Artificial intelligence (AI) is poised to reshape corporate business models, but Fitch Ratings warned that credit risks linked to AI adoption remain concentrated in specific sectors, particularly technology, media, and telecommunications (TMT).
The rating agency said in a report on Wednesday that disruption and overinvestment risks are the primary channels through which AI could influence corporate credit metrics, though for most sectors, AI is not expected to drive near-term rating changes.
Fitch defines overinvestment risk as the possibility that companies could deploy excessive capital on AI infrastructure without generating commensurate returns, potentially weakening credit profiles.
Disruption risk, meanwhile, captures threats to established business models as AI enables new competitors or substitutes.
Both risks are most pronounced in asset-light industries, such as software, media, and services, where value is concentrated in intangible assets like intellectual property, proprietary data, and human capital.
According to the report, the AI-driven capital expenditure (capex) boom is primarily concentrated among a handful of hyperscalers and cloud service providers.
Fitch noted that the “big four” — Alphabet, Microsoft, Amazon, and Meta — plan to spend roughly $650 billion in 2026 on AI-related infrastructure, nearly matching cumulative spending from 2020 to 2024.
Despite these headline numbers, the broader universe of Fitch-rated corporates is exercising prudence, with capex largely aimed at meeting visible demand rather than aggressive expansion.
Data from Fitch’s Global Corporate Cash Flow Monitor, covering over 1,500 non-financial issuers, shows that capex intensity for North American corporates (excluding hyperscalers) is projected to rise slightly to 7.4 percent of revenue in 2025–26, compared with 6 percent to 7 percent in the previous five years. These increases are largely supported by strong operating cash flows and are not expected to materially affect free cash flow.
Outside technology, power utilities are among the largest capital spenders, though their investments are driven by structural needs such as aging infrastructure, renewable energy integration, grid hardening, and extreme weather resilience rather than AI-specific buildouts.
Utilities have generally avoided committing to long-term capacity expansions solely to meet AI-driven demand, emphasizing measured investments to maintain balance-sheet flexibility, said Fitch.
Fitch also highlighted that supply chain bottlenecks across semiconductors, data centers, and power grids are likely to slow AI capacity expansion.
Lead times for memory, storage, and semiconductor fabrication remain tight, while power grid interconnections and permitting delays extend timelines for new data-center connections.
These constraints may keep hyperscaler capex elevated even if delivered capacity ramps more gradually, as firms may incur higher costs for expedited procurement or interim solutions, it noted.
While overinvestment risk is concentrated in hyperscalers, Fitch opined that disruption risk is centered in asset-light industries where AI can easily substitute human labor or existing processes.
It identified software, services, and media as the most exposed sectors.
For software, it sees AI can accelerate development of simple, narrow-scope applications, such as workflow automation or customer interfaces.
However, it opined that enterprise software with mission-critical functions and high switching costs remains resilient, particularly in regulated sectors like healthcare, finance, and the public sector.
As for services, Fitch highlighted that customer service outsourcing, data analytics, and other repeatable workflows face potential displacement by AI.
Companies with proprietary data, entrenched ecosystems, long-term contracts, or regulatory insulation are less vulnerable to rapid disruption, it added.
As for media, Fitch said AI-generated content reduces production costs, and while fully AI-generated content has not yet matched studio-quality output, incremental AI tools—like automated summaries—have already impacted digital advertising revenue and search traffic for publishers.
Fitch also noted that resilience is typically higher among firms with mission-critical offerings, regulatory integration, proprietary data, and financial flexibility, enabling them to invest in AI capabilities, absorb transition costs, and extend their competitive moats.
Despite widespread interest in AI, Fitch cautioned that transformative positive effects on corporate ratings remain limited.
While AI may enhance efficiency and reduce costs across various sectors, it noted revenue opportunities tied to AI adoption are still emerging and largely incremental.
For instance, pharmaceutical and media companies may benefit from faster research and development (R&D) or content production cycles, retailers could use AI for more targeted marketing, healthcare providers might streamline billing processes, and auto manufacturers could offer AI-enabled features on a subscription basis.
However, these incremental gains are difficult to quantify and unlikely to influence ratings in the near term, it added.
It also said most corporates are proceeding cautiously with capital investments, prioritizing visible demand and balance-sheet flexibility.
Even in sectors exposed to AI, such as semiconductors and cloud services, end-market diversification and customer breadth limit downside credit risk from a potential slowdown in AI buildouts, it said.