Pre-Market Briefing on Sept 22: Meta Up 11.34% After Muse Tops Chart in Ten Days, With Three Fed Speakers Today

The last session was Monday, September 21, and all three indexes rose: the Nasdaq Composite gained 2.26% to close at 27,122.09, the S&P 500 1.49% and the Dow 0.71%, so the Dow trailed the Nasdaq by 1.55 percentage points. The stock in focus is Meta (META), up 11.34% at $741.25, with market value rising about $192B in one session. The industry averaged 4.87% and Meta added 6.47 points on top, so this was a group led higher, not a counter-move. Three Fed officials speak in sequence from 14:05 today, the first substantial comments since the September 16 hike. Academy is the Tuesday metric class: R&D intensity. Data in this report is as of the September 21 US close, and all times are UTC.

 

Money leaned clearly towards growth. The Nasdaq now sits just 0.3% off its 52-week high, while the Dow managed 0.71%.

 

Chips were the strongest block of the day: AMD rose 9.95% to $615.52 and crossed a $1.00T market cap, Intel rose 12.14% to $121.78 and Arm rose 17.16%. ⚠ The move rests on an expectation that a senior US-China meeting this week will include several AI company heads. That is an expectation, not something that has happened.

 

At the other end, Brent crude fell 3.4% to $100.34 as the market priced a resumption of US-Iran talks, which eases the energy pressure behind the hike.

 

It closed at $741.25 and market value rose by about $192B in one session. Volume ran 2.69x the average, and the price sits just 6% below its 52-week high.

 

The driver is its own product. Muse, a paid AI agent launched on September 8, reached No.1 on the US free iPhone chart in ten days with roughly 730K downloads, priced at $20 and $100 a month.

 

⚠ A qualifier that cannot be dropped: paid conversion and subscription revenue for Muse are only disclosed in a later filing. Meta Connect opens Wednesday as the next information point.

 

Five-Dimension Score: trend position 83, peer strength 82, peer rank 100, sector valuation 92, volatility control 70.

 

→ Almost every corner is filled out, and the lowest of them still reads 70. ⚠ Lowest of the five is not the same as weak: all five sit at 70 or above, and sector valuation at 92 is the second highest. Its beta is only 1.24, which makes a move of 11.34% unusual for it.

 

The supporting name, Arm, rose 17.16%. It launched Neoverse CSS N4 and a configurable CPU platform for AI servers that day. AI agents draw more CPU than chatbots do, and Arm licenses chip designs, collecting a royalty per chip shipped on its architecture.

 

Among six larger names in the internet content and information industry, NYSE listings included, five rose and one fell: Meta (META) +11.34% at the top, then Reddit (RDDT) +5.22%, down to Match Group (MTCH) −0.09%. The ends sit 11.43 points apart.

 

→ Meta finished 6.12 points above second place (an excess return of 6.47 points, which roughly corresponds to what the company did on its own), and the only decliner closed essentially flat.

→ ⚠ The industry average of +4.87% is positive, so this was a group led higher rather than a counter-move, and Meta simply rose the most. The reading flips from the previous edition.

 

 

1. Today's reading. Meta traded 48.45 million shares against an average of 18.04 million, a multiple of 2.69x. Close to three times the usual number of shares changed hands, so the price formed after a great deal of trading.

 

2. When the multiple sits near 1. The price moved but the crowd did not grow, and a rise like that is held up by relatively few trades.

 

3. How to use it. The larger the multiple, the more trades tested that price. When a headline gain looks impressive, ask what the multiple was that day.

 

Williams speaks at 14:05, Jefferson at 14:20, and Barkin at 17:00 alongside the two-year note auction. The US opens at 13:30, so all three fall inside the session. There are no major US data today, only the second-tier Richmond Fed manufacturing index.

 

Williams is president of the New York Fed and a permanent voter on the Open Market Committee; Jefferson is vice chair of the Fed, and the two sit closest to the chair. Barkin is president of the Richmond Fed. The Fed turned back to tightening last Wednesday to 3.75%–4.00%, its first move after five straight meetings on hold, so these three are the first time policymakers explain that vote themselves.

