Goldman's AI Rollout Follows Rivals

Hey, it's Matt. I break down what’s happening in AI + Wall Street with expert interviews, curated news, and in-depth analysis.

Was this email forwarded to you? Sign up below:

AI NEWS

Goldman Launches AI Assistant for Employees

Goldman Sachs is deploying an AI assistant to 10,000 employees for tasks like email and code work. The system aims to function like a seasoned employee while augmenting rather than replacing human workers. The bank joins JPMorgan and Morgan Stanley in releasing ChatGPT-like tools to staff. (CNBC)

• Also: Goldman Sachs is using AI to draft IPO documents in minutes instead of weeks, with 95% done by AI.

← I’ve highlighted in the past that Wall Street’s AI adoption is a lot slower than people think. But I think that’s because timelines are so aggressive. It’s only been slightly more than two years since the release of ChatGPT and three of the top investment banks in the world are using the technology already.

• Citi released a new study on AI in finance, which I discuss in more detail below. It’s worth highlighting this chart showing the extent of LLM usage in financial services, based on Microsoft cloud data.

  

Trump Announces $100 Billion AI Initiative

OpenAI, Oracle and SoftBank formed a new joint venture called Stargate to invest at least $100 billion in computing infrastructure to power AI, building on major U.S. investments in the technology. (NYT) $

• What is the ‘Stargate’ AI project? (Fortune/Yahoo)

DATA

Trading Firm XTX Bets €1 Billion on Its Own AI Hub

Trading firm XTX Markets is investing over €1 billion to build five data centers in Finland, breaking from the industry norm of using third-party computing infrastructure.

The London-based firm, which trades over $250 billion in assets daily, says building its own AI computing hub will be more cost-effective and help guarantee the computing capacity needed for its machine learning-driven trading strategies.

← A decade ago, prop trading was all about speed. There were plenty of stories about firms putting their offices across the street from exchanges. I have more reporting to do on this but it seems like computation power is the next frontier.

• I have a lot of sophisticated readers so if you have a view on this dynamic, please reach out: [email protected]

Here’s some background on XTX and its stockpile of AI chips I highlighted in October:

Man Group Builds Tool to Process Billions of Stock Trades

The need to analyze an overwhelming influx of stock data—and to do it fast—pushed the hedge fund to become its own kind of tech company. The tool, ArcticDB, is now being commercialized and has attracted Bloomberg as a customer. (WSJ) $

← It’s surprising to me that hedge fund had to build this in-house. I thought we were in the era of big data.

REGULATION

Trump Revokes Biden Executive Order on AI Risks

Biden's order required developers of AI systems that pose risks to U.S. national security, the economy, public health or safety to share the results of safety tests with the U.S. government, in line with the Defense Production Act, before they were released to the public. (Reuters, Bloomberg) $

Some states have passed AI regulation. In March, Tennessee enacted protections against AI voice cloning, and in May, Colorado created a tiered system for AI deployment oversight. In September, California passed multiple AI safety bills, one requiring companies to publish details about their AI training methods and a contentious anti-deepfake bill aimed at protecting the likenesses of actors. (Ars Technica)

← Right now, there’s no significant AI legislation in the works coming from Congress. But I think that will change. The industry would rather have one set of rules at the federal level rather than 50 different state regimes.

NY Targets AI Layoffs with Transparency Rule

New York is moving to be the first state in the US to require employers to disclose when mass layoffs are related to AI adoption.

Gov. Kathy Hochul announced the plan last week that would use New York’s Worker Adjustment and Retraining Notification (WARN) law to require that employers disclose when mass layoffs are related to their adoption of AI. (Bloomberg Law)

← I think this would be tricky to implement: How do you determine that a job loss was tied to AI?

FINRA Alerts Investors to Rising Generative AI Fraud

FINRA warns investors about the increasing use of generative AI by fraudsters to commit identity theft, account takeovers, and other financial scams. (Finra)

• This follows an FBI warning last month that criminals are using AI voice cloning to impersonate loved ones requesting urgent money transfers.

← Unfortunately, as someone who wrote about financial fraud for many years at Bloomberg, I expect to see a significant rise in AI-driven scams.

SIGN OF THE TIMES

Startup Valued at $700 Million After Near Ruin Five Months Ago

Founded in 2017 as a web development software company, StackBlitz was nearly shut down five months ago. But the launch of their AI-powered website building platform in October already has 1 million monthly users and is generating tens of millions in annual recurring revenue.

← This is not so much an AI and finance story, but it’s just striking to me how quickly things change/ how challenging it is to predict where things are headed. (Yahoo)

RESEARCH

AI Using Speech and Text Improves Stock Forecasts: Study

AI that analyzes both what executives say and how they say it during earnings calls makes better stock predictions than text-only analysis, according to researchers from Stevens Institute of Technology.

