELI," but that the market clearly does not price Kaspi as a pure high-quality platform; it keeps applying a regional and governance discount. That discount is not unreasonable, but it also means that as long as the fundamentals do not deteriorate, the stock tends not to get too expensive.

Asset-Value Method

For a platform-type financial institution like Kaspi, a liquidation valuation is not the best primary method, but it works as a floor check. At year-end 2025 equity attributable to shareholders was about $4.929 billion; stripping out only goodwill, rough tangible equity attributable to shareholders was about $4 billion. On the other hand, the fair-value information the company disclosed shows the carrying value of customer loans is close to fair value, and the carrying value of customer accounts is close to fair value as well, indicating most financial assets and liabilities show no extreme divergence; what truly warrants caution is the goodwill-impairment risk from the Heps consolidation. In other words, book value has some reference value, but Kaspi's investment value comes mainly from future earning power, not liquidation assets.

Combining the three methods, I offer the following price framework:

So, at the current price around $87–88, the stock sits in a "reasonable, leaning attractive" zone rather than a "clearly cheap" one. This is the core reason I give it "Watch" rather than a direct "buy."

Risks, Comparisons, and Final Conclusion

Kaspi's most important risk is not share-price volatility, but the following scenarios that could lead to permanent loss of capital.

The first category is sovereign and regulatory risk. The vast majority of Kaspi's profit base is still in Kazakhstan, which currently remains in a high-rate environment. In April 2026 Kazakhstan's central bank held the base rate at 18%, with March inflation at about 11.0%. High rates alone need not destroy Kaspi, but they mean any regulatory change to reserves, deposit rates, and the pace of consumer credit could directly affect funding costs and margins. For ordinary investors, the hardest part is not whether they can read the financials, but whether they can consistently judge this policy environment.

The second category is Turkish expansion and M&A integration risk. After acquiring Hepsiburada, goodwill on Kaspi's balance sheet jumped, and inventory, trade payables, and other liabilities all expanded noticeably; in 4Q/FY2025 the company also explicitly said it would keep Hepsiburada near Adjusted EBITDA breakeven while investing around delivery, advertising, BNPL, and UX. This means that for some time ahead, Türkiye looks more like a "value-creation testing ground yet to be proven" than a mature asset that immediately thickens per-share value. If the Turkish business cannot turn a profit for years, or needs continuous additional capital, Kaspi's "high-quality parent" will be suppressed long-term by a valuation discount.

The third category is governance and related-party transaction risk. The founders' high ownership is alignment; but high control in an emerging market can also bring a reflexive discount. Kaspi has related-party transactions with Kolesa and Magnum, and the company concedes its board does not follow Nasdaq's general requirement for a majority of independent directors. Layer on Hepsiburada's 2024 material weakness in internal control, and all of this means investors should apply a governance discount to the company over the long run, rather than take for granted the governance standards of a high-quality domestic U.S. platform.

The fourth category is valuation-misjudgment risk. Many people will be drawn to a 7–8x P/E, but a low valuation does not automatically equal a high margin of safety. If the market grants this discount because it believes long-term that Kazakhstan/Turkey and governance risks cannot be eliminated, then even with steady profit growth, the valuation multiple may not re-rate noticeably. Put differently, Kaspi could well be a company that is "a good company, but one that can only earn a discounted valuation long-term." Returns of that kind come more from earnings and dividends than from multiple expansion.

The strongest contrarian view I would phrase this way: You think you are buying an "emerging-market high-quality platform," but in reality you may be buying "a high-quality domestic business + long-term cross-border expansion uncertainty + a country discount that will not go away." What a short seller might see is not whether Kaspi makes money today, but whether it can transplant its "Kazakhstan playbook" to a more complex Turkey, and whether it will burn through its otherwise excellent record of capital allocation in the process. This is a very strong contrarian argument, and I think it must be faced head-on.

What facts would overturn the current judgment? Quite clearly:

Comparing it with other opportunities, my view is this: Relative to the risk-free rate, Kaspi clearly offers more potential return today — the U.S. 10-year Treasury yield is currently about 4.5%–4.6%, while on a conservative owner-earnings basis Kaspi's initial yield is close to 10%. But this does not mean it is necessarily more "suitable for everyone," because those extra five points are, in essence, the regional, regulatory, governance, and M&A-integration risk you are taking on.

Relative to a broad-based index, Kaspi's potential return may be higher, but with less certainty than an index. An index's value lies in diversification, judicial transparency, and more consistent governance standards; Kaspi's value lies in a higher internal rate of return and a deeper valuation discount. If you are an ordinary investor, I do not think buying it is "clearly better than buying an index"; if you know emerging markets and are willing to follow regulation and company data closely, it can be a single stock to watch with more alpha than an index.

If a portfolio could hold only five assets, my answer is: for most ordinary investors, it does not make the top five; for those familiar with Central Asian / emerging-market internet finance, it can serve as a marginal 4th–5th slot, not a core holding.

Below is the Checklist organized as you requested.

Open questions and limitations: This report can fairly solidly judge the quality of Kaspi's domestic business, but my judgment on the "3–5-year return-on-capital path after the Turkish consolidation" can still only be rated medium confidence; moreover, the FCF, EV/EBITDA, and net-debt/EBITDA measures commonly used for traditional industrial companies do not fully fit a platform-style financial institution like Kaspi, so the related valuations can serve only as supporting tools, not as the primary anchor.

Finally, the requested final conclusion.

To distill it into the most "Buffett-style" statement: Kaspi.kz looks more like "a very good business" than "a no-brainer trade." If you understand it, it may be cheap; if you do not understand it, no price is necessarily safe.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.