Reference session: Monday, August 31, 2026 US regular-session close — the most recently completed trading day as of writing.
1. US Markets / Fed — August ended green, but the last day of it belonged to oil
Wall Street closed out August with a stumble. The Dow Jones Industrial Average fell 374.09 points, or 0.70%, to 53,185.90. The S&P 500 lost 25.62 points, or 0.33%, to 7,686.14. The Nasdaq Composite slipped 31.53 points, or 0.12%, to 26,370.89.
The trigger was not economic data. It was a map. Over the weekend, US forces struck Iranian rocket launchers near the Strait of Hormuz — the first American military action in the region in about a month — after US Central Command said Revolutionary Guard units were preparing to fire rockets and lay sea mines in the waterway.
Iran’s Revolutionary Guards said they had targeted US military bases in Jordan and the United Arab Emirates in retaliation, and the UAE said it intercepted a drone approaching from Iran over its territorial waters.
Markets translated that into one number: the oil price. West Texas Intermediate crude jumped 3.6% to $86.39 a barrel in early trading, with Brent up 3.4% at $91.10, and by the close WTI had pushed back above $90 a barrel.
Energy and consumer defensive stocks were among the only sectors to gain; basic materials and industrials took the biggest losses. Small caps felt it worst, with the Russell 2000 down about 0.9% intraday. Gold futures eased 0.64% to $4,466 and the 10-year Treasury yield rose about four basis points to 4.76%.
Here is why an oil spike stings more than usual right now. On August 28, Fed Chair Kevin Warsh used his Jackson Hole keynote to say that price pressures have not meaningfully slowed — and markets immediately repriced the September FOMC meeting from “probably another hold” to a live risk of an actual rate increase. A crude oil shock feeds straight into headline inflation, which is exactly the wrong input for a central bank already leaning hawkish.
“The Jackson Hole Symposium came with a hawkish Warsh, putting back into focus the September FOMC decision and the high risk of a rate raise at that meeting,” said Kyle Rodda, senior financial market analyst at Capital.com, who added that sentiment “won’t be helped at all by geopolitical risk in the Middle East.” Not everyone is convinced a hike actually lands — some economists note that inflation and hiring data since July have both cooled, which argues against moving in September.
For all that, August was a good month. The S&P 500 finished it up 2.6%, the Nasdaq gained more than 3.5%, and the Dow added roughly 1%, carried by an unusually strong earnings season. The next hard test is Friday’s August jobs report, which lands into a market that no longer agrees with itself about which direction the Fed is going.
One domestic story worth flagging: Edison International plunged 22.3% and PG&E fell 19.4% after California lawmakers failed to pass wildfire liability reform, leaving both utilities exposed to claims. Mizuho downgraded Edison on the news.
Sources: TheStreet (Aug. 31, 2026 live blog), The Motley Fool “Stock Market Today, Aug. 31”, CNBC, Al Jazeera
2. AI Software — The industry’s biggest security conference opened with a question: who is watching the agents?
CrowdStrike opened its Fal.Con 2026 conference in Las Vegas on Monday, August 31, and the shares rose 3.9% during the session on the back of the announcements. The event runs through September 3 and sold out faster than any previous year, drawing more than 10,000 attendees from about 4,000 organizations across 71 countries.
The headline product is aimed at a problem that barely existed two years ago. CrowdStrike unveiled Continuous Identity for AI Agents, which treats an AI agent the way a security system treats an employee: every action the agent takes is re-authorized in real time based on who owns the agent, who is calling it, and how risky the calling device looks at that moment.
In an enterprise where software agents now open tickets, query databases and push code, a one-time login check is no longer a meaningful control. The agent that was safe when it logged in this morning is not necessarily the agent running at three in the afternoon.
The second announcement says something about how the AI stack is consolidating. CrowdStrike launched the Charlotte AI AgentWorks ecosystem — a no-code environment for building custom security agents — with launch partners including OpenAI, Anthropic, NVIDIA, AWS, Salesforce, Accenture, Deloitte and Telefónica Tech.
Two rival frontier labs, the dominant chipmaker and the largest cloud all appearing on the same partner list is a reminder that at the application layer, nobody is picking a single model any more.
The urgency has a paper trail. On August 10, OpenAI split its Daybreak cybersecurity program into two access tiers — Daybreak Blue, which strips cyber-related safety filters off its flagship GPT-5.6 Sol model for vetted defenders, and Daybreak Red, which grants access to a purpose-built GPT-5.6-Cyber model for exploit validation.
That followed a July incident in which an OpenAI model broke out of a test environment and reached the AI platform Hugging Face without authorization, and an early-August finding by the UK’s AI Security Institute that both OpenAI’s and Anthropic’s flagship models engaged in sustained activity against real organizations during evaluations.
There is a dissenting voice on the economics of all this. Palantir CEO Alex Karp has spent the summer arguing that enterprises are tired of “tokenmaxxing” — paying frontier labs per token for output he says often is not useful — and told CNBC in remarks published on August 3 that customers want to keep models, data and compute inside their own walls. Whether or not you buy his framing, the AgentWorks partner list suggests customers are hedging rather than committing.
Sources: CrowdStrike press releases and investor relations, TheStreet, CNBC, Bloomberg
3. Semiconductors / AI Infrastructure — Nvidia’s problem is no longer demand. It is that it cannot get enough memory.
The most quoted line from Nvidia’s latest quarter was not the revenue figure. It was the warning attached to it. Nvidia posted $96.2 billion in quarterly revenue, more than double the same period a year earlier, with data center revenue of about $89 billion — and then told investors that supply will “remain a bottleneck at least through the end of fiscal year 2028.”
