It is now three years since the launch of ChatGPT on 30 November 2022. Since then, the rapid commercialisation of AI has become one of the dominant themes shaping global financial markets.
What began as a breakthrough in generative AI quickly translated into a surge of investment in data centres, semiconductors, and cloud infrastructure, driving a powerful - and highly concentrated - rally in global equities, led by a small group of US mega-cap technology firms.
More recently, the scale of capital expenditure required to support AI deployment has increasingly drawn credit markets into the story, with record levels of corporate bond issuance used to fund investment in AI infrastructure. This note examines how the AI boom has influenced equity market performance and corporate credit conditions to date, and considers some factors that will influence the outlook.
A concentrated equity rally
Since ChatGPT’s debut, enthusiasm around artificial intelligence has been a major engine of US equity performance. Over that period, the S&P 500 has risen nearly 70% [1] with a small group of mega-cap technology firms - Nvidia, Microsoft, Apple, Alphabet, Amazon, Meta, and Broadcom - responsible for almost half of the index’s total gains over the last three years, according to Bloomberg [2].
Nvidia, which makes the specialised chips used in data centres to train and run large AI models, has been the poster child of the AI rally. Since ChatGPT’s launch, shares in Nvidia have risen nearly ten-fold [3], and the company has become the world’s largest, with a market capitalisation just shy of US$4.5 trillion [4]. Nvidia alone now accounts for around just over 7% of the S&P500 market capitalisation [5].
However, when looking more broadly at the companies that have been directly impacted by the AI boom, the weighting is nearly half the index. A large swathe of companies, including hyperscalers and cloud service providers, AI model developers, specialist AI chip manufacturers, and data centre operators all have their fortunes tied to the broader adoption of AI.
JP Morgan has created an index of 30 S&P500 companies that are most affected by AI, spanning sectors such as tech, real estate, utilities, and consumer discretionary. These firms made up about 26% of the S&P500 in late 2022, but following a strong period of outperformance, accounted for around 44% by October 2025 [6].
Outside of the US, JP Morgan estimate that AI-heavy stocks represent around 35% of the MSCI Asia ex-Japan index [7].
Enablers vs. adopters
Against this backdrop, it is clear that developments in AI will have an outsized impact on the performance of global stock markets going forward. So far, the narrative around AI’s positive influence on equity market performance has been a ‘picks and shovels’ story: investors have piled in to the so-called ‘AI enablers’, companies that supply the essential tools, infrastructure, and inputs that make AI possible.
On the downside, there have also been some high-profile casualties regarding companies whose business models have been undermined by AI. For example, the share price of Chegg, an online education company, has fallen by around 97% over the last three years [8].
The evidence regarding a notable AI-related upswing in so-called ‘AI adopters’ - the companies that will ultimately be the end-users of AI - is more mixed. A recent analysis by Read AI concluded that shares in publicly-traded companies adopting AI to boost productivity outpaced the S&P500 by 3.9 percentage points during the year to July 2025 [9].
Read AI noted that, relative to the S&P 500, companies leading in this class of productivity AI adoption were most commonly found in financial services, healthcare, and consumer cyclical industries.
However, hopes for a broad-based AI-related improvement in performance have been undermined by other studies which suggest that AI adoption rates are flatlining, or even falling. Data from the US Census Bureau shows that the share of large companies using AI to produce goods and services has fallen by nearly two percentage points from its peak, to stand at around 12% in September [10].
Although evidence is still emerging, some surveys suggest that the payback from investment in AI initiatives has been underwhelming. A Deloitte survey of C-suite executives (CFOs, COOs and CEOs) found that 45% of respondents reported returns from AI investments that fell short of their initial expectations, while only a minority exceeded them [11].
Other surveys similarly suggest that AI investment has yet to deliver a meaningful uplift in profitability. Against this backdrop, it is perhaps not surprising that investors are starting to question whether revenues will be forthcoming to justify the huge investments currently being undertaken by the AI enablers.
The mega-cap hyperscalers - including Alphabet, Amazon, Meta, and Microsoft - are pouring enormous sums into AI, with their combined capex set to exceed US$380 billion this year according to one estimate [12]. The Organisation for Economic Co-operation Department believes that without AI investment and the construction of data centres, US Gross Domestic Product (GDP) would have contracted by 0.1% during the first half of this year [13].
AI capex pressures Free Cash Flow
This surge in investment is putting pressure on Free Cash Flow (FCF, the cash a company has left after spending money to support and maintain its operations and capital assets) and may limit how much these companies can return to shareholders through dividends or buybacks.
Share buybacks have been an important source of support for equity markets in recent years, but there are signs that this tailwind may be starting to fade; JP Morgan estimate that S&P500 buybacks rose 12% to US$1.1 trillion in 2025, but will fall back 9% in 2026 to US$1.0 trillion in part due to higher investment spending on AI [14].
