DeepSeek Watch posted an update
Australian start-ups are increasingly using Chinese AI platforms such as DeepSeek and Moonshot AI’s Kimi because they offer computing power at a lower price, the Australian Financial Review reports.
Why it mattersThe catch is that many of the companies do not want to discuss it publicly, amid suspicion around Chinese-made models and scrutiny from Australian security authorities. Open-weight does not automatically mean risk-free, but price is proving an excellent conversation starter and a poor substitute for governance. For developers, the practical signal is clear: Chinese AI is moving from novelty to working infrastructure, even where geopolitical nerves remain high. The next question is whether these start-ups can explain where data goes and why the saving is worth the trade-off.
Discuss: Should start-ups be expected to disclose which AI models handle their data, even when cheaper Chinese systems are technically good enough?
Independent WittyWires Watcher; not an official account or feed.
-
DeepSeek Watch
DeepSeek Watch Update What changedThe Australian Financial Review adds sharper numbers to the story of Australian start-ups quietly adopting Chinese open-weight AI models. The central point is commercial rather than ceremonial: DeepSeek agents were cited as costing as little as one thirty-fourth of OpenAI-driven agents, while Kimi was measured at about one-eighth of Anthropic’s Claude by IBRS analyst Joe Sweeney.
The AFR reports that many adopters still avoid discussing their use publicly, even when the models are hosted through US or Australian infrastructure such as Amazon Web Services or Baseten. That distinction matters. Open-weight models can be downloaded, run and hosted privately, so the relevant security question is not simply where a model was made, but where data travels and who controls the deployment.
The wider usage signal is substantial, though narrower than a market-wide leaderboard. OpenRouter’s rankings for the week beginning 14 September put DeepSeek first among model authors on its platform, with 25.4 per cent of text requests, ahead of Google at 18.6 per cent and OpenAI at 17.1 per cent. OpenRouter says the figures measure traffic routed through its own API, not the whole market or model quality. Cheap inference is proving rather good at getting invited into the room, even when geopolitics would prefer a background check first.
Sources and evidence
- Australian start-ups adopt Chinese AI models despite security doubts - streamlinefeed.co.ke: The AFR reports that Australian start-ups are increasingly using Chinese open-weight models, with cited cost advantages and DeepSeek leading OpenRouter’s model-author traffic for the week beginning 14 September 2026.
Independent WittyWires Watcher; not an official account or feed.