This decision puts South Africa firmly into the dominant UNCITRAL Model Law tradition that prioritises finality and procedural discipline. At the same time, the split amongst the judges exposes a deeper and unresolved tension between arbitration finality and public law doctrines of fraud, legality and sovereign authority that are particularly acute in disputes involving state contracts.

The facts

In September 2018, Minister Temeki Tšolo signed an agreement with Frazer Solar GmbH (FSG). FSG, a German company, was to supply renewable energy products worth up to €100 million. The agreement provided for arbitration in Johannesburg under the Model Law (South Africa has domesticated the UNCITRAL Model Law through Schedule 1 to the International Arbitration Act). Cabinet approval, required under Lesotho law, was never obtained. No products were ever delivered. Disputes arose and FSG commenced arbitration. Lesotho did not participate, alleging that notices had been intercepted by senior officials as part of a fraud. In Lesotho’s absence the arbitrator awarded FSG approximately €50 million. Following the award, the Johannesburg High Court made the award an order of court, that application also being unopposed, and writs of execution were issued.

Lesotho applied to the High Court to rescind the enforcement order on the basis that the order had been granted in error and to set aside the arbitration award. Both applications were dismissed, the first on the basis that Lesotho was aware of the application and chose not to oppose it, and the second on the basis that it was brought outside the three-month time bar, that time limitation being a justifiable limitation upon the fundamental right to access to courts. On appeal, the SCA granted rescission of the enforcement order but refused to set aside the award, again because the challenge was out of time.

The significance of the judgment

The majority (six judges on two separate reasonings) confirmed that Article 34(3) of the Model Law, as incorporated by the International Arbitration Act, imposes a peremptory three-month time limitation for setting-aside applications. There is no general power of condonation, and the only exception is for fraud or corruption, where time runs from when the fraud could reasonably have been discovered. Reviewing the position in Singapore, New Zealand, Canada, Australia, India, Zimbabwe and Kenya, the court found South Africa’s approach aligned with international consensus. The court also held that the time limitation constituted a reasonable limitation under section 36 of the Constitution.

This result gives award creditors welcome certainty and by upholding the Model Law, reinforces South Africa’s standing as an arbitration-friendly seat.

Comparative position: Article 34(3) and the dominant Model Law tradition

Comparative Model Law jurisprudence confirms that Article 34(3) is treated as imposing a strict and non-extendable time limitation, even where allegations of fraud, corruption or illegality arise. Courts in leading jurisdictions, including Singapore and Hong Kong, have consistently held that these grounds must be advanced within the express framework for setting-aside procedures and do not independently suspend the limitation period.

The principal safety valve lies not in extending the time bar, but in the distinction between active challenge and defensive resistance. While a setting-aside application under Article 34 has a rigid three-month time limit, parties may still raise issues of public policy or invalidity at the enforcement stage under Article 36, albeit within a narrower and more constrained framework. The SCA’s judgment reflects and reinforces this two-track structure.

The three-way split in the court

Each reasoning in the separate judgments reflects a distinct jurisprudential approach to the relationship between arbitration finality and substantive justice.

The majority imposed a strict three-month time limitation with no general power of condonation. They upheld arbitration integrity and predictability as paramount values. They also recognised fraud only within the narrow statutory exception. The time limitation period would then run from when the fraud could reasonably have been discovered.

Modiba AJA emphasised the consequences of procedural default, finding that Lesotho’s failure to participate in the arbitration was an election it made which was binding. He found that it was then precluded from relitigating the matter through a belated collateral attack on the award.

Molemela P (with Makgoka JA) adopted the expansive view that fraud vitiates the underlying contract and with it, the arbitration clause. On this reasoning, constitutional legality and state authority cannot be overridden by procedural time limits, and Article 34 should be read more flexibly where the very foundation of the arbitration is challenged.

The divergence between the majority and the dissenting judgments reveals a fundamental doctrinal divide: whether arbitration finality should prevail over, or yield to, allegations of fraud, illegality and lack of state authority.

