Markets are growing increasingly worried over the prospect of an AI-led tech bubble. Our investment analysts are less so.
Key takeaways
- The software selloff has gone too far.
- AI is not displacing human labour to any great extent at present.
- The application of AI by companies is broad and multifaceted.
Download the Analyst Survey 2026 PDF
Markets are growing increasingly worried over the prospect of an artificial intelligence (AI)-led tech bubble. Fidelity International’s investment analysts are less so.
Our analysts are still bullish on AI. Indeed, none of our analysts who focus on information technology (IT) companies believe their sector is overvalued – the only sector for which this is the case. They also report higher levels of confidence among their management teams than any other sector.
There is of course plenty of debate within Fidelity about the durability of the current boom and the longer-term sustainability of earnings. Some of our shorting analysts, for instance, spend their days looking for tech names for which the hype has gone too far.
But, taken as a whole, the findings of this year’s Analyst Survey would appear to mitigate rising concerns that valuations in the tech sector are stretched beyond repair.
Hardware justifies the hype
Terence Tsai, Fidelity’s tech sector leader and portfolio manager, argues that high levels of investment will benefit companies across the AI value chain. “With hyperscalers raising capex numbers higher than market expectations, that cascades into the capital beneficiaries – the suppliers of the AI infrastructure build and also their suppliers,” he says.
In Tsai’s view, his companies will “strongly outperform” the market this year:
“humans and markets tend to think linearly and have trouble grappling with exponential growth”.
Jonathan Tseng and Austin Kelly focus on North American semiconductors – businesses usually considered among the biggest beneficiaries of the capex spend Tsai references.
They are similarly bullish. “Although valuations are towards the upper end,” says Tseng, “I suspect ongoing earnings upgrades can drag these names upwards in spite of an elevated multiple starting point.”
For Kelly, this points to a wider market transformation. “The emergence of AI as a demand driver in semiconductors has caused and will continue to cause supply/demand tightness across many areas of technology hardware. We have already seen the impact in memory, but we will soon see it in other areas too.”
The software sell-off may have gone too far
The transformational promise of AI means the market is now discounting other parts of the tech sector previously considered among the most predictable names in investors’ portfolios, namely software. The “SaaSpocalypse” began in late 2025 and has intensified through 2026, as the market believes “Software As A Service (SaaS)” applications could be replaced by AI altogether.
But many of our analysts think this sell-off has gone too far. Clare Coleman says the market is wrong to deem all her Australian software and internet companies ‘AI losers’. Instead, after the indiscriminate selling, she’s finding value in those “with strong network effects, proprietary data, and/or regulatory and compliance moats”.
Jack Graham, who focuses on US software and IT services names, believes we may be at the start of a cyclical shift away from growth towards value following higher inflationary expectations this year, supporting some of the less favoured names he follows.
That dynamic, in his eyes, will be exacerbated by rising commodity prices stemming from geopolitical tensions and - for now at least - the conflict in Iran.
Meanwhile, many of our analysts who focus on software think the market is overstating its redundancy risk in a new AI-dominated world.
Siddharth Misra, who focuses on India IT services companies, thinks there remain valuable, fundamental use-cases for IT services names.
Misra cites the role of these companies as “bridging the gap between off-the-shelf agents and driving value in a complex enterprise” by:
- Managing and unblocking the data layer in enterprises
- Orchestrating and integrating multiple agents within the enterprise
- Creating new applications that will emerge because of AI
- And most importantly, having the right guardrails and governance in place
Noriyuki Takizawa reports a similar dynamic among his Japanese companies, explaining that “internal development capacity, governance structures, and security constraints [mean] the assumption that generative AI will quickly displace external IT services demand in Japan may be overstated”.
But both Misra and Takizawa do not think these realities will necessarily reflect in valuations immediately. In fact, Takizawa expects his companies to underperform other Japanese companies this year.
“It may take time (or prove impossible in some cases) for some IT services companies to demonstrate they are not meaningfully impacted by AI disruption.”
Demand justifies the supply
That’s the supply, what about demand? Our analysts following companies from sectors beyond tech report widespread usage of AI.
Almost 90 per cent say that at least some of their companies are seeing productivity benefits, though 62 per cent overall say this is only relevant for a minority of their companies.
There are also signs that AI is beginning to boost companies’ bottom lines, even if the majority are still yet to see that impact.
…and more are seeing a positive impact on profit levels too
Of those analysts who say they have seen their companies benefitting materially from AI, its application is broad, as the chart below shows:
Efficiency gains are seen to vary sector by sector, company by company. Many of our financials analysts report on AI’s ability to assess customers’ creditworthiness more efficiently and speed up anti-money laundering checks, for instance. Some consumer analysts see AI enhancing ad targeting. One industrials analyst notes the “optimisation of road trucking routes” improving pricing algorithms for his European freight forwarders. Another speaks of AI analysing geological data to help his energy companies make new discoveries.
Same headcount, more output
Then there is the impact on workforces. Seven in 10 report their companies either reducing workforce sizes as a result of AI, or (more encouragingly) doing more with the same headcount.
Some are reporting significant layoffs. Sukhy Kaur says one major bank she analyses has saved 100,000 hours per week across development teams.
But the reports of increased productivity with the same number of employees are more common, with some providing examples of companies upskilling workforces or redeploying cost savings elsewhere.
A few analysts see humans spending more time interacting with clients; others think people will need to remain in order to operate AI. “Work might materially change,” says Canadian industrials analyst Robert Reynolds, “as they learn how to manage AI agents.” Michael Gaynor, a fixed income consumer analyst, speaks of AI streamlining existing processes like inventory management. “All of these processes still require human input,” he explains. “But the use of AI tools will make these efforts more targeted and efficient.”
Don’t forget the upside
AI transcends all other market themes in recent years in terms of the scale of investment. Naturally, that has caused jitters. But our analysts who spend their days on the ground, tracking precisely where that spending goes, are broadly sanguine.
“The market goes around asking where are the revenues being produced by AI,” says semis analyst Jonathan Tseng, “while ignoring the fact that the biggest spenders on AI have added over USD $200bn of aggregate revenues in the last three years.”
“Sometimes the answer is hidden in plain sight.”
