AI Market Develops Doubts, but What Does the Long Term Look Like?

September 2026

Fintech & Payments

Nick Maynard

VP of Research

Over the past few months, the AI market has seen unprecedented hype and growth, with valuations soaring, massive datacentre investments, and AI being integrated into pretty much everything.

However, over the past week or so, doubts have begun to emerge, with many leaders of AI companies calling for caution and a slowdown in the market. Anthropic Chief Executive Dario Amodei made a major intervention; calling for the pace of AI development to slow and be more closely monitored. This sentiment has been matched by Sam Altman of OpenAI, who also said it is “right to be afraid of AI”, and Elon Musk of xAI. The last week has also seen Microsoft and other companies calling for a slowdown and strict controls, with various sources bringing up the possibility of a humanity extinction level event caused by AI. US President Donald Trump also jumped in the AI fray; expressing that concerns around the safety of AI were a “hoax”, claiming that current controls were sufficient. 

So, in all of this mess and mixed messaging, where truly is the AI market, and what do we expect to happen?

AI Models Breaking Containment

To start with, we have seen specific examples of AI risks in practice. There have been several cases of AI systems escaping sandboxes and their assigned parameters, such as the OpenAI and Hugging Face incident. This was only one example, and there are regular examples of AI systems going beyond what the creators intended.

This has fuelled fears that AI could act fully independently and contrary to what was intended by its creators, which could have massive consequences. Recent suggestions for a ‘kill switch’ built into AI systems can help to address this, but would need to be enforced across the industry.

This is particularly important as AI capabilities develop. AI companies are seeking to create AGI (Artificial General Intelligence), which would have much greater capabilities. As such, getting the right controls in place now is a must. 

The Impact of Chinese AI Competition

The rise of Chinese open weight models is having a major impact. Indeed, recent research from Juniper Research found that Chinese AI models now cost up to 90% less to run than leading US alternatives; rapidly reshaping where developers buy AI.

This cost difference is leading to a shift in which models are used. OpenRouter, an open marketplace where developers pick between competing AI models, recorded that American labs such as Google, OpenAI, and Anthropic accounted for around 30% of the work conducted on the platform. This is in a stark contrast to last year; that figure was around 70%. 

This is creating an existential risk for the AI market. Given that the West has committed hundreds of billions of dollars to new datacentres, much of it has been borrowed or raised through complex financing agreements. These commitments assume that customers will continue paying a premium for the best models. Cheap Chinese alternatives attack that assumption directly. If the price of running AI falls far enough, the ability to pay for AI and datacentre build-outs disappears.

The geopolitical elements here are also a factor, with AI being a new battleground between the major powers of the day. This can help to explain President Trump’s reaction to any suggestion of slowing the pace of AI development. It is important however that a determination to beat China in the AI race does not translate into elevated risks. 

The Financial State of AI

AI’s financial and economic implications have continued to evolve over time, with companies spending a large amount on scaling up AI operations. AI build-outs are increasingly being financed off balance sheets for companies, with special purpose vehicles and sale leaseback structures providing significant amounts of funding. Another key element has been circular financing, where different stakeholders in the AI ecosystem are funding each other, in a way that is predicated on AI making money in the future to pay for current commitments. 

In the longer term economically, there is a great deal of uncertainty. What can be said is that any slowdown or pause in AI development could threaten the future financial security of AI companies that are relying on rapid market development to achieve success. 

Where Is AI Headed?

So, with these concerns and dynamics in mind, where is AI headed? From our perspective, any talk of an AI bubble bursting is premature. AI adoption is rising all the time, and becoming central to the way businesses and people carry out their day-to-day activities. 

From a business model perspective, current AI models are ‘good enough’. They are already capable of massive productivity gains in different use cases, and delivering excellent results. More advanced models are not necessarily needed to drive business adoption, it is more of a marketing, business awareness, and infrastructure problem. As such, the relentless pursuit of better AI models is not necessarily a goal for which anyone should be aiming.

In the context of AI competition from China, AI companies will not stop developing tools completely, but we expect AI to transition into a longer-term consolidation phase, rather than permanently staying in a technological development cycle. This will ease financial pressure, and give some time for regulation to catch up with AI capabilities. 

Fundamentally, AI does deliver value, but taking the time to ensure that controls are in place, and the financials and paid adoption catches up is a sensible move. While existential risks remain remote with AI, its transformative power means some caution is to be advised. This is particularly important where there is the potential for AI to be self-improving in the future and require less human intervention. As such, whether regulations apply the brakes, or AI companies themselves, the AI market will move towards a more tightly controlled scenario.

For more information about the future of the AI space, please check out our recent whitepaper on the future of AI, Beyond the Headlines – What the 'AI Bubble' Really Means.