Descriptions:
Bloomberg Tech sits down with Michelle Weaver, Morgan Stanley’s US thematic research strategist and executive director, to assess where the AI investment trade actually stands in mid-2026. Weaver draws on Morgan Stanley’s proprietary CIO survey data to argue that the market is still early on the S-curve but showing clear forward progress: 25% of S&P 500 companies can now quantify the ROI they are getting from AI adoption, up from 14% a year ago, and by year-end the majority of companies are expected to have at least one AI project live in production.
On the supply side, Weaver identifies labor as the single largest bottleneck to data center construction — a shortage of electricians and specialized tradespeople — followed by power, where she estimates a roughly 40-gigawatt shortfall in capacity needed through 2028. Even accounting for innovative solutions like Bitcoin site conversions and fuel cells, she sees a 10-20% residual gap. The NIMBY and anti-data center movement is flagged as a growing political wildcard heading into the midterms, with New York State already issuing a state-level moratorium.
Weaver also addresses the open versus closed model debate, noting that 63% of US enterprises currently run both in parallel, and that employee behavior change and embedded workflows will create strong stickiness for closed models even as open-weight alternatives improve. Geopolitical concerns around foreign open-weight models add another layer of hesitation for US enterprises considering a full switch.
📺 Source: Bloomberg Tech · Published August 12, 2026
🏷️ Format: Interview
