Investors had a modicum of nervousness coming into this week, which is normal in front of a Federal Reserve meeting. Expectations for a year-end rate cut had oscillated between a 100% probability to below 50% following the October FOMC meeting and the remarks made by Chair Powell following the decision to stand pat. Even if a quarter-point reduction was to be the outcome this week, there was concern that it would be delivered with a hawkish tilt to the language in the statement by Powell. As it turns out, the nervousness was for naught. Well, we got the rate cut and guidance was left in place for potentially more easing in 2026. Good news for stocks and bonds, though the celebrations were cut short after Oracle provided mixed results in its second quarter report. Considering that equity bullishness has been based on not only the advance in technology stocks linked to AI and predictions of lower interest rates, but the risk now is also what happens if both assumptions are thrown into question?

The other day I was sifting through the Pyle wealth management archives and came across a YouTube podcast that I had recorded back in 2009. I had done this one after Nortel filed for bankruptcy and titled it Northern “Telegone” (aka Nortel). This was nine years after the implosion of the tech bubble and in discussions with clients I have often used this as an example of what happens when investors put full faith in a company or a sector when the guidance provided by either is exceptionally enthusiastic. Back in the late 90s, this company could do no wrong. Revenues and profits were screaming and management kept feeding shareholders nothing but positive guidance. Until the results didn’t match the guidance that is. While that guidance came out, analysts would at times raise suspicions but soon squash them for the sole reason that Nortel was a massive and fairly mature company.

In 2025 there has been an explosion in not only capital expenditures by companies on AI, but a surge in credit to help pay for this spending. The above chart shows Oracle’s share price over the past five years relative to the S&P500. I showed a similar comparison with the Mag 7 index back in my Tripping Over Paper newsletter on September 12th. That article dealt with the risk that rising bond yields could create conditions for a valuation adjustment, especially among those stocks that looked to be too far over their skis.

As a provider of cloud infrastructure, Oracle became more intertwined with the AI story after inking a $300 billion deal with OpenAI (the creator of ChatGPT) earlier this year. To meet the demand from OpenAI and others, Oracle ramped up its planned capital expenditures and this week announced that its total investment spend for the year ending May will be around $50 billion, which is $15 billion higher than the estimate it provided back in September. As a result of this investment binge, the company has reported negative free cash flow for three straight quarters.

So, what is the problem? Surely with an order backlog that continues to grow, a few quarters of cash flow burn can be easily made up when revenues start to pick up? For one, reliance on borrowing to increase capital expenditures may not sit well with rating agencies and there is already concern that Oracle could see its investment grade status disappear. As of October, the three main rating agencies (S&P Global, Moody’s and Fitch) all had investment grade ratings, however, both S&P and Moody’s have assigned “negative” outlooks to their ratings. The increased probability that it might lose investment grade status has sent credit default swap (insurance protection for bond investors default) prices to their highest since 2009.

Back on September 26th, Oracle issued a $4 billion 10-year bond, which carries a coupon of 5.2%. The above chart shows the price of that bond, compared to the iShares IBOXX US Investment Grade Corporate Bond ETF (LQD). I have indexed both to 100 on that day since the new issue is priced at 100. As you can see, the Oracle issue has lost value to the tune of almost 4% in the space of less than three months. Now, this is not an ultra long debenture so risks of outright business failure are still very low, but if corporate spreads widen out due to weaker economic growth and if increased debt supply from the corporate sector also pushes spreads out, the loss on this bond could get worse.

Concerns over fading momentum in the AI story is not new and there has been more talk of this being a bubble compared to the start of the year. The early signs of a fade, however, were presumed to come from the demand side. In other words, weaker economic activity would cause consumer and business spending on AI to moderate, causing a slowdown in orders for datacentre capacity and the infrastructure that goes into it. The fact that Oracle reported lower revenues in the second quarter than analysts had predicted might be an early sign.

Then again, one day after the disappointing news from Oracle, Broadcom delivered results from its fiscal fourth quarter that beat estimates on revenues and earnings and attributed most of this to demand for its custom AI chips. Investors initially breathed a sigh of relief on the headlines, but then things turned sour after Broadcom CEO Hock Tan made comments regarding the company’s order backlog. He told investors that backlog was $73 billion, which was under analyst expectations. He tried to clarify in saying this was a minimum, but the gains in the stock were reversed. As you can see in the above chart, Broadcom has done better than many of its peers in the recent AI uncertainty, but I would suggest that caution still needs to be exercised. True, Broadcom’s chips enjoy a similar market share dominance in the custom segment (delivering to the likes of Google and META) as Nvidia has in the general AI arena. That still doesn’t address the question of what happens when end user demand fades.  And that is not just an issue for equity investors betting on continued portfolio growth from AI. It also applies to how central banks view the economic benefits and costs from the AI wave.

