Deutsche Bank Research maps a ‘superintelligence by 2030’ scenario and tells investors to sort holdings into AI adopters and casualties

Your AI and this story

Key facts

• Author: Adrian Cox, Deutsche Bank Research; title as given by one summary: ‘AI boom or doom: what will superintelligence feel like and when will we get there?’
• Scenario (quoted by Yahoo Finance): ‘Under our scenario, by 2030 the best performing AI models would have the qualities of “superintelligence”’, defined as ‘better than human experts in all domains’, compressed enough to run on phones, tablets and laptops. Example: a factory retooled overnight by robots
• Constraints named (Stephen Innes’s summary): whether recursive self-improvement can overcome training time, energy, verification, capital and real-world implementation fast enough. MarketWatch’s headline: ‘Superintelligence may never arrive, Deutsche Bank says’; Futu’s: achievable by 2030 ‘with power supply and costs as key constraints’
• Three points for investors (Yahoo Finance): classify holdings as adopters or casualties instead of sizing by sector averages; do not read calm prices as low risk (Chegg lost half its value in a day in May 2023); be careful averaging down, since ‘a falling price may be information that a business model is obsolete rather than an opportunity’
• On diversification: ‘Diversification normally works because idiosyncratic risks are unrelated and average each other out’; with one dominant variable, spreading capital spreads exposure
• Where value goes as intelligence gets cheap (Innes’s summary): physical assets, proprietary data, trusted customer relationships, regulatory permissions and the infrastructure that turns compute into output
• The note itself was not read (client research); MarketWatch, Seeking Alpha and Futu pages were not readable