Rising AI capex also means that the mega-cap companies are shifting from being capital-light businesses to capital-intensive ones, with potential implications for equity valuations. For example, Microsoft’s capex now makes up 25% of its revenue, a percentage that is more than three times what it was 10 years ago, according to Bloomberg [15].
The increase in capex on the part of the big tech companies will lead to a rise in future depreciation expense, which could create a significant burden on earnings per share in the future. In recent years, several of the hyperscalers have made the assumption that their equipment’s useful shelf-life will be longer, a factor which lowers depreciation accounting costs and boosts reported earnings.
However, some observers think that depreciation should be faster, because technology is advancing so quickly that hardware like Graphic Processing Unit (GPU’s) becomes outdated sooner [16]. In turn, a quicker rate of depreciation would lower profits.
In addition, it can be argued that companies with high capital expenditure tend to trade on lower Price/Earnings (P/E) multiples because heavy investment depresses free cash flow in the near term and increases uncertainty around future returns. Investors therefore demand a valuation discount to compensate for execution risk and the longer time lag between spending today and earnings realised tomorrow.
Investors grow more selective
At present, the hyperscalers trade on relatively elevated valuations - for example, Microsoft and Alphabet both have forward P/E multiples of around 28, compared with around 22 for the S&P 500 - reflecting, in part, continued profit outperformance [17].
However, there are growing signs that investors are becoming more discerning when it comes to assessing companies’ AI capex plans, particularly when they are funded through heavy borrowing. This shift is clearly illustrated by Oracle, whose shares fell sharply despite solid quarterly earnings after it flagged a significant increase in debt-financed AI capex, raising concerns over higher leverage and uncertain returns [18].
By contrast, Alphabet has been afforded greater investor tolerance for heavy AI investment, reflecting its strong balance sheet, clearer monetisation through advertising and cloud services, and a competitive advantage from its custom Tensor Processing Units (TPUs) which reduce its reliance on, and act as a rival to, Nvidia’s high-cost GPUs.
This contrast highlights a market now differentiating between companies whose capex is financially sustainable and translates quickly into profit, and those whose spending risks balance-sheet deterioration.
AI and credit markets
Significantly, concerns over heavy AI-related borrowing are now showing up in credit markets. Concerns over Oracle’s increased borrowing have pushed up yields on its corporate bonds, and the cost of insuring against a default on the company’s debt has risen to a 16-year high [19].
More broadly, increased issuance means that AI-related debt is making up a larger share of the corporate bond market, and threatens to push up credit spreads. According to JP Morgan, AI-related companies have issued US$273 billion of Investment Grade (IG) corporate bonds this year, a rise of 50% year-on-year [20].
As a result, AI-related debt is seen making up almost 15% of the US IG bond market this year, up from around 13% in 2024 [21]. Looking ahead, JP Morgan expects rising AI-related borrowing to help push total US investment-grade bond issuance to a record US$1.81 trillion, eclipsing the previous peak of US$1.76 trillion in 2020 [22].
Partly as a result, the bank sees US IG bond spreads widening around 15 basis points in 2026 to 110 basis points [23].
When AI becomes macro-like
All of this suggests that developments regarding artificial intelligence are no longer just a sectoral theme or a technology story. Rather, AI is becoming a macro-like force shaping both equity and bond markets.
What started as a narrowly-focused equity rally driven by a handful of AI enablers is evolving into a wider market narrative, with consequences for valuations, cash flows, and credit conditions. At the same time, investors are moving from indiscriminate enthusiasm to a more disciplined assessment of which companies can fund, monetise, and ultimately generate acceptable returns from the AI revolution.
18 December 2025
[1] FE Analytics
[3] FE Analytics
[4] https://www.macrotrends.net/stocks/charts/NVDA/nvidia/market-cap
[5] https://www.slickcharts.com/sp500
[7] Asia ex-Japan 2026 Outlook – A Year on the Edge. JP Morgan
[8] https://uk.finance.yahoo.com/quote/CHGG/
[9] https://www.read.ai/post/productivity-ai-users-outperform-the-s-p-500-by-29
[10] https://www.apolloacademy.com/ai-adoption-rates-starting-to-flatten-out/
[14] 2026 Global Equity Outlook. JP Morgan
[17] https://yardeni.com/charts/magnificent-7/
[18] FE Analytics
[19] https://finance.yahoo.com/news/oracle-credit-risk-gauge-deteriorates-234104211.html
[20] US High Grade Credit 2026 Outlook. JP Morgan.
[21] US High Grade Credit 2026 Outlook. JP Morgan.
[23] US High Grade Credit 20