Goldman Sachs joins the DeepSeek frenzy as Chinese AI startup shakes up tech industry

Chinese AI startup DeepSeek has been making headlines with its latest AI models, and last week’s launch of its V3 model only fueled the buzz. But what’s really turning heads is DeepSeek-R1, the company’s latest reasoning model, which has drawn comparisons to OpenAI’s o1—not just in capability, but in offering a more cost-effective alternative. DeepSeek isn’t slowing down, and now, even Wall Street is taking notice.

Goldman Sachs Is Interested—And That’s a Big Deal

The question on everyone’s mind now: How quickly is DeepSeek’s open-source AI making its way into major businesses? The answer is becoming clearer, especially when you look at Fortune 500 companies. The assumption that major U.S. firms would steer clear of DeepSeek just because it’s based in China is proving to be outdated.

According to a new piece by Jon Victor, a reporter at The Information who covers enterprise software and AI, Goldman Sachs is now exploring the possibility of using DeepSeek.

AI at Goldman: What’s Changing?

Goldman Sachs Chief Information Officer Marco Argenti confirmed that his team is looking into DeepSeek, but any adoption would have to go through a security review. That’s a standard precaution to prevent potential vulnerabilities like prompt injections—an issue any company must consider when integrating AI into their workflows, regardless of where the model originates.

“Goldman Sachs Chief Information Officer Marco Argenti told me his team is also interested in using DeepSeek, but the model needs to go through a security screening. That’s presumably to guard against hidden data known as prompt injections, which is a security concern for any company that uses an AI model to power its applications, whether that model is Chinese or not,” Victor wrote.

If Goldman’s Interested, Who’s Next?

Goldman’s consideration of DeepSeek signals something bigger: open-source AI models are no longer playing catch-up. They’re now serious contenders.

The investment bank’s approach to AI is also a reflection of a larger trend in corporate tech adoption. Goldman has already been leveraging AI to improve productivity, especially in software development. Since introducing Microsoft’s GitHub Copilot to its 10,000+ engineers in June, the firm has seen an average coding efficiency boost of 10%. Argenti equated that gain to having an extra 1,000 engineers on staff.

And they’re not stopping there. Goldman is testing more advanced AI coding tools, though Argenti didn’t go into specifics.

Beyond Coding: AI for Bankers and Traders

It’s not just engineers who are seeing AI-driven improvements. Goldman has rolled out an internal AI chatbot for bankers and traders, helping them create client presentation slides and pull insights from lengthy reports. Employees can choose between models from Meta (Llama), Google, and OpenAI, selecting the best fit for each task.

DeepSeek’s latest model is making it clear that financial institutions are shaping AI adoption just as much as tech firms. That said, Argenti has no plans to invest in training AI models from scratch, nor is Goldman looking to build foundation models of its own.

“We’re not in the foundation models business,” he said.

AI in Finance: Hype vs. Reality

Some industry leaders believe AI could eventually be smart enough to guide investment decisions, potentially revolutionizing portfolio management. Jonathan Siddharth of Turing has even suggested that AI models could develop reasoning skills strong enough to predict asset performance.

Argenti isn’t sold on that vision.

An AI-driven chief investment officer sounds great in theory, he told The Information, but real-world limitations remain. At the end of the day, he believes these models are only as good as the data they’re trained on, and there’s no magic involved.

For now, human judgment still holds the final say.

AI’s Everyday Role at Goldman

Outside of work, Argenti relies on AI tools like ChatGPT and Google’s Gemini to stay up to date with AI research and assist with his philanthropic efforts. He even used AI to build a website for his Seattle-based rock band, Element47.

“For web development… it’s just unbelievably, unbelievably efficient,” Argenti told The Information.

With financial giants like Goldman Sachs now exploring DeepSeek, open-source AI is proving that it’s not just an alternative—it’s becoming a competitive force in the industry.

Meanwhile, OpenAI’s Sam Altman, a key competitor to DeepSeek, broke his silence on Wednesday, acknowledging the company’s AI advancements with praise. In a post on X, Altman described DeepSeek’s R1 model as “impressive, particularly around what they’re able to deliver for the price.”

Altman isn’t alone. Nvidia also commended DeepSeek’s advancements in creating cheaper and more energy-efficient large language models (LLMs). In a statement to Bloomberg, the chipmaker called it an “excellent AI advancement” and highlighted the model’s compliance with U.S. technology export controls.