Artificial Intelligence in 2026: Is AI a bubble?
Is AI accelerating innovation or quietly undermining the systems the world relies on every day?
Artificial intelligence has revolutionised the way we work, search, and function as a society.
We’ve written multiple blogs on AI in search, personalisation and automation, and how it has affected business.
But, while we often focus on the opportunities and changes that AI brings, there are many that worry about the negative effects that AI has been having.
With the rise in AI-driven job cuts, a seemingly rising number of software outages, and investigations by both UK’s The Guardian (Is AI a bubble that’s about to burst?) and The London Times (Is 2026 the year of the AI backlash?), we sat down with our Software Engineer, Muhammed Zia, to assess whether 2026 could mark a significant correction point for AI adoption after endless growth.
The AI boom: is it a bubble?
Artificial intelligence has seen explosive adoption in recent years.
At the end of 2025, Sam Altman, OpenAI CEO, announced that ChatGPT had reached a landmark 800 million weekly active users, showing that it has become more than just a novelty – it’s a widespread piece of technology.
Altman also drew attention to the uses of the AI engine, saying that “today, 4 million developers have built with OpenAI…more than 800 million people use ChatGPT every week, and we process over 6 billion tokens per minute on the API. Thanks to all of you, AI has gone from something people play with to something people build with every day.”
But these numbers don’t automatically mean that the success will continue.
In fact, as The Guardian discuss in their Today in Focus deep dive, at its peak, Myspace recorded ~800 million users before it’s downfall, proving that there is precedence of a tech bubble bursting.
And, when explosive growth occurs there is often requirement for rapid, large-scale investment, leaving the key economic question: can generative AI ever ‘generate’ enough revenue to justify the investment levels?
Big Money, Bigger Risks
The financial reality is that AI operations are currently not making money.
OpenAI, Perplexity, and Microsoft are just some of the big names that are operating at a loss – with OpenAI forecasting their first profitable year in 2030.
And it’s not just the big names, MIT NANDA reported that 95% of AI pilot projects are failing to yield meaningful results.
Continued investment in data centres and infrastructure means that operators are borrowing more than $1 trillion, according to a recent Morgan Stanley study.
The data centres are not the only cost, though, the ‘immense’ energy needs of AI made up 1.5% of the world’s electricity consumption last year, according to the International Energy Agency.
AI models are not just GPU hungry though, they are also memory intensive – particularly high bandwidth memory (HBM).
An ‘unprecedented’ surge in demand for RAM memory has been reported due to the demands of AI chips and processors. It’s been speculated that prices could increase as much as 50% as manufacturers are reluctant to produce more due to fears of an AI bubble (source).
AI is not only using huge amounts of memory, but it’s also distorting the market, and this affects everyone. It’s predicted that TVs, smartphones, and consumer electronics will be hit with shortages and increased prices.
These mammoth costs and requirements, paired with the high failure rate of projects and the lack of revenue, make AI a risky investment.
And this isn’t the only concern.
Nvidia, the company at the forefront of AI chip making, is the first company to reach a $5 trillion valuation. It invests heavily in AI development and software, but its chips are also bought by the companies it invests in.
This “circular AI economy” is a concern for financial experts.
Warning signs from the financial world
The so-called “circular AI economy” involves chipmakers, cloud providers, and AI firms being financially dependent on each other.
It may sound relatively harmless and a part of normal business, but it puts the whole system at risk of collapse if one player falls into difficulty.
This is where the ‘AI bubble’ becomes more real.
As Peter Cohen points out in Forbes, Nvidia has invested $100 billion into OpenAI which they will then use to buy Nvidia chips. Microsoft accounts for 20% of Nvidia’s revenue and noted that Yahoo! Finance is OpenAI’s largest investor. CoreWeave derives ~60% of its revenue from OpenAI and has Nvidia as an investor, using GPUs as collateral to buy more GPUs.
Michael Burry of The Big Short fame, has publicly bet against major AI players, sharing concern of the current AI hype and comparing it to previous market bubbles like the housing and dot-com bubbles.
