Amazon Web Services reserves its flagship South African small-business accelerator for firms that are 100% black-owned. StephanZA argues that the rule exposes a deeper conflict between racial redress, equal treatment and the demand for measurable economic results. One company, two claims The dispute begins with two statements that can both be true. AWS says its Equity Equivalent Investment Programme is open only to qualifying small and medium enterprises that are 100% black-owned. It also says any South African business can work with AWS through its wider commercial and partner ecosystem. That distinction is central. Solidarity has announced an investigation into the programme, arguing that its ownership threshold excludes a business as soon as a white person holds a share. AWS answers that the accelerator is one compliance mechanism among several, not a racial barrier around the company’s entire South African operation. The unresolved question is whether a selective development programme can be defended as remedial policy when eligibility is drawn at an absolute racial line. The programme is substantial rather than symbolic. AWS said it launched the initiative in 2019 with a plan to invest R365 million through December 2026 in black-owned ICT and digital businesses. Support includes training, certification, technical assistance, operating subsidies, cloud credits and help entering the AWS Partner Network. AWS told The Citizen this week that 47 black-owned firms had been accelerated since launch, while separate programmes such as AWS Activate and AI for Good remained open to all. What the law requires Equity-equivalent programmes exist for multinationals whose global policies prevent them from selling local equity. Instead of transferring shares, an approved company can make recognised investments in enterprise development, skills, research or other forms of economic inclusion. The contribution then counts towards the ownership element of its B-BBEE scorecard. This is where both the corporate defence and the criticism need precision. The Department of Trade, Industry and Competition says the B-BBEE Act does not place a general legal duty on private companies to comply. It does bind organs of state when they make decisions about procurement, licences, incentives and partnerships, creating powerful commercial pressure throughout supply chains. Nor does the law simply require every company to be majority black-owned. It provides a weighted system in which ownership is one element, with equity equivalents available to qualifying multinationals. AWS’s 100% threshold is therefore a condition of this particular programme, operating inside an approved transformation framework. It is not a universal statutory threshold for every business in the country. That does not dispose of Solidarity’s objection. It defines it more accurately: the argument is over whether AWS chose or needed an absolute ownership rule for access to a valuable development programme, and what public benefit that rule has produced. From eligibility to evidence StephanZA’s strongest challenge is not merely that the programme classifies applicants by race. He asks what South Africans receive in return for the money, preferential access and institutional effort directed through transformation policy. Announcing expenditure and counting participants are useful, but they are not the same as publishing survival rates, revenue growth, jobs created, certifications earned or contracts won after support ends. The debate is no longer only about Amazon. It turns on whether transformation policy can be judged by eligibility rules alone, or by what happens after money, training and market access are allocated. That diagnosis deserves scrutiny because the figures in the recording do not match the official TIMSS data. In 2019, Grade 9 learners at no-fee schools averaged 365 in mathematics and 335 in science, not 277. The score of 500 is the international scale centrepoint, not a minimum mark for university admission. The verified results remain deeply troubling: 72% of learners at no-fee schools lacked basic Grade 9 mathematics knowledge, and 78% lacked basic science knowledge. The same evidence also complicates a story of simple decline. Mathematics performance in no-fee schools rose by 41 points between 2011 and 2019, while the share reaching basic knowledge increased from 12% to 28%. Progress and failure coexist. South Africa’s schooling system has improved from a very low base while still leaving most learners in poorer schools without the foundations that an inclusive digital economy requires. The test Amazon cannot avoid B-BBEE grew from a legitimate historical problem: apartheid deliberately blocked black South Africans from ownership, skilled work, land and commercial opportunity. A development programme aimed at black-owned technology firms therefore has a coherent remedial purpose. The strongest defence of AWS is that a general programme would not neces