South Africa e-Invoicing: SARS proposes a 5-corner model, with implementation from 2030

What SARS published

On 17 August 2026 the South African Revenue Service published its Consultation Paper on VAT Modernisation: e-Invoicing, Interoperability Framework and e-Reporting. It runs to 27 pages and follows the VAT Modernisation Discussion Paper of 8 September 2023.

The paper names an operating model, allocates responsibilities corner by corner, and sets out a dated roadmap. The architecture itself was settled earlier, in definitions the Tax Administration Laws Amendment Act 4 of 2026 inserted into the VAT Act on 1 April 2026. What the paper adds is who occupies which role, what each owes the others, and over what timescale.

Written comments close on 16 October 2026 and are submitted through a Qualtrics survey rather than an email address, so responses are structured around SARS’s own questions. The 2023 Discussion Paper set no implementation dates, so the dates below appear for the first time in the August 2026 paper, and everything that follows is taken from it.

The proposed Digital VAT Model

SARS proposes three pillars, e-invoicing, an interoperability framework and e-reporting, combining into a Decentralised Continuous Transaction Control and Exchange model, abbreviated DCTCE throughout the paper. Invoices are validated and cleared by accredited service providers rather than by a government platform, and the tax-relevant data reaches SARS through an access point of its own.

A Network Authority governs the framework. It sets, maintains and enforces the technical and legal rules, accredits the access points, publishes the list that taxpayers choose from, and contracts with the providers, whose obligations a service level agreement regulates. The paper notes that the revenue authority then integrates with the service providers rather than with every vendor individually. The stated end point is pre-filled VAT returns and, over time, automatic assessment, which the taxpayer must confirm or edit to preserve the self-assessment principle. SARS already issues auto-assessments in personal income tax, where a taxpayer who disagrees files a corrected return.

Decentralised validation and centralised reporting

The paper uses the word clearance throughout, in a sense that differs from the Italian and Brazilian systems. SARS does not sit between the supplier and the buyer, and no invoice waits for state approval before reaching its recipient. Validation happens at the accredited provider, and an invoice that passes is transmitted onward. The paper puts forward the absence of a single point of failure as an advantage of the decentralised design: where many service providers are involved, the others continue to operate if one fails.

Decentralisation applies to validation rather than to the data. Corner five connects to every provider in the network and receives the same cleared invoice from both the supplier’s access point and the buyer’s, which the paper calls duplex clearance. Chapter 5 adds that the revenue authority may choose to receive only a tax-related subset of the complete message, once the providers have assured data quality.

The paper is direct about the use of that data. The executive summary states that AI and analytics tools will work in the background to analyse whole-of-value-chain data for discrepancies and risks. Chapter 5 adds that this will be combined with relevant third-party information to support AI-driven VAT assessment. The comparison of the two models describes automated processing across the supply chain, from order through goods receipting, invoicing and payment, matched with customs clearance data.

The buyer’s obligation to report VAT treatment

Corner four does not simply receive a cleared invoice. The paper requires the buyer to have the capability to receive, process and respond to e-invoices within its accounting system, including indicating the VAT status of the acquisition. The buyer sends a response to C3 confirming receipt and the accounting VAT treatment, for example fully, partially or not claimed, and C3 then transmits the invoice and that VAT status to C5.

France provides the nearest comparison in a live reform. It makes a small set of invoice lifecycle statuses mandatory, including refusal by the recipient, which must cite one of the reasons the standard permits. The South African proposal asks the buyer for the deduction position rather than for a delivery or refusal status.

For businesses that apply apportionment or partial exemption, that moves the deduction decision from a periodic return to a per-transaction response sent close to the point of receipt. The paper also envisages the original e-invoice as the supporting document that evidences the VAT liability on the supplier’s side and deductibility on the recipient’s end. It states that the technical specifications will address zero rating, deemed supplies and apportionment, without saying how.

What has not been decided

For a paper this specific about roles, several central questions remain open, and they are the ones a submission can still affect.

The standard. E-invoices must adhere to an e-invoice standard or specification, and the paper offers EN 16931 CIUS, the UN/CEFACT Cross-Industry Invoice and Peppol PINT BIS as examples rather than as a selection. It does not adopt Peppol, and it describes no central tax hub: the requirements are to be prescribed by regulations that have not been proposed yet.

The Network Authority. The paper treats it as a given, since C1 and C4 choose from its published list and C2, C3 and C5 all contract with it, but it never names it. Chapter 9 states that the IT decision will be informed by the preferred Network Authority, the interoperability framework, the associated standards and the outcome of the selection and procurement processes currently being finalised by SARS.

Thresholds, penalties and timing. Phase 5 is described as guided by mandates and turnover thresholds, and no threshold appears in the paper. Neither does any penalty. Near real-time is defined in the glossary as a short timeframe close to the point at which a transaction occurs or an invoice is issued, received or processed, "as will be defined in legislation and subject to public comment". Digital signature authentication of e-invoices appears once, inside the table comparing traditional and digital VAT administration, and nowhere in the narrative text.

