The 10-Year Touched 5 Per Cent, the First Time Since 2023: What Is Doing the Pushing?
The indices barely moved on Monday. $S&P 500(.SPX)$ closed 0.48 per cent lower, $Dow Jones(.DJI)$ 0.29 per cent lower and $NASDAQ(.IXIC)$ Composite 0.56 per cent lower. A layer below, the difference was large: $Philadelphia Semiconductor Index(SOX)$ closed 5.53 per cent lower, its biggest one-day fall since 1 July, while CrowdStrike closed 13.85 per cent higher at a record. Two sectors were priced in opposite directions on the same day.
On Saturday 12 September, Dario Amodei, the chief executive of Anthropic, published "We Must Pace the Frontier", arguing for slowing the rate at which model capability improves so that alignment work, third-party verification and operational rigour can keep up. He set out three steps: outside evaluators placed inside frontier companies, common safety benchmarks and limits on capability growth agreed among the leading laboratories, and wider coordination after that. Elon Musk, chief executive of Tesla, reshared it with the words "Dario is right"; Sam Altman of OpenAI said pacing had lately been a main topic inside the company, and pushed the listing back to 2027 on safety grounds; Demis Hassabis, chief executive of Google DeepMind, said he agreed with the ideas and that the details need work. President Donald Trump rejected the case, saying publicly that fears of AI destroying humanity are a "hoax" and that existing legal and regulatory powers are sufficient.
Memory fell hardest, and the reason is not hard to find. In SanDisk's most recent quarterly statement, revenue rose about half from the quarter before, and the company said about two-thirds of that came from higher prices. Valuations in this tier rest on the assumption that prices keep rising, so once the pace of training spending is publicly up for debate, that assumption is the first to be questioned. $SK hynix(SKHY)$ closed 7.60 per cent lower, $Micron Technology(MU)$ 5.25 per cent lower and $SanDisk Corp.(SNDK)$ 4.98 per cent lower; most other chip names fell between 3 per cent and 8 per cent.
Hewlett Packard Enterprise and $Dell Technologies Inc.(DELL)$, both named on Oracle's earnings call on Friday and both at records that day, handed the gains back on Monday, $Hewlett Packard Enterprise(HPE)$ closing about 11 per cent lower at US$55.41. There was a direct cause: Evercore ISI cut the shares from Outperform to In Line while leaving its price target at US$65, close to a fifth above Monday's close. The analyst did not argue that the business had deteriorated, and praised management's execution through the first year of integrating Juniper Networks; the case rested on price, with the shares up 158.5 per cent for the year and the multiple up from a five-year average of eight times earnings to 13 times forecast financial year 2027. He also noted that memory, NAND and wafer capacity are all tight, so HPE cannot turn its record AI backlog into revenue in the near term.
What got bought was the side that does not live off a training budget, with Alphabet, $Meta Platforms, Inc.(META)$ and $Microsoft(MSFT)$ all closing higher. Cybersecurity rose most, Palo Alto Networks closing about 14 per cent higher, its biggest one-day gain since April 2025. Software's margin over semiconductors set a record, the two matching ETFs finishing 10.67 percentage points apart. The bid was not broad, though: $First Trust NASDAQ Cybersecurity ETF(CIBR)$ rose only about 4 per cent the same day, far less than the leaders, which says the money went into a few names rather than a re-rating of the sector. George Kurtz, chief executive of CrowdStrike, did not fully accept the case for slowing down either, arguing that what matters more is securing what has already been deployed.
Dearer memory does not cut one way only. In the fourth-quarter guidance Broadcom gave on 2 September, gross margin falls to 73 per cent from 78 per cent a year earlier, which the company puts down to the high share of memory in its XPUs. The same guidance sets fourth-quarter AI semiconductor revenue at US$21.7 billion. A distinction is worth making here: guidance is the company's forecast for the coming quarter, not the value of contracts already signed, and Broadcom has not disclosed how much of it comes from any single customer.
A second line of pressure has nothing to do with AI. The 10-year Treasury yield broke above 5 per cent during Monday's session, reaching 5.014 per cent, the first time since October 2023, then eased back to around 4.95 per cent. Oil did the pushing: geopolitical tension escalated and Brent came close to US$108 a barrel. Goldman Sachs moved its call from no change to a hike after Friday's inflation figures, and the market puts the odds of a rise this week between 89 per cent and 92 per cent. The decision comes on the afternoon of Wednesday 16 September, US Eastern time.
$Oracle(ORCL)$ closed 3.65 per cent lower, down for a third session since its results, with the question now centred on cash. Turning about US$664 billion of remaining performance obligations into revenue means building the capacity first, and the capital budget already runs at more than twice the cash the business itself generates, with the gap borrowed. The higher rates go, the harder that is to make add up.
Taken apart, both sides of Monday were betting on the same thing that has not happened. As of the close no company had announced a cut to capital spending and none had delayed a model release; four men had stated a view. Chips were priced on "it will slow down", cybersecurity on "somebody has to manage what is already deployed", and both assume the statements will turn into budgets. Wednesday's rate decision, by contrast, will certainly happen, and its effect on these valuations is more direct.
The above is personal analysis, not investment advice.
💬 【Talking Point】
The 10-year yield touched 5 per cent on Monday and finished below it, with oil and the Fed both pushing. How much of that move would you put down to each?
💰 【Bounty】
Drop your view in the comments and there are Tiger Coins in it for you! 🎁
🔔 Better shared than saved — tag a friend and split the coins!
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
What’s interesting is the market reaction underneath the indices: money rotated away from memory and semis and toward cybersecurity/software. That tells me investors aren’t simply becoming risk-off — they’re becoming much more selective about where AI spending creates sustainable returns.
For me, that rotation is more important than the headline index moves.
Oil/geopolitical tension: 60%
Brent oil moved close to US$108.
Higher oil prices can push inflation higher.
That makes investors expect higher interest rates for longer, which pushes Treasury yields up.
Fed/rate expectations: 40%
Stronger inflation data increased expectations of a rate hike.
Markets were pricing around 89–92% probability of a hike this week.
This directly supports higher Treasury yields.
My view
The oil shock was the main trigger, while Fed expectations amplified it.
The important point is that a 5% 10-year Treasury yield is a big deal for expensive growth stocks. Higher yields make future profits worth less today, so high-valuation technology and AI stocks can face pressure.
For investors:
Short term → I would be cautious about chasing expensive AI/software stocks.
Long term → I would watch for good companies to fall because of higher rates, rather than panic-selling.
Bottom line: Oil 60% + Fed 40%.
I find the cybersecurity move more interesting. The more AI systems and agents enter production, the more security and monitoring they will need. That makes CRWD and PANW interesting to me, although after such a strong one-day rally, I would rather wait for a better entry than chase.
I am also watching the 10-year yield and oil closely because they are more immediate valuation risks. If yields stay near 5%, high-growth AI stocks could face more pressure. I would treat semiconductor weakness as a potential DCA opportunity, but remain selective and focus on companies with strong long-term demand.
@TigerStars @TigerClub @Tiger_comments