This semi-monthly column highlights news, government documents, NGO/IGO papers, conferences, industry white papers and reports, academic papers and speeches, and central bank actions on the subject of AI’s fast paced impact on the banking and finance sectors. The chronological links provided are to the primary sources, and as available, indicate links to alternate free versions.
NEWS:
Financial Times (FT.com)
How the AI ‘bubble’ compares to history (no paywall). December 31, 2025. US stock valuations are higher than before 1929 Wall Street crash but the dominance of a single sector has precedents. Surging artificial intelligence stocks have driven the US market to record highs this year, drawing comparisons on some metrics to infamous periods of investor mania in the past. The huge gains of 2025 — in which Nvidia’s market value has more than doubled from its April lows, making it briefly the world’s first $5tn company — have prompted warnings from central bankers and some investors that the AI sector could be in a bubble and that a stock market correction could pose a threat to financial stability. The US blue-chip S&P 500 index is now more expensive on a cyclically-adjusted 10-year price/earnings ratio — a commonly used valuation metric — than it was before the 1929 Wall Street crash and well above where it was on the eve of the 2008 global financial crisis, according to data group Finaeon. In data going back to the 1840s, the only time valuations have been more stretched has been during the dotcom bubble in 1999.
AI boom adds $500bn to net worth of US tech billionaires in 2025 (no paywall). December 31, 2025. America’s wealthiest tech billionaires added more than $550bn to their combined net worth this year, as they benefit from an investor frenzy around leading artificial intelligence companies. The top 10 US tech founders and chief executives possessed nearly $2.5tn in cash, equity and other investments at the close of trading in New York on Christmas Eve, according to Bloomberg data. The figure is up from $1.9tn at the beginning of this year and comes as the S&P 500 climbed more than 18 per cent. Silicon Valley’s leaders have profited from the hundreds of billions of dollars spent globally on AI chips, data centres and products, even if some of their gains were trimmed in recent months over concerns about an AI-fuelled investment bubble. “This is all speculative and correlated to the success of AI,” said Jason Furman, an economics professor at Harvard University and consultant for start-up OpenAI. “There’s a huge question mark over whether this is all going to pay off, but investors are betting that it will.”
Tech groups shift $120bn of AI data centre debt off balance sheets. December 24, 2025. Creative financing helps insulate Big Tech while binding Wall Street to a future boom or bust. Tech companies have moved more than $120bn of data centre spending off their balance sheets using special purpose vehicles funded by Wall Street investors, adding to concerns about the financial risks of their huge bet on artificial intelligence. Meta, Elon Musk’s xAI, Oracle and data centre operator CoreWeave have led the way on complex financing deals to shield their companies from the large borrowing needed to build AI data centres. Financial institutions including Pimco, BlackRock, Apollo, Blue Owl Capital and US banks such as JPMorgan have supplied at least $120bn in debt and equity for these tech groups’ computing infrastructure, according to a Financial Times analysis. That money is channelled through special purpose holding companies known as SPVs. The rush of financings, which do not show up on the tech groups’ balance sheets, may be obscuring the risks that they are running — and who will be on the hook if AI demand disappoints. SPV structures also increase the danger that financial stress for AI operators in the future could cascade across Wall Street in unpredictable ways.
CNBC: December 30, 2025.
- SoftBank has completed its $40 billion investment commitment to OpenAI, sources told CNBC’s David Faber.
- CNBC reported in February that the Japanese firm was finalizing a $40 billion investment in the ChatGPT maker at a $260 billion pre-money valuation.
- The rise of artificial intelligence applications has created a rush to invest in more data centers and connectivity solutions to support booming demand.
SoftBank Group has completed its $40 billion investment in OpenAI, a source familiar with the matter said on Tuesday, marking one of the largest private funding rounds ever and deepening founder Masayoshi Son’s bet on AI. SoftBank has been building one of the largest private technology investment programs in the world, with a particular focus on artificial intelligence and related infrastructure such as data centers. Visa says it has completed hundreds of secure, AI-initiated transactions with partners, arguing this proves agent driven shopping is ready to move beyond experiments. The company believes 2025 will be the last full year most consumers manually check out, with AI agents handling purchases at scale by the 2026 holiday season. Nearly half of US shoppers already use AI tools for product discovery, and Visa wants to extend that shift all the way through payment using its Intelligent Commerce framework. The pilots are already live in controlled environments, powering consumer and business purchases through AI agents tied to Visa’s payment rails. To prevent abuse, Visa and partners have introduced a Trusted Agent Protocol to help merchants distinguish legitimate AI agents from bots, with Akamai adding fraud and identity controls. While the infrastructure may be ready, the bigger question is whether consumers fully understand the risks of letting software spend their money.
