Research Publication – Artificial Intelligence – Economics, Causality, Ecosystem & China

28th September 2026: RFM updates its in-depth coverage of Artificial Intelligence. RFM subscribers will receive their copy by email.

• The economics of the data centre are coming into sharp focus as shortages have caused pricing to rise to unsustainable levels, potentially causing demand weakness.
• Although Anthropic has had a stellar 2026, OpenAI is sneaking into pole position for 2027, which could delay Anthropic’s IPO further.
• The models are getting much better, creating great economic opportunity, but they still have no grasp of causality, meaning that superintelligence remains a pipe dream.
• The growing opportunity is triggering fierce competition for the attention of the enterprise and the consumer, which will be won by the few players that are able to create a thriving AI ecosystem.
• Highly capable open-source models can now run on relatively cheap hardware, but this intelligence is already a commodity.
• Even with only 20% of the tokens, RFM thinks that the frontier can still capture 80% of the value for as long as it can stay meaningfully ahead.
• China now dominates open source, but its actions are contradictory to the Chinese state’s stated aims. RFM thinks that China intends to centralise open-source AI in China and gain control of it, as it did in solar panels.
• The problem is that its silicon remains far behind, and the gap is getting wider, leaving China dependent on importing compute from overseas. This is an obvious target for US export restrictions.