Home

Deck

Kaspi.kz · KSPI · NASDAQ

Kaspi.kz runs Kazakhstan's dominant consumer super-app — payments, marketplace and consumer credit on one bank charter, used monthly by 78% of the country's adult population and now expanding into Türkiye through Hepsiburada.

$78.58
Price (ADS, 10 Jun 2026)
$15.1B
Market cap
$7.8B
Revenue (FY25)
15.7M
Monthly active users (KZ)
Listed on NASDAQ in January 2024 at $96; ran to $138 by July 2024 — now $79, 43% below the peak and 18% below the IPO 17 months in.
2 · The tension

7x P/E on a 51% ROE — but which earnings are you paying for?

  • Bank multiple, software margins. KSPI trades at 7.1× trailing P/E versus an EM consumer-finance peer median near 19.5× (Nu 28×, MELI 51×), yet generated 51.2% ROE and 55.4% operating margin in FY25 — held through an NBK rate hike from 15.25% to 18% at a 6.0% cost of risk.
  • The Hepsiburada step-down is the live debate. Pre-deal Kaspi printed ~75% operating margin every quarter of FY24; consolidating the Turkish marketplace from Q1 25 reset the band to 54–60% across five consecutive quarters. Bull calls it a transitional floor; bear calls it the new ceiling.
  • The Q2 print is the arbiter. Consolidated operating margin above 58% on ~10 August 2026 would validate the engine and open the door to sell-side margin upgrades; at or below 52% with Hepsiburada missing EBITDA breakeven supports the bank-multiple anchor.
An engine that earns 51% ROE through an 18% base rate at 7× earnings is either a mispriced franchise or the right price for a governance and Türkiye-execution binary. That is the whole debate.
3 · Money picture

Bank balance sheet, super-app margins, cyclical-commodity multiple.

51.2%
Return on equity FY25 (79% FY24)
55.4%
Operating margin 75% pre-Hepsiburada
7.1×
P/E (TTM) peer median 19.5×
93%
Loan book deposit-funded 100% loan-to-deposit

The 20-point margin reset from FY24 to FY25 is the Hepsiburada consolidation, not engine decay — quarterly operating margin has stabilized in a 54–60% band across five post-deal quarters. The balance sheet still funds itself on KZT customer deposits ($15.0B against $15.0B of loans), so the lending book grows without wholesale debt or equity dilution. Free cash flow conversion has dropped from 100%+ pre-deal to 46% — the variable to watch is whether that recovers as Türkiye integration completes.

4 · The engine

Five interlocking pillars stacked under one Kazakhstan regulator.

  • One bank, one app, three profit pools. Payments, Marketplace and consumer credit run on the same identity and the same deposit book — 78% of Kazakhstan's adult population on the app, DAU/MAU at 68%, 77 monthly transactions per active user.
  • Deposit funding is the engine. $15.4B of customer deposits, 93% in KZT, fund the loan book at sub-market cost; the spread between that funding cost and a 24% Fintech yield is where most of net income lives.
  • AI underwriting, six seconds. 99.9% of consumer-credit decisions are made in under six seconds against in-app behavioural data; cost of risk held at 6.0% through the NBK rate move from 15.25% to 18%.
A Kazakh competitor would need a bank charter, a daily-habit app, an AI underwriting stack and government-services distribution at the same time. No one has cleared all four in 18 years.
5 · The overhang

A 7× multiple buys you a real franchise and an unresolved disclosure tail.

  • Live US class action. Krivenok v. Kaspi.kz (24-cv-10926, C.D. Cal., filed December 2024) alleges undisclosed Russia exposure and related-party transactions; Rosen Law was still soliciting class members on 25 May 2026 with no motion-to-dismiss ruling 18 months in.
  • The Culper short report has never been answered. Management has not directly addressed the September 2024 Culper allegations across seven CEO letters, and the 20-F still discloses an unremediated material weakness in internal control over financial reporting.
  • Insider flow is one-way. Co-founder Kim sold ~$12.3M of ADSs in May–June 2026, including into post-rally weakness; 70+ insider and major-holder filings since IPO with zero open-market buys, while the founders together still own 43%.
Tencent, US endowments and a $600M Eurobond oversubscribed at 5.90% in April 2026 say outside capital does not view the disclosure tail as unmanageable — but a clean motion-to-dismiss is what would actually clear it.
6 · Catalyst calendar

Three observable events in the next six months decide the rerate.

  • Q2 2026 earnings, ~10 August. Consolidated operating margin above 58% with Hepsiburada at EBITDA breakeven would open the door to sell-side margin upgrades; at or below 52% would confirm the structural reset and support the 5.5× bank-multiple frame.
  • Rabobank A.Ş. BDDK approval, mid-2026 guided. The Turkish banking licence is what converts Hepsiburada from a loss-making marketplace into a deposit-funded super-app in an 85M-population market — without it, the ~$1.1B paid for the option becomes a multi-year drag.
  • Krivenok motion-to-dismiss, window open. 17 months past filing and inside the typical 12–24 month MTD band; a clean dismissal with prejudice would clear the only binary multiple-cap ahead of any operating data, while a denial would confirm a multi-year litigation tail.
The cleanest positive setup since IPO — a Tencent anchor block, $97 peak, golden cross on 18 May — round-tripped to $79 in two weeks on rising volume. The tape has not yet validated the post-overhang frame; only the events above test it.
7 · Bull & Bear

Lean long, wait for confirmation — the engine is real, the binary keeps it on a watchlist.

  • For. 51% ROE survived a full NBK tightening to 18% on 6.0% cost of risk and 93% deposit funding — no other EM consumer-finance peer prints these economics at this multiple.
  • For. Tencent, Lomtadze, US endowments and a $600M BBB- Eurobond at 5.90% in April 2026 are independent third-party validation that the post-Culper governance overhang is manageable, not disqualifying.
  • Against. Group operating margin has now printed inside a 54–60% band for five consecutive post-deal quarters; pricing this as a 75%-margin business is anchoring to a print that no longer exists.
  • Against. The Krivenok motion-to-dismiss is genuinely binary on a single docket event management has not publicly addressed; the May–June round-trip from $97 to $79 came on rising volume — distribution, not consolidation.
My view: a 51% ROE franchise at 7× is too cheap to dismiss, but a confirmed position needs both a Q2 operating margin above 58% and a clarifying Krivenok docket event. Without those, this is a long bias on a watchlist, not a position.

Watchlist to re-rate: Q2 2026 consolidated operating margin (≥58% bull / ≤52% bear); Rabobank A.Ş. BDDK approval announcement; Krivenok 24-cv-10926 docket activity.