Bajaj Finance — the compounding machine recalibrates

The Pulse

Bajaj Finance is the closest thing Indian lending has to a compounding machine: assets just crossed ₹5 lakh crore, profit grew ~24% to ₹19,332 crore in FY26, return on equity sits near 20%, and bad loans stayed under 1.1% gross — a combination almost no lender sustains for a decade. FY26 was the year the machine recalibrated rather than sprinted. Management pulled back hard on one stressed business (unsecured small-business loans), wound down a money-losing one (captive two-wheeler finance), and — most tellingly — voluntarily took a ₹1,406 crore provision hit to permanently “bulletproof” the balance sheet. The result was a deliberately ugly December quarter masking a still-excellent core. The forward story has two engines: a fast-scaling gold-loan book and a genuine, company-wide bet on AI. The only real debate is the price — the stock trades at ~4.8x book, a multiple that already assumes the compounding continues.

The Business

At its core Bajaj Finance does the simplest thing in finance — borrow at one rate, lend at a higher one — but it does it across an unusually wide and high-yielding spread of products: consumer durables EMIs, personal loans, small-business loans, mortgages, gold loans, two- and three-wheeler finance, plus a securities-broking and a housing-finance subsidiary. That breadth produces a fat ~33% financing margin, well above what a plain secured lender earns, because the mix leans toward high-yield consumer and SME credit rather than thin-spread collateralised loans.

What actually makes it special is captured in one number management repeats: the gap between “4% and 13%.” Bajaj holds roughly 4% of the system’s loan balances but originates about 13% of new loans — meaning it mines its own ~119-million-customer base far harder than rivals, cross-selling product after product to people it already knows and has scored. Rajeev Jain calls the ambition being “the Amazon of financial services in India.” The moat, then, isn’t any single product — it’s the franchise: 119 million customers, 70-plus products, 4,200-plus locations, a top-five financial-services app, and a risk culture that has kept credit losses low even while growing ~25% a year. Ownership is stable and concentrated: parent Bajaj Finserv and the Bajaj family hold ~54.7%, barely moving for years, while institutions slowly accumulate from a broadening retail base.

How Management Thinks

If there’s one self-description that recurs, it’s “risk-first.” Management treats the willingness to deliberately shrink a business as a point of pride, not an apology — “we have never shied away from cutting business… that’s the power of a diversified model.” Rajeev Jain insists he is “a bigger believer in micro than macro… risk is a choice that you make,” and the operating obsession is vintage analysis: tracking how each cohort of new loans performs at 3, 6, 9, 12 months against the explicit benchmark of beating pre-COVID (FY20) loss levels. “I keep reminding them I want to go back to FY20 credit costs — now they are all delivering.”

The defining act of FY26 was the December quarter’s voluntary ₹1,406 crore accelerated provision — a permanent minimum loss-floor applied across every business, deliberately structured as a permanent policy change rather than a reversible “overlay,” because overlays “invite judgment and reversal.” That’s a management choosing to look worse this quarter to be sturdier forever. On candor they grade well: Bajaj was the first large NBFC to flag the small-business-loan stress (Q1), openly admitting it was “a little too suddenly… we are also a little surprised by it,” and disclosing the restructuring quarter by quarter — though by Q2 Jain was visibly impatient to stop the calls being “hijacked by MSME.”

There is also a leadership subtext worth noting: Jain, the architect of the modern franchise, returned to the operating MD role through March 2028 after his designated successor resigned within months — reassuring for continuity, but a reminder that the succession question is unresolved.

Capital allocation is disciplined to the point of being a rulebook: every business gets the same leverage, must clear a return-on-assets and return-on-equity hurdle, and must reach ~$2 billion of scale or they “wouldn’t get into the business.” Earnings are largely retained to fund the book (payout ~15–19%), capital adequacy is kept well above regulatory norms, and the FY24 listing of the housing subsidiary surfaced value while keeping control. The numbers back the words — guidance has been hit or beaten, and they cut guidance honestly (AUM growth trimmed from 24–25% to ~22% mid-year) rather than pretending.

