Key takeaways:
• The AI value gap is sparking debate about whether we are in an equity market bubble.
• AI is driving efficiency gains at companies.
• As the rise of AI continues, its computational power needs, environmental impact, and training costs cannot be ignored.
During BNP Paribas’ Global Markets Conference 2025, leaders from politics, financial services, academia and technology, looked at how the AI value gap can the world.
The AI value gap: Is the bubble about to pop?
Even though AI is driving efficiency gains at companies, experts at BNP Paribas’ Global Markets Conference 2025 were divided about whether the technology is fuelling an equity market bubble or not.
As one analyst pointed out, since the launch of ChatGPT, the S&P has gone up by 67%, and most of that growth is being driven by the ‘Magnificent Seven’ companies, with Nvidia leading the charge. They also highlighted concerns about the pattern of circular investing, which has become particularly endemic within the AI sector – “OpenAI buys Nvidia chips, which increases the value of Nvidia, and increases the financing of OpenAI. This circular financing is troubling.”
“I would argue that we are not quite blowing AI bubbles, but with such a big focus on so few companies, we are at risk of inflating some fragile balloons,” they concluded.
Citing comments made by Amazon Founder Jeff Bezos, one asset manager described the rise of AI as being a ‘good bubble.’ “We should separate the potential impact on bond and equity markets from the real economy and the innovation that will come out of it,” they said.
Just as the dotcom bubble in the early 2000s sparked a significant market dislocation in the short-term, it paved the way for some truly disruptive companies to thrive, which ultimately – over the long-term – transformed the real economy, they continued, highlighting that the AI value gap is merely a phase expected for such a disruptive technology.
Use case 101: AI is driving efficiency gains at companies
While a lot of emerging technologies – such as blockchain – have in the past been derided for being ‘solutions in search of problems’, experts believe AI is different.
Although a recent study by MIT revealed that 95% of AI pilots are failing, speakers at the BNP Paribas Global Markets Conference 2025 said the technology has plenty of tangible use cases.
Organisations may have struggled to generate much revenue from AI, but the productivity gains are hard to falter. A technology company founder said the deployment of Large Language Models (LLMs) and AI is driving efficiency gains at companies by expediting workstream automation, allowing for previously manual or inefficient processes to be streamlined. They noted that a repeatable task at one client, which previously took five minutes to complete each time, is now being wrapped up in less than 30 seconds, all thanks to AI.
The rise of AI is driving efficiency gains at asset management companies too.
A fund manager said that investment research and portfolio construction are ripe candidates for disruption by AI. “We know there are research agents out there, which a few large asset managers are deploying. There are also fintechs in the market which are managing money based on AI-enabled research agents,” they continued.
At a time when asset management is facing significant cost pressures, several speakers highlighted the rise of AI is driving efficiency gains at companies in the sector.
“We were working on a project, which was quite data intensive. Previously this particular project would have required two or three teams to come together over a period of time. Now I have one person, who is well-versed in the project, use an AI assistant and then verify the work accordingly. This meant the project was completed in a much shorter timeframe,” noted an asset manager.
However, one CIO described Generative AI as being akin to a “super-keen intern”, insofar as it is very useful for processing and summarising straightforward information, but it struggles when handling complex or more nuanced business challenges.
Are we at peak AI yet?
AI keeps getting better and more powerful, with some industry leaders speculating that Artificial General Intelligence (AGI) – a type of AI which could match or even exceed human intelligence – will emerge in the next 5-10 years.
But there are barriers which could impede – or worse derail – the rise of AI and bridge the AI value gap.
For AI to keep accelerating at its current pace, it needs computational power, facilitated by data centres. Although an alternative asset manager said AI’s insatiable appetite for computational power is creating huge underwriting demand for hyper-scalers, this does create logistical problems. For example, the technology company founder said AI businesses in Europe are struggling to obtain the power needed to scale their operations, primarily due to constraints and capacity issues at national electricity grids.
The growth of AI also has serious environmental implications, according to speakers, which need to be considered. Having just unveiled a $500 billion, 10GW data centre in Texas, the technologist highlighted OpenAI now plans to build an additional data centre with a capacity totalling 30GW, rivalling the power consumption of New York.
Another challenge is that as AI models get more powerful, the costs of training them will rise too.
However, the technology company founder said there are several ways to mitigate the environmental and cost impacts of AI – either companies train fewer or more specialised AI models, leverage open-source AI models when building out AI solutions, or increase how much they charge customers.
The AI value gap remains a contentious issue, with experts divided on whether the hype surrounding AI will ultimately lead to a boom or a bust. As the rise of AI continues to drive efficiency gains at companies, it’s clear that this technology has the potential to revolutionise industries and transform the real economy.
However, concerns about computational power, environmental impact, and training costs cannot be ignored. As we navigate this complex landscape, one thing is certain: the AI value gap is a phase that will likely accompany the growth of any disruptive technology. Whether we are at peak AI or on the cusp of something even more significant, remains to be seen. Nonetheless, the power of AI is undeniable, and its implications will be felt across markets, economies, and societies for years to come.
Attended by 761 of the bank’s institutional and corporate clients, the BNP Paribas Global Markets Conference 2025 was an excellent opportunity for colleagues and clients to both reconnect and gain invaluable insights into some of the main dynamics influencing geopolitics and markets today.