This week’s FOMC press conference was always going to be a careful balancing act for Powell. In addition to the persistent threat to the Fed’s independence from the Administration, cracks have been developing in the labour market and the momentum in bringing inflation down was also fading away. As economists anticipated, there was dissent behind this week’s rate decision, with three voting members going against the call. Trump’s handpick, Stephen Miran, again wanted a half-point reduction in the target rate, while the Chicago and Kansas City Fed presidents voted for no change.

Behind the official vote there was actually more dissent. Alongside the decision, the Fed also published the latest quarterly projections presented by members, voting and non-voting. Six of them had called for the funds rate to be in the 3.75% to 4% range at end of this quarter. That is where the range was before the decision to lower the range to 3.5% to 3.75%. Assuming that the two dissenting voters were in this bunch, it tells us that another four also didn’t think rates should be cut. And they have valid reasons for holding this view.

The Summary of Economic Projections (SEPs) saw an upward revision to economic growth estimates from the September meeting, with 2026 now 2.3% (vs 1.8%), 2027 now 2% (vs 1.9%) and 2028 edged up to 1.9% from 1.8%. PCE inflation is expected to be a little cooler in 2026 compared to the September estimates, but the 2027 and 2028 projections remain the same. Looking at this, there doesn’t appear to be much need for further rate cuts, though the median estimate is for one more next year. In other words, the marginal influence of lower rates on the equity market appears to be done, even though market participants still believe that there are two or more rate reductions coming down the pipe. Hence why we have the major indices trading at new record highs. Thanks Santa.

Powell’s push back against criticism of this week’s cut seemed to align with other pundits in recent weeks who claim that the so-called Phillips curve is broken. This was essentially a model that represented a trade-off between inflation and unemployment in the short run. If the unemployment rate rises, reflecting weaker growth, then inflation would decline and vice versa. The argument that this relationship doesn’t hold leans on the assumption that AI is going to create a boost in labour productivity. In other words, we can have stronger growth without an increase in demand for labour, higher wages and hence higher inflation.

This is indeed possible, but we are going to have to wait a while to see empirical evidence. Even still, there is nothing in this argument that makes a case for continued rate cuts. If productivity growth is improving, leading to a sustained stronger economy without more than one single additional rate cut (according to the Fed’s forecast), then it kind of sounds like a new equilibrium. The only way you would get more easing in rates is if the assumption was false – productivity growth was coming more from declining employment (real GDP / employment) and not just technological progress. That would imply weaker economic growth or contraction and eventual disinflation or deflation. Result – decisive action by the Fed to stimulate. As I have argued before, it’s hard to grasp how equity investors would view this scenario as particularly constructive.

But we don’t have to limit our focus on the Fed for an appreciation of how the rate cutting cycle might fading into the distance. Our own Bank of Canada decided this week to leave rates unchanged as it balances a stronger than expected performance by the economy in the last quarter against nagging inflation signals. In Australia, market participants are now anticipating actual rate hikes in 2026, and Japan may also have to nudge rates higher to curb inflation.

So, let’s bring the two themes together. If we do not see a downward path for US rates next year, this could have two effects – create a downward revision to valuations in the tech sector and make additional borrowing by this sector for AI-related investment more expensive. As other countries respond to residual inflation concerns by resisting the desire to lower rates or, in fact, raise them, bond investors may be drawn more to those markets and away from the US. That is over and above the fiscal risks that remain elevated in the US, made worse by this week’s revelation that Washington is engineering a work around that will provide tax relief on R&D by year-end.

This is why our year-end positioning involves a pivot away from US fixed income exposure and maintaining an underweight exposure on the equity front. If the AI momentum fade from lower realized demand comes first and tech stocks retreat, then we will be looking for opportunities in that space. If it is a valuation revisit from flat to rising yields, it might take longer to run its course, but the opportunities will present all the same.

On behalf of the Pyle Wealth Advisory team, have a wonderful and calmly organized weekend.   

Andrew Pyle