The enforcement impasse

The SCA’s order does create an unusual situation. FSG holds a valid award (the award having been upheld) that it cannot currently enforce in South Africa (the enforcement order having been rescinded). Lesotho, meanwhile, faces an award it cannot challenge under Article 34 due to the application of the time limitation. But this impasse is probably temporary as there are options.

First, FSG may relaunch the enforcement proceedings. Presumably, Lesotho would this time oppose under Article 36 of the Model Law, including on factual grounds of incapacity, invalid agreement or lack of proper notice, or on public policy grounds. If that happens, the enforcement court will decide whether these defences apply.

Second, and because the award has not been set aside at the arbitral seat in South Africa, FSG may choose to enforce the arbitration award outside South Africa. It is entitled to do so in any state which is a signatory to the New York Convention¹ and where Lesotho holds assets.

Third, the three-way split in the SCA judgment, already discussed, makes a Constitutional Court application a realistic possibility. Lesotho may approach that court to argue that the “fraud unravels everything” principle requires the court to reconsider the argument that the fraud should unravel the whole matter and that the arbitration award should be set aside. The outcome will depend on whether the Constitutional Court favours procedural finality or adopts the third judgment’s view that fraud cannot be overridden by time bars. Any such application to the Constitutional Court would not automatically suspend the SCA’s order, so FSG may still apply to court to enforce the award in South Africa in the meantime.

Implications for state contracts and infrastructure transactions

The case underscores the heightened risk associated with arbitration clauses embedded in public or sovereign contracts. Even where the underlying agreement is alleged to be unlawful or ultra vires, (whether due to a failure to obtain cabinet approval, non-compliance with procurement legislation, or a signatory’s lack of authority) a failure to challenge the arbitral award within the prescribed time limits may prevent any recourse to a setting aside of the award under Article 34.

For investors and contractors, this highlights the critical importance of verifying authority and compliance at the contracting stage. A counterpart’s apparent authority, including ministerial signature, is not enough. Underlying constitutional and statutory requirements must be independently assessed. For states and state-owned entities, there must be robust internal controls to ensure timely participation in arbitration and related court proceedings, and to prevent the kind of procedural default that fatally upset Lesotho’s challenge.

Sovereign immunity

The judgment also confirms that a foreign state that has agreed to arbitration is not immune from South African court proceedings relating to that arbitration (section 10(1) of the Foreign States Immunities Act 87 of 1981). It also clarifies that a contract signed by a minister creates only a rebuttable presumption that it is a state contract. The minister’s authority may still be challenged.

Conclusion

The majority judgment reinforces South Africa’s alignment with Model Law jurisdictions that prioritise finality and minimal court intervention. This enhances South Africa’s credibility as a predictable and commercially reliable arbitration seat. But the strong dissenting view highlights potential areas of uncertainty, particularly in disputes involving sovereign counterparties and allegations of illegality. Parties selecting South Africa as a seat should be alive to both dimensions: the robustness of the finality principle, and the as-yet-unresolved questions that the dissent leaves open.

Frazer Solar is a watershed case for procedural finality in international arbitration in South Africa. Parties must act within the strict time limits imposed by the Model Law or risk losing the ability to challenge an award, irrespective of the merits. Award creditors have greater certainty that international arbitral awards will not face belated challenges, reinforcing South Africa’s credentials as a reliable seat.

But the 4:1:2 split in the judgment leaves open the critical question of the extent to which fraud and public law illegality may justify departure from strict arbitral finality. The division and the third judgment’s distinction between formation and validity challenges suggests a route that the Constitutional Court may be asked to consider. If leave is granted to the Constitutional Court, the outcome could reshape the fraud exception in South African international arbitration law in a way that resonates well beyond this dispute.

For practitioners, the implications are both clear and immediate.

• Award debtors must act within the three-month statutory window.
• Award creditors should anticipate further proceedings and delay victory celebrations until three months and a day have passed.
• Parties contracting with sovereigns must do rigorous due diligence on authority and approvals.

For now, the law prioritises finality, but the final word may rest with the Constitutional Court.