The question on everyone’s lips, amid Big Tech rivalries and market swings, is whether we’re in an AI bubble – and if it’s going to burst.
Even some of the industry’s most influential leaders see bubble-like signs. OpenAI’s CEO Sam Altman has cautioned that “someone’s gonna lose a phenomenal amount of money,” while Meta’s Mark Zuckerberg admits an AI bubble is “definitely a possibility.” At the same time, Nvidia’s share price has already wobbled on fears it may be losing its lead in the AI race.
Caution extends beyond the tech sector. The European Central Bank has pointed to “stretched valuations” that could lead to sharp price corrections, while the Bank of England recently warned that current conditions resemble the peak of the dotcom bubble in the late 1990s.
Others argue the fears are overstated. Analyst Arthur Lai said concerns are “overdone,” noting that many investors “overlook the positive economic returns from AI investment”, and Federal Reserve Chair Jerome Powell described the AI boom as “different” from the dotcom bubble.
The bubble won’t pop – because there isn’t one
From my perspective the very notion that there is an AI bubble is incorrect, let alone the idea that it’s bursting. I would call the current situation more of a reality check.
Day to day, the world only sees Generative AI, especially chatbots, but AI has been embedded in our everyday lives for far longer. Our phones have relied on it for years, powering everything from biometric security and personalised recommendations to something as simple as predictive text.
There may be some truth in that businesses are starting to notice the downsides of flashy general-purpose tools, but the AI sector is so much broader than those consumer applications. Purpose-built AI, often powered by small language models, has existed for decades and supports tasks like document processing, chatbots, and translation.
We expect investments in purpose-built AI to hold strong as the c-suite reassess their priorities and focus on measurable impact, rather than the promises that have been driven by hype.
Generative AI fuelled the hype, but purpose-built AI is the answer for enterprises
Generative AI was definitely the catalyst for the hype and boom for investments. Nearly every business is using some form of AI, with Generative AI being the most prominent, and in this year’s survey, 98% of IT decision makers said they are generally happy with their tools. They won’t be tossing them out anytime soon.
However, many also admitted that working with Generative AI is harder than expected. To close the gaps and make it effective for real enterprise challenges, they’ve had to bring in supporting technologies like Process Intelligence and Retrieval-Augmented Generation (RAG).
LLMs may get the hype, but small, purpose-built models are often where the real value lies. A study from Nvidia and the Georgia Institute of Technology noted that AI agents are being used for narrow, repetitive tasks, which small language models are much more suited to. This can be very costly for businesses, for no reason.
Rather than a burst bubble, what we’ll see more of is investment shifting toward complementary technologies that make AI truly enterprise-ready. Not just for meeting summaries and regurgitating content, but for adding quantifiable value in improving procurement processes, accelerating goods to market, helping patients see their specialists faster for better health outcomes, and meeting KYC and AI regulatory compliance standards easier without adding expense.
The new era of AI is moving from broad, general-purpose tools to targeted, purpose-built solutions. As we head into the new year, more organisations will recognise they don’t always need to train a massive model on 30,000 documents, for example. Sometimes a well-crafted regular expression can achieve the goal faster, cheaper, and with far less compute.
Companies need to take a strategic approach
We’ve seen what can happen when businesses jump all in without having a strategy – chaos among staff and IT departments, and tools that don’t actually solve problems. In fact, 60% of IT decision makers in our 2024 survey admitted that their driving factor for investing in AI was FOMO.
Many of them invested in AI because they wanted to be up to date with the latest trends, but now business leaders are starting to realise that generalist Generative AI tools aren’t delivering the outcomes they hoped for, with numerous analysis pieces pointing out that many AI startups and initiatives have high funding but low proven revenue or ROI yet.
What’s often missing is the most important step – a clear, solid AI strategy. What problem are you trying to solve? Are the tools you’re choosing suited to that exact problem?
OpenAI and other major platforms will continue to disrupt, bringing new ways to solve real world problems – but they will never be a one-stop-shop. Other vendors and technologies will always be needed to get tools where they need to be.
Going into next year, I think we will still see ample investment in AI, but more will be spent on purpose-built tools that are focused on solving a business problem.