The study introduces the "ECC Analyzer," a framework that leverages large language models (LLMs) to dissect earnings conference calls (ECCs) by integrating audio recordings, transcripts, and targeted financial queries. The system reduced prediction errors by 27.7% compared to current methods, particularly improving short-term volatility forecasts (3- and 7-day windows).

"Different types of data reveal different insights," Yupeng Cao, a Stevens Institute doctoral candidate and co-lead author, tells me. "Text gives us clear statements and numbers, while audio patterns might reveal additional context not captured in transcripts alone."

How It Works

The ECC Analyzer processes audio using speech-to-text models like Wav2Vec2 and analyzes transcripts with tools such as SimCSE. It then employs a two-tiered approach:

1. Summarization: LLMs condense lengthy call transcripts into key themes.

2. Targeted Extraction: A "Question Bank" co-developed with financial experts—asking, for example, "Did the company raise dividends?" or "What risks were mentioned?"—guides a retrieval-augmented generation (RAG) system to pinpoint critical sentences.

These insights are fused with audio features (e.g., speech rate, pitch) and fed into a neural network to predict volatility. While the model outperformed text-only benchmarks and traditional quantitative approaches like GARCH in tests on S&P 500 earnings calls from 2017, these results are from backtesting and haven't been validated in real-world trading conditions. The research was done in coordination with Fin AI, a group promoting open-source AI tools for finance.

← There’s a long way to go but multimodal AI—combining text, audio and eventually video—could become as ubiquitous as sentiment analysis.

RESEARCH

Citi: AI Impact May Dwarf Internet Era

“AI and agentic AI could have a bigger impact on the economy and finance than the internet era."

That prediction comes from a recent Citi GPS report, which draws on 30+ interviews of AI startup founders, BigTech executives, and other industry experts.

Here are some key findings:

• BigTech's AI Focus: References to agentic AI by BigTech companies increased 17x in 2024, signaling massive industry momentum.

• AI Adoption Surge: Financial services are the second-largest adopters of AI globally, using it for fraud detection, compliance, and hyper-personalized products.

• VC Funding Boom: AI startups attracted 37% of all venture capital funding in 2024, the highest on record. Autonomous agents and digital co-workers led the growth.

• Automation in Action: AI is enabling real-time decision-making, tailored financial products, and efficient compliance processes—reducing costs and boosting productivity across the sector.

• Risks on the Horizon: Cybersecurity challenges and governance risks remain significant as firms accelerate adoption.

MARKETS EDITION

Last Sunday, I broke down how to use OpenAI's new GPT-4o scheduled tasks feature, which automates recurring tasks. The feature can generate regular market reports and research alerts delivered via email at specified intervals.

• The feature can effectively replace traditional Google Alerts with more structured, analytical updates.

• Users can set custom frequencies for automated market summaries and alerts.

• Setup process involves crafting a detailed prompt and selecting delivery frequency.

• Tasks can be managed through chatgpt.com/tasks

• I shared a detailed prompt that requires exact dates, times, and numerical data, for precise market summaries.

An example output included detailed coverage of S&P 500 movements, CPI data, and major bank earnings reports from JPMorgan Chase, Goldman Sachs, and Morgan Stanley.

PODCAST

Next week, I’ll release the first episode of my new podcast, The Alpha Intelligence Show, co-hosted with Francesco Fabozzi, an expert in AI + finance.

ICYMI

With the new podcast, I’m retiring the Five Minutes with interview series.

Five Minutes with Interviews

• Fitch’s Jayeeta Putatunda on causal AI reducing hallucinations

• Former JPM Exec Tucker Balch on scaling investment analysis with AI

• USC's Matthew Shaffer on using ChatGPT to estimate “core earnings”

• Moody’s Sergio Gago on scaling AI at the enterprise level

• Ravenpack | Bigdata.com’s Aakarsh Ramchandani on AI and NLPs

• PhD candidate Alex Kim on signals with executive tone in earnings calls

• MDOTM’s Peter Zangari, on AI for portfolio management

• Arta’s Chirag Yagnik on AI-powered wealth management

• Finster’s Sid Jayakumar on AI agents for Wall Street

• Sov.ai's Derek Snow on AI for fundamental investors

• Bain’s Richard Lichtenstein on AI adoption in private equity

• Snowflake’s Jonathan Regenstein on AI building novel datasets

• Skadden’s Dan Michael on the SEC’s AI stance

• Stardog’s Matt Lucas on hallucination-free AI

• Celent’s Monica Summerville on AI Adoption in capital markets

• Aveni's Joseph Twigg on building a finance LLM

• Persado’s Assaf Baciu on tailored AI marketing at banks

• Professor Alejandro Lopez- Lira on AI-driven stock predictions