Here’s the rally; what’s the reason?
The Fed has chosen to hold rates steady for a fifth straight meeting, with FOMC consensus beginning to divide. While President Trump has been vocal in criticising the Fed, Jerome Powell argues that the full inflationary impact of tariffs has yet to be felt. Despite this pessimistic undertone, the US has so far recovered from every potential stumble relating to its self-declared trade war. At the time of the infamous April 2nd Liberation Day, all fingers pointed to an impending economic downturn, all but confirmed by equity markets worldwide falling off a steep cliff.
Figure 1 – Effective tariff rates in the US.
Fast forward to present day, the US has an effective tariff rate of around 17%, the highest level in 95 years – see Figure 1. Surprisingly, however, trade retaliation by the rest of the world has been immaterial, with the US holding all the cards and the stock market reaching all-time highs once again. Economic theory would paint global markets (including the US) as being in an altogether worse scenario due to deadweight loss – with tariffs leading to inefficient economic allocations that result in a net loss in overall economic welfare. A deadweight loss damages both domestic US consumers, and foreign importers. However, President Trump would likely ignore this economic theory, in favour of the more tangible implication of US trade domination, and surmise this as enough evidence for tariffs to remain the new status quo following August 1st.
The US continue to step over the cracks
Despite what equity market prices may suggest, economic signals have not all shown green lights. As referenced in our May Market Synopsis, soft data, such as consumer surveys, has been negative for a while now. Among other signals, the recent June JOLTS report provides evidence that some hard data is starting to finally agree with the negative outlook of consumers – see Figure 2. The recent reading suggests a weakening US labour market momentum, with job openings falling more than expected to 7.4 million (from a downwardly revised 7.7 million), while quits declined to 3.1 million. Layoffs remained roughly unchanged.
Figure 2 – June data confirms “low hiring, low firing” labour trend.
On August 1st, 2025, the Bureau of Labor Statistics (“BLS”) reported that US employers added only 73,000 jobs in July, well below forecasts. Additionally, the combined May and June readings were revised down by 250,000, one of the largest non-COVID-era adjustments since 1979. That same day, President Trump fired the BLS head, Erika McEntarfer. The President took issue with the magnitude of revisions and further suggested McEntarfer was behind an effort to inflate job statistics to aid the democratic image during the 2024 election. While fair criticism has been made regarding the outdated estimation methods and the economic implications of this data release, President Trump’s “shooting of the messenger” raises troubling questions regarding institutional integrity.
Other economic indicators have been showing similar signs of pressures. Despite bouncing back in June, real retail sales were 0.5% lower on a seasonally adjusted basis versus December 2024, speaking to the realised consequence of negative consumer surveys and other soft data releases – Figure 3.
Figure 3 – Retail sales begin to face pressure.
Additionally, revolving credit growth is decelerating, signalling a potential shift in consumer borrowing behaviour – see Figure 4. At the same time, delinquency rates on credit cards and auto loans are nearing levels last seen during the 2008 Global Financial Crisis. These trends, coupled together, reflect the growing cracks in the economy that the US equity market has so far ignored, in favour of the more positive headline signals.
Figure 4 – Delinquency rates approach 2008 highs alongside weakening credit growth.
AI: another reality check
If hard data is starting to show the cracks, why has the rally remained mostly persistent in its upward trajectory? A primary reason lies in artificial intelligence (“AI”), with the fall-and-recovery in the US market largely tracking the overarching trends in the AI trade. Unsurprisingly, AI investment also remains one of the few areas of the US economy experiencing strong growth. In fact, over the past 12 months, capital expenditures (“CAPEX”) by the largest publicly listed hyperscalers in the AI trade (Microsoft, Google, Meta, Amazon, and Oracle, among others) reached $295 billion, nearly 1% of US GDP. The contribution of investment spending on AI is likely even higher when including expenditure by smaller or privately held firms such as OpenAI and xAI.
Figure 5 – Historic CAPEX schedules for AI hyperscalers.