The one genuine red flag management itself keeps naming: “consumer leverage continues to remain an area of concern.” They treat the sheer number of loans a customer carries as the single best predictor of future losses, and prune against it.

Where It’s Going

The forward arc is “cruise mode after a calibration.” After an 18-month clean-up, management guides FY27 to 20–24% AUM growth, return on assets of 4.4–4.6%, and return on equity of 19–20%, with credit costs stepping down to roughly 1.45–1.6% — entering the year, in Jain’s words, “with tailwinds, not headwinds.” Two specific growth engines stand out. Gold loans are exploding (up 115% in FY26 to ~₹18,000 crore) and targeted to exceed 5% of assets, much of it originated digitally through the app by existing customers. And AI (“FinAI”) is the multi-year thesis: an AI call-centre agent that costs a third of a human, processing 600,000 loans a day at Diwali versus 100,000 before, 363 dedicated AI staff, and 600–800 autonomous agents planned for FY27. The pitch is that AI eventually lowers both operating costs and credit losses.

The honest tensions: the small-business book is still healing (management expects the “worst behind us by June 2026”); the captive two-wheeler business, though now under 1% of assets, still throws off an outsized share of bad loans as it runs off; and management explicitly made FY27 guidance contingent on geopolitical and macro stability, naming “one year, three crises” as the new normal. Gold-price volatility is a live planning headache too — “gold went from $5,500 to $4,500 in one week… makes it hard to do a budget planning session.”

The Four Checks

Quality & moat (gate). Yes — a genuine, durable moat, and a rare one in lending. It is not any single product but the cross-sell franchise: a vast, already-scored customer base mined at ~13% of system originations off ~4% of balances, layered on data, an app ecosystem, and a low-loss risk culture. Sustaining ~25% growth for a decade without sacrificing asset quality is the proof. This is a high-quality business.
Returns on incremental capital & runway. Excellent and durable. Return on equity has run in the high-teens to low-20s for a decade (18.2% headline, ~20% in the latest quarter), profit has compounded ~34% over five years, and the runway is long — management’s own framing (“barely scratched the surface” of India’s middle class; a 200-million-customer ambition by FY30) is credible given India’s low credit penetration. Incremental capital still earns well above the cost of capital.
Capital allocation for the stage. Rational for a still-fast-growing lender. Retaining most earnings to fund a high-return book is correct; the modest payout and above-norm capital buffers are appropriate; the rulebook (same leverage, hurdle rates, $2bn scale test) imposes genuine discipline; and the voluntary ECL floor is exactly the kind of conservatism you want from a leveraged compounder. No buyback — which is right when the stock trades at ~4.8x book and capital still earns ~20%.
Price. Demanding — and knowingly so. At ~28.6x earnings and ~4.8x book against an ~18–20% ROE, the market prices Bajaj as a quality compounder, not a typical NBFC (the snapshot flags valuation as the main concern). The maths only works if the growth-plus-quality combination persists for many more years. It has so far, and the FY27 setup (easing credit costs, gold + AI engines) supports the case — but there is little margin of safety left in the multiple itself. The valuation rests on durability, which is a bet on the moat holding.

Sources

Concall transcripts read: Q1 FY26 (call 24 Jul 2025), Q2 FY26 (10 Nov 2025), Q3 FY26 (3 Feb 2026), Q4 FY26 (29 Apr 2026).
Annual reports read: FY23, FY24, FY25 (high-signal sections).
Financial snapshot: screener.in, fetched 2026-06-09 (FY26 revenue ₹81,982 Cr, PAT ₹19,332 Cr, ROE 18.2%, P/E 28.6x, P/B 4.83x, GNPA/NNPA 1.01%/0.41%).
Research dumps (not published): vault/Sources/Earnings/Bajaj Finance Ltd/.
Notes: the Dec-2025 quarter’s reported profit fell on the voluntary ₹1,406 Cr ECL provision + a ₹265 Cr labour-code charge, partly offset by a ₹1,416 Cr housing-subsidiary stake-sale gain booked below the line — core profit grew ~23%. A FY26 recovery-netting presentation change shifts the optics of the credit-cost ratio (no profit impact).