Markets
Pikachu Psyduck Togepy Squirtle In The Animated Movie Pokemon:The First Movie Ph
Getty Images

US stocks: Weird, flat, and setting confusing new records

The stock market goes up when stocks go down, and down when stocks go up.

On the surface, this has been a very boring start to the week: a relatively small down day followed by a modest gain has the S&P 500 up 0.1% through Tuesday.

Under the hood, the price action has been so weird that we haven’t seen the likes of it in at least 27 years

Let’s start with Tuesday on its own: A fierce snapback in recently beleaguered tech shares, punctuated by Nvidia’s 6.8% gain, propelled the S&P 500 up by 0.4% on the day. The five biggest stocks in the S&P 500 rose, with Meta and Alphabet each up more than 2%. 

But the advance-decline line for the S&P 500 (the number of stocks up on the day less those that fell) was a whopping -274. There’s never been a session in which the S&P 500 rose this much on a day when that many stocks were actually down, in data going back to January 1997.

And now let’s look at Monday’s tape: the mirror image of Tuesday.  The advance-decline line was above +200, but the S&P 500 fell 0.3%. The success of the many could offset the pain in megacap semiconductor companies.

Tuesday was a superlative unto itself; putting the two days together yields another. In the past 27+ years, we’ve never had a session in which the advance-decline was above 200 but stocks fell followed by a day in which it was below -200 in which stocks rose (or vice versa).

What does this mean? Well, for one, it means we are somehow not running out of fresh ways to point out how market breadth has been (largely) terrible lately. 

More importantly, this dynamic also speaks to an underlying fragility within the stock market. The top-line market environment are calm, the inter-market environment is downright violent.

The trailing 20-day realized volatility of the S&P 500 information technology sector is in the 68th percentile relative to its long-term history (that is, well above average). The 20-day realized volatility of the S&P 500 is in just the 12th percentile, or very below average. That’s the biggest gap between tech sector and index level vol since at least October 2001 (the period for which we have realized volatility data available for all 11 S&P 500 sectors).

The seeming “magic” of high dispersion and low correlations between important parts of the market — that is, megacap tech, in particular Nvidia, versus everything else, is playing an increasingly important role in preventing major fireworks for US stocks at the headline level.

More Markets

See all Markets
markets

Data center trade deep in the red

The data center trade is seeing its steepest sell-off since the market rout that was ignited by President Donald Trump’s Rose Garden tariff announcement back in April.

Goldman Sachs’ themed basket of AI data center shares was down more than 6% at around 12 p.m. ET, putting it on track for its worst day since the tariff announcement.

Losses hammered seemingly every form of input needed for the sprawling concrete server warehouses at the heart of the investment boom.

Hardware makers including data storage companies like Sandisk, Western Digital, and Seagate Technology Holdings, as well as DRAM maker Micron — some of the best-performing stocks in the S&P 500 this year — were taking a licking, as were networking stocks Cisco and Arista Networks and data center builders such as Vertiv Holdings and electrical and mechanical contractor Emcor.

Optimism for all things AI has seemed to evaporate throughout the week, as the stock market greeted lackluster quarterly numbers from Oracle and Broadcom with jittery sell-offs and concern about growing debts that could crater cash flows.

Those worries seem to be spreading to ancillary beneficiaries of the AI boom on Friday, gouging a chunk out of charts that retail dip buyers have not — at least so far — stepped in to buy as we head into the weekend.

markets
Luke Kawa

Oracle denies Bloomberg report that it’s delaying some data centers for OpenAI to 2028 from 2027

Getting a multi-hundred-billion-dollar backlog for cloud computing revenues from data center projects is easy. Building them is hard.

Oracle extended declines to as much as -6.5% on the day on the heels of a Bloomberg report that the cloud giant has pushed back the completion dates for some of the data centers it’s building for OpenAI to 2028 from 2027, citing people familiar with the work. Oracle denied this report, telling Reuters that there have been no delays to any sites required to meet its contractual commitments and that all milestones remain on track.

Shares had fully pared their report-induced drop ahead of Oracle’s reply, but remain in the red for the day.

Bloomberg said the reported postponement was attributed to labor and material shortages.

Oracle has been spending more on capex than Wall Street had anticipated, leading to higher-than-expected cash burn. Management boosted its full-year capital spending plans by $15 billion after reporting Q2 results earlier this week.

Oracle’s cloud infrastructure sales came in short of estimates in its fiscal 2026 Q2, a signal that markets already had reason to doubt its ability to quickly turn its humungous RPO (that is, remaining purchase obligations) into revenues.

Traders also seem to be of the mind that potential delays to data center completions are going to limit sales for what goes into them.

Some of the bigger losers since the Bloomberg headline hit the wires include:

markets
Luke Kawa

Broadcom’s post-earnings tumble is weighing on Google’s entire AI ecosystem

Broadcom’s post-earnings plunge is prompting a sharp pullback in Google-linked AI stocks, which had been on fire thanks to the warm reception to Gemini 3.

The stocks getting hit hard:

A basket of these Google-linked AI stocks compiled by Morgan Stanley is suffering one of its worst losses of the year. This brisk retreat also follows the release of GPT-5.2 by OpenAI.

markets

Citi initiates coverage of Planet Labs with “buy” rating

Planet Labs was up after aerospace and defense analysts at Citi initiated coverage with a “buy/high risk” rating and $19 price target.

The stock is up more than 40% this week, after a strong earnings result that spotlighted the company’s growing opportunity in linking its core business of capturing daily images of the planet with AI technologies.

Citi analysts noted the potential for a positive flywheel effect for Planet Labs as it deepens its focus on integrating AI into its offerings:

“AI is accelerating the conversion of pixels to decisions, where Planet’s daily scan and deep archive offer a uniquely large training corpus and broad-area foundation for automation. AI-enabled solutions (MDA/GMS/AMS) are gaining traction with customers such as NATO and the U.S. DoW, validating the approach of integrating AI into broad-area monitoring products... These AI moves create a compounding advantage: more coverage generates more training data, which improves models, which in turn increases product utility and addressable demand.”

The stock has also caught the attention of some of the retail trading crowd, with call options activity spiking on Thursday as traders rode the market reaction to the results.

Latest Stories

Sherwood Media, LLC produces fresh and unique perspectives on topical financial news and is a fully owned subsidiary of Robinhood Markets, Inc., and any views expressed here do not necessarily reflect the views of any other Robinhood affiliate, including Robinhood Markets, Inc., Robinhood Financial LLC, Robinhood Securities, LLC, Robinhood Crypto, LLC, or Robinhood Money, LLC.