OP-ED - The space race is a vanity dash. The vault is the prize.
A Chinese open-source model wiped a percentage point off the S&P 500. Three weeks earlier, the world's largest hedge fund beat every frontier model it tested with a fine-tuned Alibaba base and its own expert-labelled data. African banks and telcos should take notes.
Last Friday, a Chinese startup shaved a percentage point off the S&P 500 with a model launch. Moonshot AI unveiled Kimi K3, an open-source model the company says closes much of the gap with OpenAI's ChatGPT and Anthropic's Claude. On the news, the Nasdaq dropped 1.4% and Taiwan's benchmark index closed down more than 6%.
You might recall DeepSeek triggering similar market ripples in January 2025, although US markets recovered quickly enough. At this point, it’s evident that signs are pointing to frontier-grade LLM prowess commoditising, with Kimi K3 serving as yet another harbinger for enterprise incumbents. That begs the question, if the smartest model in the room may soon be free, is the LLM space race to global dominance any more than a vanity dash to nowhereville?
Andile Masuku
Three weeks before last Friday's Kimi K3-induced excitement, the AI research unit at the world's largest hedge fund, Bridgewater, working with Mira Murati's Thinking Machines Lab, published results showing that a fine-tuned version of Qwen3-235B, Alibaba’s open-weights base model, outperformed every frontier model it tested on six financial judgment tasks drawn from investors' daily work.
Astonishingly, the custom model averaged 84.7% accuracy against 78.2% for the best frontier system, at roughly one-fourteenth the running cost. It’s worth flagging that these measures were gauged by Bridgewater's own internal evaluation, rather than an independent benchmark.
Nevertheless, this places the critical role of proprietary data and insights, exercised with turnkey judgment, in the spotlight. In specialist domains those elements can rarely be articulated into a prompt. Those have to be trained in from examples labelled by your in-house experts, drawn from data exclusively held by or singularly handled within the organisation. When that’s done, even a relatively modest model, finessing sovereign proprietary data, is capable of beating the giants at a fraction of the cost.
This got me thinking of banks and bankers. During a live fireside chat at VivaTech in Paris last month, The Atlantic's CEO Nicholas Thompson pressed Inrupt CEO John Bruce and his co-founder Sir Tim Berners-Lee (yep, the web's inventor, in town collecting a Visionary Award) on what AI does to banking. Inrupt works with banks and other financial institutions to implement decentralised data infrastructure.
Thompson drew out Bruce’s assessment that the bank's nightmare is not a rival chatbot but the customer uploading their statements into ChatGPT to shop for a loan, taking the data relationship, and the power that comes with holding it, out the door with them. Inrupt's answer is Charlie, a personal AI agent that sits between you and the model providers and decides what data leaves your device.
The banks as David to the LLM Goliath… Who would have thunk it? Meanwhile, African banks are fighting a second front as well, fending off a mobile telco industry that’s long since stopped being content with monetising phone calls and text messages. I’d say it’s a tricky time to be a banking CEO.
Andile Masuku
The thing is, though, incumbents on both sides of that tug-of-war are tempted to lean on what they perceive as birth-right exploitation privileges. There’s an instinctive manoeuvring around banking licences, spectrum, regulatory moats, distribution muscle, driven by the assumption that legacy clout and scale will win out by default. However, the Bridgewater result points at that being a commitment worth revisiting for C-suites.
It’s growing apparent that the prize sitting inside an African bank or telco is decades of transaction data nobody else on earth holds. Think repayment histories, agent-network float patterns and the seasonal cash-flow rhythms of informal traders. Organising the judgment of their best credit officers into labelled training data is slow, unglamorous, intrapreneurial work, with painfully little upside in the short term, accompanied by the awkwardness of empowering insiders to build the thing that might cannibalise their own fee lines before an outsider does. In other words, it’s the type of work done in service of a leadership agenda based on placing long-termist ecosystem beneficiation over short-term profit-taking.
When I profiled AI-native data and analytics platform Papermap in January, co-founder Benedict Quartey argued that AI would eventually become a commodity, and that once that happens, "the only thing that truly has value is what you do with it." With the Kimi K3 and Bridgewater revelations landing within a fortnight of each other, it does seem like that sensibility coming to fulfilment.
It remains to be seem whether Africa's incumbents have the bandwidth, patience and will to do the unglamorous work that leans into this emerging trend. The finesse layer appears on track to get built either way.
Editorial Note: A version of this opinion editorial was first published by Business Report on 21 July 2026.