PAPERS:
NBER
The Emerging Market for Intelligence: Pricing, Supply, and Demand for LLMs. Mert Demirer, Andrey Fradkin, Nadav Tadelis & Sida Peng. Working Paper 34608. DOI 10.3386/w34608
Issue Date December 2025
We document six facts about the structure and dynamics of the LLM market using API usage data from OpenRouter and Microsoft Azure. First, we show rapid growth in the number of models, creators, and inference providers, driven by open-source entrants. Second, we show price declines and persistent price heterogeneity across and within intelligence tiers, with open-source models being 90% cheaper than comparable closed-source models of the same intelligence. Third, we document market dynamism, with frequent turnover among leading models and creators. Fourth, we present evidence of horizontal and vertical differentiation, with no single model dominating across use cases, and demand for intelligence varying widely across applications. Fifth, we estimate preliminary short-run price elasticities just above one, suggesting limited scope for Jevons-Paradox effects. Finally, we show that although the share of firms that use multiple models increased over time, most firms concentrate their use on a single model, consistent with experimentation rather than persistent reliance on multiple models.
Bank for International Settlement (BIS)
Artificial intelligence and growth in advanced and emerging economies: short-run impact. BIS Working Papers | No 1321 | 19 December 2025. We study how generative artificial intelligence (AI) affects short-run growth across countries. Our analysis covers 56 economies and 16 industries. We combine two elements. First, industries differ in how much they rely on cognitive and knowledge-based tasks, and so differ in their exposure to AI. Second, countries differ in their readiness to adopt new technology, based on their digital infrastructure, human capital, innovation capacity and regulatory frameworks. We measure industry exposure using data from the United States and country readiness using the International Monetary Fund’s AI preparedness index. We then link these measures to the change in real value added in each country-industry pair between 2022 and 2023.
Generative economic modeling. BIS Working Papers | No 1312 | 02 December 2025. Advances in artificial intelligence (AI) offer significant opportunities for economic analysis by pushing the boundaries of modelling capabilities. In particular, deep learning has emerged as a powerful tool for addressing dynamic economic models that were previously deemed intractable. However, effectively applying deep learning in practice often requires meticulous adjustments tailored to the model in question, which can present considerable challenges. In contrast, traditional solution methods are specifically designed and optimised for certain types of economic models but struggle with high-dimensional models. To bridge this gap, we propose a novel approach that combines the strengths of both approaches.
GOVERNMENT DOCUMENTS:
UK Office for Budget Responsibility, Briefing paper No.9. Forecasting productivity. November 2025…While the outlook for productivity growth is one of our most important variables in terms of its impact on the public finances, it is also one of the most uncertain. To illustrate the uncertainty around this central forecast, we present an upside scenario where more of the recent weakness in productivity growth was due to temporary shocks, the fading of which alongside a larger boost from artificial intelligence (AI) pushes potential productivity growth up to 1.5 per cent. We also present a downside scenario where potential productivity growth stays around its post-financial crisis average of 0.5 per cent over the forecast period …We also expect artificial intelligence to begin having a positive effect on productivity growth within the forecast period. There is significant uncertainty around both the size and timing of this effect – our central estimate is that it will build over time as adoption grows to reach an estimated 0.2 percentage points by our forecast horizon…
OECD
OECD.AI – Macroeconomic productivity gains from Artificial Intelligence in G7 economies. June 30, 2025. The paper studies the expected macroeconomic productivity gains from Artificial Intelligence (AI) over a 10-year horizon in G7 economies. It builds on our previous work that introduced a micro-to-macro framework by combining existing estimates of micro-level performance gains with evidence on the exposure of activities to AI and likely future adoption rates. This paper refines and extends the estimates from the United States to other G7 economies, in particular by harmonising current adoption rate measures among firms and updating future adoption path estimates. Across the three scenarios considered, the estimated range for annual aggregate labour productivity growth due to AI range between 0.4-1.3 percentage points in countries with high AI exposure – due to stronger specialisation in highly AI-exposed knowledge intensive services such as finance and ICT services – and more widespread adoption (e.g. United States and United Kingdom). In contrast, projected gains in several other G7 economies are up to 50% smaller, reflecting differences in sectoral composition and assumptions about the relative pace of AI adoption.