Data protection. The paper states that legislative rules will address objections, data protection, certification and authorisation of service providers, secrecy and related governance requirements. It does not mention the Protection of Personal Information Act 4 of 2013, and it does not say how that Act will apply to the accredited providers that would hold transaction data as it passes between the trading parties and SARS.

The implementation roadmap

What falls in 2028/2029 is phase 3, validation and quality assurance testing in a controlled environment with voluntary participants, expected to run for approximately six months. Phase 4, the pilot with voluntary participants from priority segments, follows for approximately six months in 2029/2030. Only then does phase 5, phased implementation, commence during the 2030 calendar year, extending over approximately 36 months.

Within phase 5 the sequence is large B2B taxpayers first, government as a recipient second and possibly alongside them, MSMEs third and B2C last. Large taxpayers are expected to adopt voluntarily at first, with mandatory adoption introduced later, and voluntary participation by any sector or size of business will be available. SARS states that the order may change subject to ease of adoption, risk of compliance, VAT gap indications and other factors.

On the paper’s own figures, 36 months from 2030 runs to about 2033, with B2C last in the sequence. Each preceding phase carries a stage gate, shown in the roadmap figure as design approved, build complete, testing complete and pilot signed off. The same figure carries a disclaimer that the roadmap is indicative and subject to change, and that timelines, phases and activities may be revised as planning and implementation progress. The paper cites the VAT gap among the reasons for the programme but does not quantify it or explain how the timescale relates to it.

What applies today

The Tax Administration Laws Amendment Act 4 of 2026, promulgated in Government Gazette 54447 on 1 April 2026, inserts definitions of e-invoice, e-debit note, e-credit note, e-reporting and interoperability framework into the VAT Act, each subject to further requirements the Minister may prescribe by regulation. The definition of an interoperability framework is the use of a network of service providers where decentralised exchange of those documents occurs, and that can facilitate clearance and interoperability between supplier and recipient. Decentralised exchange is therefore already the statutory definition.

The Act also adds section 74(1B), which allows the Minister to make regulations prescribing the requirements for participation by a vendor in a voluntary e-reporting system. There is no power in the Act to compel structured e-invoicing or transactional reporting. Draft regulations are due in phase 1 and promulgation in phase 2, which places the first binding text in 2027/2028.

Electronic invoicing is meanwhile permitted and largely unregulated by SARS. The VAT 404 Guide for Vendors states at section 13.9 that vendors do not need prior approval from the Commissioner to implement e-invoicing, that electronic transmission and retention are regulated by the Electronic Communications and Transactions Act 25 of 2002, and that SARS is not in a position to rule or advise on whether an EDI system meets that Act’s technical specifications. Records are retained for at least five years. Two claims found in vendor material do not hold: EDI invoicing does not require authorisation from SARS or National Treasury, and no National Treasury instruction requires suppliers to submit digital invoices for public procurement, which arrives in phase 5b.

What changes for a finance function

Chapter 9 describes a shift from periodic, manual VAT return preparation to continuous transaction monitoring: overseeing automated invoice validations and addressing exceptions the system flags, rather than compiling data at month end. Internal controls may move earlier in the process, and the paper expects the finance role to focus more on data quality and exception handling than on routine tax calculation. The benefits SARS sets against that are reduced compliance effort, faster refunds and oversight concentrated on exceptions, anomalies and high-risk behaviour.

On systems, accounting and ERP software must be compatible with the new e-invoicing standard and connect to an accredited provider. SARS expects the change for large businesses to consist of upgrades rather than substantially new implementations, with legacy and new systems running in parallel across segments until the transition completes, and points smaller taxpayers towards service-provider portals and simplified digital tools. It states that it will engage software providers on tiered subscription models and consider subsidised or low-cost options, without giving eligibility, cost or timing. The paper contains no cost estimate, and says SARS will use the consultation to understand cost impacts.

What to do next

Chapter 5 of the consultation paper is the section to read closely. The corner four response has no equivalent in current South African practice, and it is the obligation that reaches furthest into existing accounts payable processes.

The parts still open are worth a submission. Transactions that carry no invoice are expressly deferred to the design process: the paper names deemed supplies, sector-specific supplies and deductions, so a sector that sets out its own cases now has the opportunity to shape the rules. Beyond that: the timeframe within which a buyer can realistically return a deduction position, given that near real-time is undefined; where liability sits when a provider rejects a valid invoice, which the paper raises as a topic without allocating; how POPIA obligations will apply to accredited providers; and which standard is selected, since the paper lists cross-border compatibility among the things the model must support.

There is no technical specification to test against. The process the paper sets out confirms the sequence: written submissions run for 60 days, consolidation and analysis of feedback covers weeks 1 to 14, and sector-specific engagements run from week 15 until policy adoption. The three deliverables that follow, a consultation findings report, draft regulations where applicable, and publication of the VAT digital model and technical specifications, are each dated "to be communicated". The work that holds regardless is mapping which transactions carry a tax invoice, where apportionment decisions are made and by whom, and how far existing systems are from producing structured data.