The AI market is expected to fragment by 2026 following a volatile end to 2025, marked by tech sell-offs, rallies, circular deals, debt issuances, and high valuations that raised concerns over an AI bubble. This volatility may signal changes in AI investment as investors increasingly scrutinize who is spending and earning money in the sector, according to Stephen Yiu, CIO at Blue Whale Growth Fund. Many investors, particularly retail ones invested via ETFs, have so far not distinguished between companies merely having AI products without business models, those burning cash on AI infrastructure, and those benefiting from AI spending. Yiu emphasizes the importance of differentiation as AI is still in its early stages, noting that while "every company seems to be winning, " market participants will likely begin to separate the various types of AI-related firms. Yiu categorizes the AI landscape into three groups: private startups like OpenAI and Anthropic, which attracted $176. 5 billion in venture capital in the first nine months of 2025 (PitchBook data); publicly traded AI spenders such as Amazon, Microsoft, and Meta; and AI infrastructure providers like Nvidia and Broadcom that receive AI investments from the big tech spenders. Blue Whale Growth Fund evaluates companies by comparing their free cash flow yield (the cash generated after capital expenditure) against stock price to assess valuation fairness. Most of the so-called Magnificent 7 companies trade at significant premiums amid heavy AI investments. Yiu prefers to invest in firms that benefit from AI spending rather than those heavily investing, believing that positioning "on the receiving end" is wiser as AI-related costs increasingly affect company finances. Julien Lafargue, chief market strategist at Barclays Private Bank, notes AI "froth" is concentrated in specific segments rather than broadly spread, with higher risk around companies that secure AI-era funding but lack earnings—such as in quantum computing, where optimism outpaces tangible results.

He stresses the critical need for differentiation. Big Tech’s business models are evolving as they become more asset-heavy by acquiring technology, data centers, and computing power for AI strategies, shifting from asset-light software firms to hyperscalers. This change impacts their risk profile and valuation approaches. Dorian Carrell of Schroders cautions against valuing these firms with old models typical for software or low-capex companies, highlighting uncertainty about funding AI plans despite belief in their long-term success. Tech companies have turned to debt markets for AI infrastructure funding in 2025; Meta and Amazon raised debt but remain net cash positive, unlike more leveraged companies. Carrell expects private debt markets to be particularly important next year. Yiu observes that if AI revenue growth fails to outpace the rising expenses of infrastructure and hardware depreciation, profit margins could shrink, pressuring investor returns. He anticipates wider performance gaps among companies as AI spending increasingly factors into financials, with greater differentiation becoming necessary going forward.