AI is the hottest topic in equity markets right now but for bond investors the rationale for investing is far less appealing, according to PGIM’s Greg Peters.
With any nascent technology, there will be winners and losers but this dynamic could be more pronounced this time, with large tech giants repeatedly implying through their vast expenditure that the AI market will be a winner-takes-all situation.
For equity investors, this doesn’t matter providing they diversify, as the winners should broadly balance out the losers over time and lead to significant gains. For bond investors, however, this is not the case.
Peters, co-chief investment officer of fixed income at PGIM, warned that if these tech companies are right and there can only be one or two winners, it means bond investors are “knowingly financing a bunch of losers”.
“We don't get the [same] upside. Our upside is just getting our money back, so the losers will overwhelm the one or two winners,” he said.
“If you're an equity investor a few losers won't matter in the scheme of things if you pick the right ones [too]. So there's a very, very different asymmetry between the two. That's why bond investors are just much more bearish – because they have to think about what goes wrong, because what goes right just means you get your money back.”
The total cost spent on the AI buildout so far amounts to around $5.7trn, according to Peters, with companies having to take out more debt to finance their spending.
This has transformed the bond market and how fixed income investors view these tech behemoths. Once considered safer than US treasuries (at least if credit ratings are to be believed), companies such as Microsoft are becoming far riskier.
Microsoft’s bonds were once “considered great credit” because the company had no debt, but now it “has a lot”, said Peters. The same can be seen in Alphabet and Meta, the latter of which he said is will be free cashflow negative this year.
Yet the bond market is “falling all over itself to help finance” the plans of these enormous tech companies and even PGIM has partaken in buying some of these bonds.
It is backing those companies involved in the data centre buildout, although Peters warned that investors “have to be very careful” when investing in these tech giants.
In particular, he said investing through the ‘unsecured space’, which includes most traditional investment-grade credit markets, carries greater risk than those that are ‘secured’ (backed by assets).
“Where we see value is in deals where we can control terms. These are actually providing good opportunities for bond investors as you can have collateralised debt investments with covenants and control,” he said.
“It’s typically short tenured and we're getting paid an extra 100 to 250 basis points over unsecured, which shouldn't be the case.”
The credit market is not the only place where AI could have a material impact, however. The rapidly evolving technology could also heavily change the landscape in sovereign bonds too.
At present, government bond markets are focused on three main areas: inflation (and the controlling of inflation), debt and deficits.
Peters explained there are three ways that a government can get itself out of debt problems. The first is to inflate it away, but since the 1970s a growing proportion of new debt has been index-linked, meaning it moves in-line with inflation.
The second is to raise taxes, which “people generally don't like”, while the third is to cut entitlements “which people also don't like”.
Taking the US as an example, he noted that there could be some scope for higher taxes, particularly for the ultra-wealthy, but warned that despite bipartisan agreement “we'll see if it ever gets through”. The other option is cutting Medicare and Medicaid, which he said is “really hard [to do] politically”.
A final – and pain-free (politically speaking) option – is to grow out of debt. This is where AI could come in.
“I think the way out is through this potential AI boom. We're seeing this [data centre] buildout and the amount of money involved is absolutely eye-popping and staggering. But that could lead to a higher productivity plateau. And if it does, then you grow your way out,” said Peters.
“We had a stat: if you add 50 basis points in productivity over the next 10 years, you basically stabilise US debt at around 117% debt-to-GDP versus 176% just like that. Hope springs eternal, but that's something that could potentially change the game – and without extreme austerity, which I don't think any government right now is willing to pursue, given their seemingly precarious political positions.”