These experts have cause to be concerned with experts predicting a potential $20 trillion impact on the US economy in the event of an AI crash and the Bank of England similarly estimating a tens of billions loss in the UK should the worst happen (source).
AI, layoffs, and fragile infrastructure
There is a common narrative around artificial intelligence that it is going to ‘steal’ jobs and replace humans.
While this is a broad generalisation and not something that most of us have to worry about, there has been a recent trend of tech companies replacing human expertise with AI-driven systems, cutting engineering and operations staff.
AI systems still rely heavily on human monitoring, and the removal of experts can create operational blind spots.
There are many variables in system failures and outages but is it a coincidence or correlation?
Let’s explore the facts:
1. Amazon Web Services (AWS)
The outage:
In October 2025, there was an AWS outage that affected multiple services including Snapchat, Slack, and Atlassian (Trello and others).
The outage caused significant disruption and was tracked by ThousandEyes, which described the incident as a ‘complex cascade of infrastructure failures across multiple dependent systems, demonstrating how a single technical defect in critical infrastructure can create ripple effects throughout interconnected cloud services’.
The root technical issue was identified as an automated DNS management system which, even after it was restored, caused cascading failures across the systems that rely on it.
ThousandEyes pointed out that ‘the total recovery time exceeded 15 hours not because the DNS race condition took that long to fix, but because each subsequent phase required time to complete’.
This shows that modern infrastructure and systems can be incredibly complex and fragile, requiring time and attention to fix.
The report highlights that any attempted quick fixes or steps to ‘accelerate recovery by skipping phases or rushing restoration of normal traffic levels would likely have triggered new problems.’
These decisions and insights require human oversight.
The workforce changes:
Also in October 2025, Amazon released news that around 14,000 jobs were being cut across the business.
Beth Galetti, Senior Vice President of People Experience and Technology wrote that ‘this generation of AI is the most transformative technology we’ve seen since the Internet…we need to be organized more leanly, with fewer layers and more ownership, to move as quickly as possible for our customers and business.’
While workforce changes are, unfortunately, a natural part of doing business, this highlights the approach that many are taking as AI develops: slimming teams down.
2. DigitalOcean
The outage:
DigitalOcean reports on software outages and any disruption on their website. A major incident was reported in October 2025 involving Cloud Control Panel and various services.
While they do not share what the issue was or how it occurred, it was classified as a major incident and affected more than one service.
The workforce changes:
There were also layoffs announced at DigitalOcean in October 2025.
Details have not been published, however Kuray Karaaslan points out that lay-offs were likely to have affected multiple teams and include the renowned documentation team.
As Karaaslan highlights, the undermining of documentation and human teams presents a multitude of risks, including:
- Loss of institutional knowledge
- Slower updates and lack of bandwidth
- Increased reliance on external sources
- Lower trust and disruption for users
- Increased internal bottlenecks
3. Microsoft Azure
The outage:
Another cloud computing service, Azure suffered an outage at the end of October 2025 and was reported here.
The outage was caused by ‘a specific sequence of customer configuration changes, performed across two different control plane build versions, resulting in incompatible customer configuration metadata being generated’ and affected the Azure services as well as Microsoft 365, Copilot for Security, and more.
The workforce changes:
In July 2025, it was announced that Microsoft was cutting 9,000 jobs across the business to better position it for AI investment.
A Microsoft spokesperson told the BBC that ‘we continue to implement organisational changes necessary to best position the company for success in a dynamic marketplace,’ as they set out plans to invest £86.6bn in data centres to train AI models.
What happens to AI in 2026?
But what do these incidents, warnings, and changes mean for artificial intelligence in 2026 and beyond?
It won’t be the end of AI, but it may be the end of unchecked optimism.
The companies that thrive will be those that balance AI capability with human expertise and robust infrastructure.
AI’s growth is actively reshaping global supply chains and changing how teams are built, creating new points of failure in systems that were never designed for this level of strain.
