Aug 31, 2026 · 33 min
Markets question the Fed as AI and oil face hard constraints
Bloomberg Surveillance TV: August 31st, 2026
The episode connects doubts about monetary-policy credibility with rising demand for hard assets and examines whether AI and energy investment can overcome physical and financial limits.
- 1Gold’s strength reflects declining confidence in policymaking institutions, geopolitical risk, and preference for hard assets over traditional dollar exposures.
- 2AI infrastructure demand remains strong, but TSMC capacity, financing costs, and uncertain frontier-model profits challenge the spending outlook.
- 3Oil-market pressure centers on refining capacity, export flows, and logistics rather than simply the availability of crude reserves.
Don't miss
CJ Muse frames the AI boom’s central unresolved question: whether frontier-model companies can generate profits that justify the infrastructure spending and financing costs.
The brief
Paul Sankey argues that rising gold prices signal doubts about Federal Reserve credibility and broader institutions, as investors favor hard assets over bonds, stocks, and traditional dollar exposures.
Inflation measures and strong nominal earnings leave equities facing a high bar, while markets may be interpreting the Federal Reserve’s data and guidance selectively.
CJ Muse sees durable demand for compute but highlights constrained TSMC capacity, NVIDIA’s ecosystem strategy, and the unresolved economics of frontier-model companies.
The oil discussion shifts attention from Venezuela’s crude reserves to refining capacity, then weighs China, Russia, export policies, and attacks on refining infrastructure.
The standout tension is whether AI’s infrastructure boom can earn adequate returns as higher rates and credit spreads raise the cost of financing data centers.
What was said on this episode
33 statements · 19 positive · 12 negative · 1 mixed · 1 neutral
The Fed should raise rates given persistent inflation and relatively strong employment.
“if you've missed your target for 60-plus months and inflation is above what your mandate is by a lot and the employment conditions are not so bad, it seems like they should hike.”
Listen at 2:45
Taylor-rule estimates imply short-term rates should be 100–200 basis points higher.
“you'd expect interest rates on the short end to be up 100 or 200 basis points.”
Listen at 3:19
The data supports substantial, relatively rapid interest-rate hikes.
“The data is unambiguous. It's meaningful hikes relatively quickly.”
Listen at 3:45
Failure to act on data after missing the mandate undermines Fed credibility.
“when you stand up there and you say, say, we failed at our mandate for 60-plus months, and you don't deliver based upon the data, that's the thing that's actually, the credibility is what matters a lot more than the Ford guidance.”
Listen at 4:25
Gold’s recent acceleration signals doubts about policymaking institutions.
“gold in general is giving us an indication. It's reacceleration over the course of the last couple of weeks.”
Listen at 5:00
Gold and oil are outperforming bonds and stocks in the current environment.
“hard assets... are outperforming soft assets, right, where you see gold and oil continuing to be strong, where you see bonds and stocks soft in this sort of environment.”
Listen at 5:55
Investors are shifting from dollar trades toward hard assets.
“investors are starting to move away from trying to express the U.S. dollar and the fiat currency trade across crosses, and they're moving more to hard assets.”
Listen at 6:30
A credible 200-basis-point tightening would reverse the hard-asset trend, but is unlikely.
“A credible 200 basis points of tightening. But it ain't going to happen.”
Listen at 6:47
Nominal growth and inflation exceed levels consistent with the Fed’s mandate.
“nominal growth is too high. Inflation is too high for the Fed's mandate.”
Listen at 8:35
Markets price in 25% earnings growth over the next two quarters.
“we've priced in 25% earnings growth for the next two quarters”
Listen at 8:58
The equity market is overextended after high-flying stocks’ earnings reactions.
“that highlights just how overextended this market is.”
Listen at 9:37
U.S. consumer activity is weakening rapidly beneath headline economic strength.
“the consumer is starting to fade pretty quickly.”
Listen at 10:23
TSMC capacity is effectively sold out through 2029.
“TSMC is effectively sold out until 2029.”
Listen at 13:24
Frontier-model developers are seeking maximum compute as quickly as possible.
“they are trying to get as much compute as fast as they can.”
Listen at 13:49
AI infrastructure spending will continue through at least 2030.
“our view is that this spending will continue until 2030 minimum.”
Listen at 14:06
AI-related stocks are very cheap if spending extends through 2030 and beyond.
“if you think this can extend into 2030 and beyond, then these stocks are incredibly cheap.”
Listen at 14:21
NVIDIA’s ecosystem financing addresses insufficient AI supply rather than circular financing.
“this is more of a demand problem in terms of just not enough supply.”
Listen at 16:21
NVIDIA effectively operates as a virtual hyperscaler.
“NVIDIA is a virtual hyperscaler.”
Listen at 17:46
Jensen Huang wants global AI usage centered on NVIDIA’s platform.
“He wants the world on NVIDIA's platform.”
Listen at 18:19
Frontier models generate 30% of tokens but capture 90% of economic profits.
“the frontier models probably do about 30 percent of token generation, yet earn about 90 percent of the economic profits.”
Listen at 18:43
The most capable AI systems will capture most economic profits.
“true real intelligence will garner the lion's share, the economic profits.”
Listen at 18:52
AI users will select task-specific models offering the lowest token cost.
“in the future, you know, we're going to optimize. to the question to each specific model that is best for that and offers the lowest cost per token.”
Listen at 19:17
The future AI ecosystem will combine closed frontier and open models.
“we're in a world of closed frontier plus open will be our future.”
Listen at 19:29
Higher interest rates will limit AI data-center investment.
“if interest rates do continue to push higher, you know, that will limit the magnitude of investments.”
Listen at 20:25
Additional Venezuelan crude is not the main solution to current fuel problems.
“we don't need more crude at this point. That's not the issue.”
Listen at 24:00
The major current oil-market problem is insufficient refining capacity.
“we're short refining and that's the big issue.”
Listen at 24:14
Chevron is well positioned and its Venezuelan volumes are growing rapidly.
“Chevron's in very good shape here, and you've seen their volumes from Venezuela grow rapidly.”
Listen at 25:12
Venezuelan oil production could reach two million barrels per day.
“You can probably get to 2 million barrels a day.”
Listen at 25:34
A U.S.–China agreement concerning Iran is necessary to resolve the Hormuz problem.
“the only way you can sort this out is a U.S.-China agreement to do something about Iran”
Listen at 26:32
Russia, formerly the world’s largest diesel exporter, is currently struggling.
“Russia was the second. and biggest diesel exporter in the world, and it's really, really struggling right now.”
Listen at 27:45
Major oil companies have exceptionally strong management and strategy.
“they've never been in better shape in terms of the management and the strategy.”
Listen at 28:27
An oil price of 90 is highly favorable for major oil companies.
“90 is a very good price for these guys. And so they're doing very, very well.”
Listen at 29:25
Oil companies have been the market’s best-performing sector this year.
“the oils have been the best performing sector in the market this year”
Listen at 30:40
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.