The Good: Generation Zero Alcohol

Generation Z, born between 1997 and 2012, is making a significant impact on drinking culture. Unlike previous generations, Gen Z tends to consume alcohol more moderately, drinking 20% less than millennials did at the same age. This shift begs the question – is it a passing trend, or does it signal a deeper cultural change that’s reshaping our approach to alcohol consumption?

Research shows that while preceding generations may have turned to alcohol to cope with stress and anxiety, Gen Z is more likely to reserve drinking for special occasions or skip it altogether. A 2024 Forbes study revealed that around 30% of Gen Z respondents drank less than the year before, and 13% had quit alcohol entirely. A key driver behind this shift is health consciousness, with 74% of participants citing a desire for a healthier lifestyle as their reason for cutting back. Social media has further fuelled this trend, with influencers and other online figures promoting non-alcoholic alternatives and even alcohol-free lifestyles.

Despite shifting preferences, 60% of this age group still frequents bars and restaurants weekly, with data showing that younger consumers are simply 6% more likely to opt for a soft drink. Alcoholic beverage giants within our holdings, AB InBev and Heineken, have recognised this trend and responded by introducing non-alcoholic options and ‘healthier’ alternatives to traditional beers and spirits. These new products have seen a surge in sales, partly driven by Gen Z’s growing interest, and are poised to become important growth drivers for these companies.

The Bad: GNU Handling Errors

The 2024 South African national election sparked a wave of hope and optimism, as many envisioned a fresh chapter in governance. Expectations of collaboration between political parties inspired a vision of unity and teamwork reminiscent of South Africa’s national rugby team, the Springboks. This optimism gained momentum, with signs of progress materialising—the Rand rallied against major currencies, buoyed by an improving economic outlook that lifted the national mood.

Recent weeks have brought challenges for South Africa’s leadership, highlighting the importance of collaboration and open communication in governance. However, the Government of National Unity (GNU) seems to be navigating some rocky waters. The majority party’s focus on advancing significant legislation, including the BELA Bill, the Expropriation Bill, and the National Health Insurance (NHI) Bill, has sparked debate over whether the promised spirit of collaboration within the GNU is being upheld.

While these bills can be argued about ad nauseam, they do highlight the fragile nature of political alliances and the potential ramifications for the average South African. History has demonstrated that negative sentiment, coupled with governance missteps like those experienced during the State Capture era, can significantly undermine GDP growth prospects and stock market performance. Compounding this, the Rand has faced a challenging trajectory, depreciating by an average of 5.6% per annum since 1995 – affecting the real wealth of South Africans.

Despite the US accounting for about 65% of the global equity market and South Africa representing less than 0.5%, it’s notable that more than 80% of South African equity allocations remain concentrated domestically—despite the relaxation of exchange controls that were in place before the 1990s. Beyond the stark difference in market size, the performance gap is equally striking, with the S&P 500 outperforming the JSE by 4.5% per annum in Rand terms since 1995. This focus on domestic markets limits individuals’ exposure to the broader global markets that are driving economic growth.

We understand the difficulties South Africans face in protecting their wealth amid these challenges. This is why we remain steadfast in our Rand-hedge mandate, offering a product such as the High Street Balanced Prescient Fund designed to navigate these complexities. Only time will tell if current trends persist or if the GNU can draw inspiration from Siya Kolisi and Rassie Erasmus, leading South Africa toward a story of success.

& DeepSeek and You Shall Find

The fast-paced world of AI experienced a major disruption in January, when Chinese research firm DeepSeek introduced a novel approach to AI model training that prioritised efficiency over raw computing power. To date, US dominance in AI has largely been defined by huge investments, access to leading chip technologies, and a scaling strategy that emphasised bigger models and more data. However, DeepSeek’s approach emphatically challenged that paradigm, indicating that smaller, more efficiently trained models can still perform at a high level. Their “R1” model scored well in benchmark tests despite allegedly being trained with a budget of only $6m — a fraction of what US counterparts require. DeepSeek were able to do this by leveraging key innovations like quantisation, multi-token prediction, and Mixture-of-Experts (MoE) architectures, which allowed for substantial efficiency gains. This triggered a sharp reaction in global markets, with the Nasdaq Composite dropping 4% and chip giant Nvidia plunging 16% in response.

The market’s strong reaction was fuelled by fear that DeepSeek’s efficiency breakthroughs could have severe implications for the broader AI hardware sector. However, in the days following the announcement, several leading technology figures have been quick to highlight Jevons paradox — the economic principle that greater efficiency in a resource usually leads to higher overall consumption. In this instance, cheaper AI models are likely to expand AI adoption, ultimately increasing demand for computing hardware rather than reducing it. As the market started to absorb this new narrative, Nvidia and other US tech names rebounded sharply. Ironically, the very advancements that sparked these concerns may have been driven by US export controls, designed to stifle China’s AI efforts. As DeepSeek CEO Liang Wenfeng noted, “the problem we are facing has never been funding, but the export control on advanced chips.” However, rather than slowing technological development, these restrictions may have actually pushed Chinese engineers to develop more resource-efficient techniques.

Outside of the geopolitical tech jostling, it is becoming increasingly clear that AI models are starting to get more commoditised. Much like smartphones, their real value lies not in the models themselves, but rather the applications built on top of them. Despite breakthroughs in efficiency, data centres remain a crucial part of the infrastructure that will power these applications. Companies like Meta, Amazon and Microsoft continue to ramp up AI-focused capital expenditures, committing $65bn, $75bn and $80bn respectively in recent weeks. Government interest is also starting to build, with President Trump announcing a landmark “Stargate” venture between OpenAI, SoftBank and Oracle that would invest $500bn in AI infrastructure across the US. While DeepSeek’s innovations mark a shift in AI development strategy, they don’t diminish the importance of compute power. The future of AI will be defined not just by more efficient models but by the transformative applications they enable.