While acknowledging future risks, Principal Asset Management has argued strong fundamentals are cushioning the artificial intelligence (AI) boom – at least for now.
Although comparisons with the dot-com era persist, the asset manager says that despite some bubble-like traits, the current macro backdrop and fundamentals for the AI boom are far more supportive, though with risks further down the road.
In a recent note, the firm’s market strategist Magdalena Ocampo said that with the benefit of hindsight, today’s mega-cap tech stocks – including the ‘Magnificent 7’, whose collective share prices have risen 300 per cent over the past three years to 16 December – look less overblown than during previous bubbles.
“While valuations across the Mag 7, the broader tech sector, and the S&P 500 are at their highest since the dot-com era, expectations remain contained and well below prior extremes,” Ocampo said.
She explained that at the peak of the dot-com bubble, tech stocks traded at a 2.5x premium to the S&P 500, with leaders such as Cisco, Oracle, Microsoft and Intel seeing their multiples surge from ~30x to over 80x between 1997 and the peak.
By contrast, the tech premium to the S&P 500 today is closer to 1.3x, with the earnings multiples of hyperscalers such as Microsoft, Meta and Google remaining rangebound around 30x.
“Importantly, their earnings growth is keeping pace with multiple expansion, suggesting rising prices are grounded in fundamental strength rather than speculation.”
She added that tech fundamentals are also far stronger than during the dot-com era, with profit margins of about 28 per cent and cash flow margins near 22 per cent, compared with roughly 7 per cent and 4 per cent in the late 1990s.
Moreover, even if AI monetisation falls short, Ocampo argued that these firms have shown they can rein in spending and rebuild cash reserves, making a correction possible but a deep, dot-com–style bear market less likely.
While she acknowledged ongoing uncertainty around the returns on heavy capex spending, she said the revenue opportunities from AI are real.
In particular, she pointed to a recent report by audit tech company AuditBoard which found that large audit companies that adopted AI early are already seeing tangible and sizable benefits, with estimated productivity gains of 20 to 40 per cent and annual savings of $3.7 million.
Overall, she said that with the economy in an easing cycle, the AI rally appears early-stage, noting that market bubbles have historically burst during periods of tightening, such as in the dot-com era.
Contained for now, but risks are rising
At the same time, Ocampo noted that although tech fundamentals remain strong and free from bubble-like conditions at present, the ongoing AI boom could bring emerging risks.
She highlighted the shift in capex funding for AI infrastructure as a key area to watch, explaining that while it was initially funded from cash reserves, cash drawdowns and debt financing have increased this year.
“Hyperscalers have issued an estimated $121 billion in debt in 2025, well above the five-year average of $28 billion,” Ocampo wrote.
Morningstar previously echoed this concern in its 2026 outlook, warning that the rapid expansion of AI infrastructure – particularly without clear monetisation plans – poses significant downside risk. Still, with credit markets currently stable, Ocampo said the rising reliance on debt is a concern worth monitoring, though more so in the future.
Meanwhile, she noted that the circulatory of deals among AI players is raising concerns about interdependence.
An example of this process is NVIDIA’s recent investment into the users of NVIDIA GPUs (graphics processing unit), culminating in its $150 billion investment into OpenAI. In turn, these funds will be used by OpenAI to build data centres which will ultimately house many thousands of NVIDIA GPUs – boosting NVIDIA’s financial performance.
As Ocampo explained, arrangements like this can obscure true end demand, while also obscuring vulnerabilities.
“When leading players become tightly linked through capital, supply, and demand, stress at one firm can quickly spread across the ecosystem – creating the kind of domino effect that has characterised past bubbles and, in extreme cases, led to systemic risk,” she said.
However, she added that not all circularity is necessarily problematic, as some arrangements reflect strategic investments by market leaders aiming to expand scale and secure supply – making it simply an area to monitor.
Selectivity and diversification
As the AI cycle moves into its next phase, Ocampo said that being selective will be far more important than broad exposure in identifying the winners and losers of the boom.
Schroders has made a similar point, noting that widespread fears of an AI bubble are prompting investors to scrutinise companies’ AI returns on investment (ROI), a process it expects to intensify in the coming months, driving volatility and divergence.
Finally, Ocampo concluded that diversification will be key, allowing investors to stay exposed to the AI story while managing concentration risk.
She highlighted cyclical sectors, value and small-cap stocks, and regional diversification to Europe and other markets with lower tech exposure in their equity indices as ways to achieve this heading into the new year.
