PIMCO: The AI split between US dollar and euro investment grade
As AI-related issuance reshapes bond markets, differences in US dollar and euro performance offer new insights into the roles of supply and technical factors.
By Lotfi Karoui, Multi-Asset Credit Strategist, PIMCO
The AI capital expenditure cycle remains solidly on track to eclipse the telecom boom of the late 1990s and become the largest investment cycle since the railway buildout of the 19th century in inflation-adjusted terms. This year alone, AI-related issuers accounted for six of the eight non-financial U.S. investment grade (IG) companies that raised more than $20 billion in the USD corporate bond market. Yet the ultimate scale of the buildout remains deeply uncertain. Bloomberg consensus forecasts imply that hyperscaler capital spending alone will surpass $1 trillion per year from 2027 onward, with no clear signs of moderation.
In fact, capital needs have been so large for the hyperscalers that they started to diversify their debt issuance into a range of foreign currencies. While the most non-USD issuance has still taken place in the EUR market, as a group, these firms have also tapped primary markets in Canada, the UK, Japan, and Switzerland, albeit to varying degrees.
Beyond just the balance sheet re-leveraging impulse that is already widely acknowledged, large issuance figures have led to technical factors becoming an increasingly important driver of relative performance within the AI ecosystem on both sides of the Atlantic.
That said, while 2026 has been a record net supply year in both markets for these firms, it has been far more pronounced for USD denominated bonds. Collectively, hyperscalers now account for nearly 5% of the USD IG index vs. just 1.2% for the EUR IG market.
One side effect of the rapid growth in the share of hyperscalers is that it has started to drive relative performance across both sides of the pond. For example, consider bonds issued by Amazon and Alphabet (Google’s parent company) in both the EUR and USD market.
In theory, the underlying company fundamentals that drive credit spreads should be largely identical regardless of currency. However, a gap in spread performance has started to emerge. This performance differential is difficult to attribute solely to firm-related risks. Instead, it points to indications of demand fatigue in the USD IG market on a relative basis vs. its EUR IG peer.
Index performance and spread dispersion
In addition to the relative value between issuer-matched USD and EUR bonds, there are two other effects that the wave of AI issuance is having. The first is overall index performance. The outsized index share of hyperscaler capital structures has effectively introduced a new risk factor into the USD IG market, whereby index level spreads can drift higher due to the underperformance of just a handful of large issuers exposed to the AI theme.
Since the start of the year, a gap has opened between the USD IG index, both inclusive and exclusive of the five hyperscalers. The EUR IG index doesn’t yet have a large enough concentration for a similar effect. Of course, for the USD IG index this is symmetric, and if the hyperscalers’ spreads started to outperform, they could tighten overall index spreads.
The second effect is spread dispersion. While headline USD IG spreads continue to trade near their recent tightest levels, under the surface there has been significant dispersion at the individual bond level, indicating a time series of the duration-weighted share of the USD and EUR IG markets that have widened by more than 30 basis points (bps) relative to the index over the prior two months.
What is striking about this figure is not that the share of the USD IG market that has widened relative to the index over recent weeks is higher than in its EUR IG peer – it typically is – but rather that the share has recently reached levels not seen since April 2025, the immediate aftermath of the 'Liberation Day' tariff announcements.
At the same time, the under-the-surface dispersion in the EUR IG market, at least as viewed through the lens of this metric, has barely moved at all and remains well-behaved. To be clear, since this measure uses a rolling two-month window, it tends to retrace over time as we have seen over the past few weeks, but it nevertheless remains elevated relative to the past 20 months.