 

→ The midpoint of the policy rate, 3.875%, already sits above both rulers for inflation: August headline CPI ran 3.4% and core only 2.4%, and what separates them is mostly energy.

→ Two things to listen for: whether they commit to the one more hike the projections carry, since that would reprice long-duration technology at a higher discount rate; and how they treat oil, given that Brent fell 3.4% to $100.34 in the prior session. If officials treat higher energy as something that fades on its own, core at 2.4% does not carry the case for another move.

→ ⚠ The dot plot median points to one more hike before year end, but that is a forecast, not a commitment. This edition's card gives no figure for the dot plot, and this report does not supply one.

 

The September 21 gains run from Arm at 17.16% down to the Dow at 0.71%: four single names in double digits, two indexes at 2.26% and 0.71%. That distribution is what money leaning towards growth looks like.

 

→ 🔴 The card puts it in one sentence: this hike presses the valuation of technology rather than its prices. So today's question is not whether the data are good, but whether the three speakers commit to that extra move.

 

 

For the most recent full fiscal year, R&D expense divided by revenue runs from TSMC (TSM) at 6.5% to Synopsys (SNPS) at 35.1%, with Nvidia (NVDA) 8.6%, Broadcom (AVGO) 17.2%, AMD 23.4% and Intel (INTC) 26.1% in between.

 

→ The lowest and the highest are about five times apart, and both are indispensable to the same chip.

→ 🔴 Only compare within the same spending structure. TSMC is a foundry and its money goes into fabs, $40.9B of capital expenditure in 2025; Synopsys sells the design tools. Ranking those two ratios means nothing : one spends on buildings, the other on people.

 

1. It does not measure effort. The ratio measures the annual cost of staying in business : skip a year and next year the product may not sell.

 

2. A low ratio has two causes. The work genuinely costs little, or sales grew faster. Nvidia spent $18.5B on R&D, more than Intel's $13.8B in dollars, yet its ratio is only 8.6%.

 

3. A falling ratio is not automatically good. Intel went from 31.2% to 26.1% last fiscal year with revenue barely moving, because about $2.8B came out of the R&D line.

 

A transferable test: ask where the money has to go first : research, fabs or distribution : then ask how long the company survives after stopping. Months, three years or ten decides whether the ratio should be high or low.

 

 

It sells the tools that draw chips: EDA software plus ready-made IP blocks, and its customers are the other five. Revenue last fiscal year was $7.05B, up about 15%. R&D is the product : 35.1% of revenue goes back into it, the highest of the six, which also means it feels an industry slowdown first.

 

A: Both. The industry averaged 4.87% and Meta added 6.47 points on top, which roughly corresponds to what the company did on its own. ⚠ But the industry average is positive, so this was a group led higher, not a counter-move.

 

A: The card does not say that. It gives No.1 on the US free iPhone chart in ten days, roughly 730K downloads, and pricing of $20 and $100 a month. ⚠ Paid conversion and subscription revenue are only disclosed in a later filing, and Meta Connect on Wednesday is the next information point.

 

A: Not as such. 70 is the lowest of the five dimensions, which is not the same as weak : all five sit at 70 or above. Its beta is only 1.24, which is why a move of 11.34% is unusual for it.

 

A: The comparison does not hold. The ratio measures the rent a business pays each year, not effort. TSMC's money goes into fabs; Synopsys sells the design tools. 🔴 The two ratios sit in different spending structures, so ranking them means nothing.

 

A: No. The dot plot is a forecast, not a commitment, and this edition's card only says the median points to one more before year end, without giving a figure. This report does not supply one, nor does it infer a policy path.

 

Disclaimer: This article is compiled and written by the MEXC RealStocks team. The data in this article is based on the closing of the US stock market on September 21, 2026. The content is a compilation of public market information, and individual stocks are publicly discussed targets, which do not represent the recommendation or opinion of MEXC and do not constitute any investment advice. More US stock content: @MEXC | @Alpha_MEXC | @MEXC_Research