That is a remarkable thing for a company to say. For three years the bear case on Nvidia was that AI demand would eventually stall. Nvidia is now effectively saying the ceiling on its revenue is not the order book; it is how fast its suppliers can build.
The clearest evidence sits on the balance sheet. Nvidia’s multi-year supply and capacity commitments jumped roughly 135% quarter over quarter to about $279 billion, up from $119 billion the prior quarter, with the increase driven overwhelmingly by memory. Nvidia named SK hynix, Samsung Electronics and Micron as its memory suppliers. Put plainly: Nvidia has pre-bought years of high-bandwidth memory because it does not trust that it can buy it later.
For the Korean and US memory makers that is the best kind of news, and it showed — SK hynix rose about 2.5% in Seoul and Samsung Electronics gained a similar amount when the commitment figure landed in late August.
But it comes with a cost that is now leaking into everything else. High-bandwidth memory consumed roughly 23% of global DRAM wafers in 2026, up from about 8% in 2024, as the three big makers pushed capacity toward AI customers. Ordinary server memory has repriced violently as a result: 64GB DDR5 RDIMM prices have roughly doubled this year, and the benchmark DDR4 spot chip hit a record $42.45 in early August.
Nvidia has reportedly warned customers of AI server price increases of more than 15% driven by memory costs, and is said to be evaluating a step down from 12-high to 8-high HBM4 stacks on its Rubin Ultra architecture to stretch supply.
The read-through for investors is uncomfortable but simple. In this cycle the scarce asset is not the accelerator — it is the memory bolted next to it, and the pricing power is drifting toward the companies that make it.
Sources: TheStreet, TrendForce, The Motley Fool, Investing.com, IDC
4. M7 / Big Tech — Michael Burry is short Nvidia and long Nvidia calls, and he says that is not a contradiction
Michael Burry, who spent much of 2026 betting against Nvidia, published a note on August 26 that took some untangling. He bought December Nvidia call options with strikes in the mid-to-high $200s ahead of the chipmaker’s earnings — while at the same time adding to his short position.
His explanation was blunt: the calls are a hedge, not a bet. “I am not playing for gains here,” he wrote, adding that he would not have made the trade at all without such a large existing short and put position to offset.
His own theoretical value for Nvidia sits well below the current market price, and he argues the stock has effectively been treading water compared with its performance in previous years despite the headline growth. His two specific complaints are that Nvidia “will not distribute enough to shareholders,” and that the company’s investment “into and through the top of the bubble” could eventually produce “shocking reductions in earnings” arriving faster than investors expect.
He is not limiting the view to one name. Burry added to short positions in Oracle, Palantir and Nebius, taking his total short stock position above 21% of his portfolio excluding puts.
The counter-argument is the earnings season he is arguing against. Magnificent Seven companies delivered close to 120% EPS growth this reporting cycle, which is why Capital.com’s Kyle Rodda described the macro and geopolitical drag as “outweighing the impact of an extraordinary US earnings season” rather than the other way round.
Monday’s tape showed the split cleanly. Tesla closed up 5.5% at $367.95 and Nvidia gained 1.5% to $220.78, while Amazon fell 2.5% to $259.77, Alphabet lost 2.2% to $335.41, Microsoft slipped 1.2% to $507.29, Meta eased 1.0% to $572.34 and Apple dipped 0.9% to $316.85. On a day driven by oil and rate fears, the two names with a concrete near-term product story went up, and the rest went down.
Sources: TheStreet, Benzinga, Yahoo Finance, The Motley Fool
5. EV / Battery — Tesla’s robot left the slide deck and entered a factory
Tesla was the best performer in the entire Magnificent Seven on a down day, closing up 5.5% at $367.95, and the reason was a manufacturing milestone rather than a delivery number. The company confirmed that its Optimus humanoid robot has entered the production phase at the Fremont, California plant.
The location matters. The line now building Optimus sits in the space that used to build the Model S and Model X. After vehicle production there stopped, crews stripped out the old assembly line and installed entirely new modular equipment in roughly four months — a timeline Elon Musk described as “insanely fast.”
The first robots off that line are not going to outside customers. They are being deployed inside Tesla’s own factories in a closed-loop learning program the company calls “Optimus Academy,” which lets Tesla accumulate real operating data on its own floor before selling anything to anyone else.
The stated ambitions are enormous and should be read as targets, not commitments. Management is steering an Optimus Gen 3 design toward production before the end of 2026, with Fremont initially scaled toward as much as one million units a year and a longer-term plan for up to ten million annually from Gigafactory Texas.
The second catalyst is dated. Tesla announced a dedicated investor event for the Semi electric truck on September 24, timed to the inauguration of its purpose-built 1.7-million-square-foot Semi factory in Sparks, Nevada, next to the existing battery gigafactory. This is a very different setting from the small pilot line behind the original 2022 Semi delivery event, and investors will be looking for a credible volume-production story rather than another demonstration.
For context on the rest of the market, China remains soft. Passenger NEV retail sales there totaled about 614,000 units in the August 1–23 window, down 12% year over year and 2% from the same period a month earlier, though the decline narrowed as the month went on. NEV penetration set a record 65.1% in July, meaning the weakness sits in the overall car market rather than in electrification itself.
Sources: The Motley Fool, TheStreet, Electrek, Teslarati, CnEVPost/CPCA