This level of AI-aligned spending may reasonably suggest that the US is reaping the secondary benefit of enhancing productivity and infrastructure. However, the headline number risks overstating the actual impact on domestic infrastructure. For example, a large share of this aforementioned CAPEX is directed towards general purchases of Nvidia chips and other semiconductor equipment; much of which is manufactured outside the US. When the $295 billion figure is decomposed to spending on specifically US-based data centre construction, the number accounts for just 0.1% of GDP (trailing twelve months till May 2025). The implication that AI spend is not even materially adding to US infrastructure comes as a contradiction to President Trump’s initial wishes on the US seeing a manufacturing re-shore. Interestingly, overall manufacturing construction has trended downward since November 2024 – Figure 6.
Figure 6 – Manufacturing construction trending downward since late 2024.
Capital expenditures could increase if more companies go forward with a reshore to the US to avoid tariffs. However, many firms are likely to hesitate on allocating to such a significant investment of building new facilities until there is a greater certainty that the tariffs will remain in effect after President Trump leaves office.
AI, geopolitics and everything in between
Throughout recent history, the S&P 500 has become increasingly global in exposure, with its US-listed constituents generating a significant share of revenue from international markets. This has made the index far more geographically diversified than its US-based domicile implies. As a consequence, global geopolitical risk has seen a growing negative correlation with the US broad market. In contrast to the historic assumption of US assets as a safe haven, we have seen a rising trend that any escalations in geopolitical events actually generate negative returns for US equities, on average.
Figure 7 – Daily returns on the S&P 500 have recently become more negatively correlated with spikes in geopolitical risk.
The AI tech race among various competing nations may further intensify this negative correlation. Given the influence that AI trends already play in the global market, it is a non-zero probability that the new-age tech will motivate the next string of future geopolitical flashpoints. This may result in similar market reactions to what we have already seen with China’s unveiling of DeepSeek. Right now, the AI trade is arguably still priced to perfection in the US, on assumptions that this near future in AI development is 1) frictionless, and 2) capital light.
Taking Figure 5 and Figure 8 into account, we can already see fallacies in these assumptions. Relative to the market’s prior expectations, AI investment does not appear as inexpensive as once thought, with hyperscaler CAPEX being material enough to classify as a GDP component itself. Furthermore, the previously assumed guarantee that AI breakthroughs would be found solely in the US is brought into question by the growing competitiveness of Chinese Large Language Models (“LLMs”). Figure 8 suggests that while the US had the head start, competitors like China did not waste time in catching up.
Figure 8 – A narrowing of the performance differential between US and Chinese AI models over the last two years.
Currently, the AI valuation trend may be exhibiting a “front-running” style of attention, where investors who are afraid to miss out on catalyst events (such as the achievement of true “general intelligence”) are choosing to price in future earnings far before any true justification could be warranted. This pushes a sizable portion of current market valuation further into the realm of speculation, rather than fundamentals. While it is unsound to label a bubble before it has burst, this bias in the trade-off between speculation versus fundamentals has still formed a pattern eerily similar to those of historic bubbles – Figure 9.
Figure 9 – A comparison of the patterns from various historic bubbles, versus the current AI-centric trend between 2020-2025.
Investment takeaway
Despite the proverbial cracks in certain facets of the US economy appearing, it has undoubtedly stood the test of time so far. While this may point to a recession probability being less than previously implied, this does not remove the fact that equity valuations remain stretched by speculatory front-running, particularly in the US. While AI remains an exciting frontier, its current perfectionist pricing exposes the sector to significant risk in the case of a market correction occurring. This rings true especially for AI-aligned companies who rely on components heavily influenced by tariff trends. We remain cautious on US equities due to its over reliance on uncertain drivers, and lean more in favour of valuation on solidified fundamentals.
If you are interested in finding out more about how cognisance of the macroeconomic backdrop impacts our investment decision making process, connect with Integrity Asset Management and let us help you navigate your investing journey.
For more information on this synopsis or to discuss solutions provided by Integrity Asset Management, please contact us at:
Tel: (021) 671 2112
Cell: 072 513 2684 / 084 601 1025
E-mail: nic@integrityam.co.za / herman@integrityam.co.za
Source: Bloomberg, 31 July 2025