ERPNext for Large Enterprises in India (2026): Proof, Scale Limits & the SAP Question

If you run a large Indian business, your ERP shortlist probably reads SAP S/4HANA or Business One, Oracle NetSuite, Microsoft Dynamics — and, increasingly, one open-source line item someone on your team added: ERPNext. This guide is the due-diligence file on ERPNext for large enterprises: who runs it at scale, where it genuinely strains, what your auditors will ask, and what the numbers look like over five years.

Who already runs ERPNext at enterprise scale in India?

The strongest evidence isn't a feature list — it's who bet on it.

Zerodha is the flagship. India's largest stockbroker runs its internal operations on ERPNext and the Frappe Framework, loading 25–30 million records a day, with a famously small technology team. CTO Kailash Nadh has been public about why: full source access, no per-user tax, and the ability to bend the system to their workflows instead of the reverse. Zerodha's conviction went beyond usage — its Rainmatter arm put ₹10 crore into Frappe Technologies, the Mumbai company that builds ERPNext. For a CFO asking "will this vendor exist in ten years?", an open-source codebase plus a profitable maintainer with patient Indian capital is a materially different risk profile from a venture-funded SaaS.

Beyond fintech, Frappe's published customer stories include Indian manufacturers at real scale: Banaraswala Wire Mesh — running ERPs since the 1980s — moved to ERPNext after evaluating TCS and Salesforce builds; A-2 Agro runs a 2-lakh-eggs-a-day operation on it; a valve manufacturer standardised 90,000 SKUs on a customised deployment. These aren't pilot projects; they're primary systems of record.

Two things worth noting about this list, honestly. First, most public ERPNext references are India-headquartered — if you need twenty country localisations run by one global instance, SAP's reference base is deeper. Second, "runs ERPNext" at enterprise scale almost always means "with a serious implementation partner or in-house Frappe team", not a self-service install. Budget accordingly.

Can ERPNext handle 500–1,000 users?

Yes — with engineering. This is the question where you should distrust both the cheerleaders and the dismissals.

The architecture is standard horizontally-scalable web stack: Python application servers, MariaDB/Postgres, Redis queues, background workers. Frappe's enterprise tier runs deployments with thousands of concurrent users on Frappe Cloud, with priority SLAs and database-scalability consulting. Self-hosted, the same results are achievable on Indian cloud or on-prem infrastructure — we've written about the mechanics in our MariaDB tuning guide — but they are designed, not defaulted.

Where a large deployment actually strains, from experience:

• Unoptimised custom reports. A query report that scans five years of GL entries will hurt at 50 users and paralyse at 500. Report discipline (indexes, date-bounded defaults, scheduled exports for heavy analytics) is an operating rule, not a nice-to-have.
• Permission complexity. Hundreds of users with row-level User Permissions multiply query cost. Design the permission model around roles and companies first; per-user exceptions are the expensive path.
• Month-end concurrency. Stock revaluation, depreciation runs, and GSTR preparation stack up at the same time. Worker-queue sizing and scheduling windows solve this; ignoring it produces the "ERPNext is slow" review.

The realistic framing: at 500+ users you are operating a business-critical platform, and it needs the same platform discipline you would apply to SAP Basis — just with dramatically cheaper licences and full access to the source when something needs fixing.

What does a large Indian company need that an SME doesn't?

Four things come up in every enterprise evaluation we've been part of — and all four are core-product territory, not add-ons.

Multi-entity consolidation. Group structures — a manufacturing company, a trading arm, an LLP for services — run as separate companies in one ERPNext instance with inter-company transactions, shared or separate charts of accounts, and consolidated financial statements. Our multi-company setup guide covers the mechanics, including what to keep separate for audit cleanliness.

GST at enterprise shape. Multiple GSTINs per entity, e-invoicing with IRN generation, e-way bills, TDS/TCS sections, and GSTR-1/3B preparation are native. At enterprise volume the differentiator is that compliance changes ship in the open codebase within days of a mandate — no waiting for a localisation pack release through a partner channel.

The MCA audit-trail mandate. Since the Companies (Accounts) Rules amendment took effect, Indian companies must use accounting software with a non-disableable audit trail (edit log) on every transaction. ERPNext's version-tracking on documents provides this, and your statutory auditors will test it — have your partner demonstrate the edit log and its immutability settings in the demo, not after go-live. We cover the broader control surface in How Secure Is ERPNext?

Maker-checker and workflow control. Multi-level approval workflows (amount-banded PO approvals, credit-limit overrides, journal approval) are configuration, not customisation. The practical enterprise question is discipline: who owns workflow definitions, and how are changes themselves approved? Treat workflow config as controlled change, the way you would an SAP transport.

When should a large company still choose SAP or Oracle?

An honest section, because credibility here is worth more than a conversion.

• Global consolidation across many localisations. If you close books across 15–30 countries, SAP and Oracle ship maintained localisations for nearly all of them. ERPNext's localisation depth outside India and a handful of markets varies, and you'd be underwriting some of it yourself.
• A parent or investor mandate. If the board's reporting stack is SAP-aligned, the integration and audit familiarity argument usually wins regardless of TCO.
• Certified industry add-ons with no Frappe equivalent. Pharma serialization/track-and-trace, automotive EDI networks, and similar certified ecosystems have decades of SAP investment. Verify the specific capability exists in the Frappe ecosystem — or price building it — before deciding.
• Appetite for accountability over ownership. With SAP you buy a throat to choke; with ERPNext you own the system and choose your partner. Some leadership teams genuinely prefer the former. Know which you are.

If none of those four conditions apply to you, the economics of ERPNext for large enterprises become very hard to argue against. And if you're currently on SAP and evaluating the reverse move, we've documented what a SAP-to-ERPNext migration actually involves — timeline, what migrates cleanly, and what doesn't.

What does it cost at 250 users over five years?

Approximate 2026 figures for an Indian deployment, in lakhs — treat these as modelling inputs and verify against live vendor quotes, which vary by negotiation and scope:

Three observations that survive any quibble with individual numbers. The licence line is structurally zero and stays zero as you add users — headcount growth doesn't reprice your ERP. The implementation line is where ERPNext projects go wrong when scoped naively; our hidden costs breakdown applies double at enterprise scale. And the internal-team line is honest: at this size you want 1–2 in-house Frappe-literate engineers regardless of partner support — still a fraction of a proprietary suite's annual maintenance alone. For the ROI arithmetic, the ERP ROI framework works unchanged at enterprise inputs.

How do large companies de-risk the switch?

The pattern that works, in order:

1. Name the sponsor and the owner. A director-level sponsor who can kill scope creep, and a full-time internal project owner. ERP projects fail from under-resourcing the buyer's side far more often than from software.
2. Pilot one entity or one plant. Pick the unit with clean-ish data and a motivated GM. Run it as the reference implementation; let internal word-of-mouth do the change management.
3. Rehearse the data migration. Two full dry runs minimum — masters, open balances, open documents — against a staging instance, reconciled by finance, before anyone schedules a cutover weekend. The data migration guide is the checklist.
4. Parallel-run finance for one close. One month of dual books is cheap insurance and is usually what converts a skeptical CFO.
5. Contract for outcomes, not hours. Fixed-scope phases with named deliverables, an SLA for post-go-live support, and a defined escalation path. Partner quality is the single biggest variance in enterprise outcomes.

Frequently asked questions