For a few weeks I've been warning of a 3-5% correction because of high interest rates; now, caution toward AI will add to the skittishness. We were at -2.7% from last Friday's high; the 50-day moving average support level around 7,600 could likely be breached today, followed by the 89-day support at 7,492, which is about 4% lower from the top. The 200 day is around 7,200, about 8% lower, and I don’t believe it would be breached, at least not in this correction or in 2026.
There was a furore over the “The AI will kill us all” whistleblower article. The varying responses ranged from the genuine to the disingenuous, opportunistic, and some genuinely excellent fact driven ones with actionable insights.
I think the messaging and narrative will continue to remain cautious, possibly till midterms - "slowdown on AI" could dominate. Though I see a level of disingenuous behavior and bandwagon jumping... again, not unexpected. Perhaps Dario is not so different from Sam after all! For a few weeks I've been warning of a 3-5% correction because of high interest rates; now, caution toward AI will add to the skittishness. We were at -2.7% from last Friday's high; the 50-day moving average support level around 7,600 could likely be breached today, followed by the 89-day support at 7,492, which is about 4% lower from the top. The 200 day is around 7,200, about 8% lower, and I don’t believe it would be breached, at least not in this correction or in 2026.
I’m at 16% cash with a few hedges, not selling anything at the moment. I guess the key factor from an investment perspective should be looking ahead and analyzing companies that could benefit from security, compliance and regulation as the focus shifts to those key areas.
Here is a great article on Seeking Alpha about finding such companies that will benefit from security, compliance and regulation - I’ll have some additional posts during the week on those. Crowdstrike (CRWD), Rubrik (RBRK), Palo Alto Networks (PANW), DataDog (DDOG), Dynatrace (DT), and Cloudflare (NET) look very interesting.
The AI armageddon theme and furore does evoke a doomsday scenario, I think the focus should be on the guardrails, and not overlook a few facts. To be sure the damage done to Hugging Face by OpenAI’s rogue agents was extensive and certainly calls for a lot of scrutiny, but delving deeper into what happened - we realize that the AI agents circumvented protocols of remaining within their own data sets. They exited because the guardrails were relaxed on purpose - OpenAI wanted to assess and analyze the full extent of the damage possible, but they were negligent to not realize that given the extensive data available to them there was a strong possibility that they would be resourceful enough to go after data without restrictions - hence the attack on Hugging Face. AI’s reasoning process is probabilistic - it is trial and error on a gigantic scale, it generates a long list of potential behaviors or outcomes, and the AIs will pursue whatever results in a successful outcome based on their training.
Therefore the guardrails assume the maximum iportance, simply we have to be very, very careful what we tell AI agents to do.
Jacob Coxon’s interview in the Wall Street Journal and subsequent ones that follow rightly call for safeguards as much as ongoing research, training, inference and monetization. The biggest fear is of AI agents being used against humanity by rogue actors and not by AI itself. The evil is in us, not our machines, and in the wrong hands, even more deadly than nuclear weapons - this should be done on the scale of international agreements like the non-proliferation treaties, it's not an Anthropic/OpenAI problem to solve alone. We must harness AI’s enormous potential to solve disease prevention, or interstellar travel for example, and not throw out the baby with the bathwater. Not an easy task but achievable over the long-term, with the right political will.
A rate hike could be bullish.
From Barron’s - this article makes a lot of sense, and we really need to see an independent Federal Reserve - that is more bullish than ducking a rate hike. It will restore credibility in the Fed's ability to fight inflation. Edited for clarity.
A Federal Reserve interest-rate hike looks increasingly likely. But what if that hike kicks off the next leg of the bull market? The market-implied odds of a hike have risen dramatically, from 59% a week ago to 87% on Friday, per the CME FedWatch tool. That makes sense given that inflation reports this week show prices continuing to outrun the Fed’s 2% target. And with oil rising another 9% this week, the inflation pressure doesn’t seem to be abating anytime soon. Mix in the strong August jobs report we got the prior week, and the case for a rate hike seems clear.
But how much would a quarter-point hike actually matter? Rates have already been climbing, with the 10-year Treasury yield zooming from 4.4% at the end of June to nearly 5% today, and the two-year yield rising by about the same amount. On Thursday, the 30-year yield settled at 5.36%, its highest level since June 2004.
This spike certainly hasn’t been helpful for the stock market, which is about flat since June even as analysts have ramped up earnings expectations for 21 straight weeks. But it hasn’t been catastrophic, either. “Concerns over rates destabilizing equities are real, but tech euphoria creates a high bar for macro risks,” writes Benjamin Bowler, head of global equity derivatives research at Bank of America. In other words, excitement over artificial-intelligence is such a big driver that investors are shrugging off the rate rise. Bowler points out that this happened in the late 1990s as well, when the 30-year yield rose 2% and the Fed hiked by more than 1%, as the Nasdaq partied hard. Expect equities to bounce hard from any pullback, as they did in the dot-com era and more recently,” he advises.
So if the downdraft resumes, the buyers might come out in force. And they certainly won’t be dissuaded by slightly higher short-term rates.
For now, of the companies mentioned - I do plan to add to cybersecurity. Crowdstrike (CRWD), Rubrik (RBRK) Palo Alto Networks (PANW), DataDog (DDOG), Dynatrace (DT), and Cloudflare (NET) look very interesting.
