Boosted by Muse, Nasdaq’s “quiet surge” raises doubts about a return to AI trading; next week’s Micron earnings become the focal point

Meta AI agent Muse was launched just 12 days ago, with its daily active users peaking at 448,000 and single-day downloads hitting a record 264,000—far outpacing ChatGPT’s performance in the same period and sparking a market frenzy betting on consumer AI reaching a mainstream tipping point.

On September 21, the Nasdaq 100 closed up about 2.83% in a single day, returning to above 30,000 and hitting a new high since June. However, Goldman Sachs trading desk analyst Peter Callahan pointed out that this was a “silent surge”—trading activity scored only 3 out of 10, and the S&P equal-weight index rose just about 18 basis points, with gains highly concentrated in large-cap tech stocks.

This is one of the rare “quiet” nearly 3% Nasdaq surges in my memory.

The explosion of Muse-like products directly opens up a new demand space for CPU chips. Arm Holdings surged over 15% that day, Intel rose more than 12%, and AMD gained nearly 10%.

Meanwhile, a divergence emerged between the credit market and equities: investment-grade CDS spreads widened to a five-month high, indicating that the credit market is cautious about the current wave of optimism; next week’s Micron earnings report will be seen as a crucial validation point for the sustainability of the AI infrastructure trade.

Muse: From Chatbot to Task-Handling AI Agent

All of this was set off by Meta’s consumer-grade AI agent Muse.

Launched on September 8, Muse’s positioning is vastly different from traditional conversational AIs—it can autonomously execute multi-step tasks such as booking trips, managing calendars, and making e-commerce purchases, running in the background instead of waiting for user instructions.

Wells Fargo analyst Ken Gawrelski cited SensorTower data in a client report:

• On September 19, Muse reached 264,000 downloads in a single day, exceeding 200,000 daily downloads for the third day in a row

• On its 10th day after launch, Muse’s daily active users peaked at 448,000

• By contrast, ChatGPT took nearly a year to reach 200,000 daily downloads and 49 days to reach 450,000 daily active users

Ken Gawrelski raised Meta’s price target from $640 to $796, maintained an “Overweight” rating, and wrote: “Muse marks the transition of consumer AI from chatbot to AI assistant, and whoever gets to market first holds an advantage—the more you use your AI assistant, the more valuable it becomes.”

Goldman Sachs trading desk’s Rich Privorotsky directly pointed out the strategic importance of Muse:

Meta has essentially built a direct-to-consumer AI agent factory—shopping, calling, subscribing, administrative tasks, increasingly acting on your behalf. The end state is clear: everyone will have dozens of agents helping you buy groceries, pay bills, book vacations, and arrange your kids’ schooling. Your productivity becomes you plus your team of agents. This is a huge release of efficiency and an enormous multiplier of computing power.

Meta’s share price rose about 11%–12% that day, reaching a new high since October 2025.

Why CPUs, Not Just GPUs

At the same time, the composition of the chip sector rally reveals the market’s logic for betting on a shift in AI architecture.

On September 21, Arm Holdings rose more than 15% in a single day, Intel rose more than 12%, and AMD nearly 10%—these three are all core players in the CPU field, not Nvidia, which is the lead in traditional AI computing power narratives.

According to market intelligence firm Vital Knowledge: Early-stage generative AI mainly relies on GPUs for isolated text and image generation tasks; AI agents, however, must continuously run background programs, logic loops, and local data scheduling, making the demand for CPUs far higher in agent AI than in GPU-intensive workloads.

In other words, the boom of Muse-like products directly opens up a new demand space for CPU chips.

AMD’s market capitalization hit $1 trillion for the first time, surpassing JPMorgan and joining Eli Lilly and Berkshire Hathaway. Peter Callahan from Goldman Sachs’ TMT trading desk noted: “If AMD surpasses these two, the entire top 10 of the S&P 500 will be TMT companies, all with market caps exceeding $1 trillion.”

In addition, reports say OpenAI is accelerating its development of consumer-facing agent products similar to Muse, and the anticipated industry arms race has further pushed up valuations across the entire CPU/cloud ecosystem.

“Silent Surge”: Concentrated Gains and Sharp Internal Market Divergence

Peter Callahan’s dispatch from Goldman Sachs TMT trading desk summed up the session in one sentence: “Nasdaq returns to 30,000, but this is one of the rare quiet 3% surges I can recall.”

Data for the day showed:

• S&P 500 equal-weight index was up only about 18 basis points, while the S&P 500 itself rose about 130 basis points—the gap was among the top five single-day gaps in over five years

• The SOX semiconductor index gained nearly 12% over five trading days, its strongest five-day performance since the rebound in early August 2025

• Options activity hit a near one-month high, with AMD and Intel each seeing more than 1 million call options traded, near yearly highs

According to media reports, some traders aggressively bought technology-related call options at the open, triggering the rare combination of “spot prices up, volatility up”—a typical feature of options-driven rather than fundamental-driven rallies.

Goldman Sachs trading desk records showed overall buy and sell flow was balanced—long-only funds (LO) were net buyers in communication services and information technology, hedge funds (HF) were net buyers in communication services, financials, and consumer discretionary, but overall activity remained low.

Investors Skeptical: Solid Fundamentals, But “Rising Too Fast”

Goldman Sachs TMT’s note directly captured investors’ real-time reactions from that day:

Investors acknowledge the robust fundamentals, but sentiment indicators are more cautious. I got a lot of questions that day like “Isn’t this rising too fast?” or “Where’s the price target?”

Peter Callahan further noted that this reflects recent conversations at Goldman Sachs' technology conferences—there’s clear tension among investors over whether to fully “return to AI infrastructure trades.” The reason: macro pressures from oil prices and interest rates have recently outweighed positive signals about healthy supply and demand, capital returns, and product rollouts at the fundamentals level, “creating what seems like a ‘buyers waiting at higher levels’ trading pattern.”

In other words: the fundamentals are fine, but macro noise keeps investors from chasing prices higher.

Credit Markets Send a Different Signal

The optimism in the stock market isn’t echoed in the credit market.

Bloomberg macro strategist Frank Monkam analyzed: Despite the S&P 500 being close to historical highs, the credit quality of investment-grade debt—as measured by 5-year investment-grade CDS—has deteriorated to a five-month low, and this divergence has continued to expand since the Fed’s rate hike earlier this month.

Frank Monkam wrote:

The stock market continues to price in resilient growth and profitability along with a more optimistic narrative. The credit market, however, is doing the math and demanding higher corporate risk premiums. This clear pricing error raises a question: which market is wrong about risk premium? My tendency is to trust the signal from the higher layer of the capital structure, which will ultimately have the final say.

Two markets, one company balance sheet, two completely different answers.

Oil and Rates: Temporary Relief or Real Turning Point?

Another macro background to this tech rally is falling oil prices.

Brent crude fell below $100/barrel, and according to Bloomberg, satellite data showed that Saudi oil loading at Persian Gulf ports surged over the weekend to its highest level since June, indicating that after Red Sea pipelines were knocked out by drone strikes, Saudi exports shifted back to the Strait of Hormuz.

The drop in oil prices led the 10-year U.S. Treasury yield to fall below 5%. Goldman’s Rich Privorotsky characterized rates and rate volatility as “the biggest macro risk right now,” noting: “The risk of rates decoupling from energy prices is still present—yes, oil prices have come down, but the impact of secondary inflation and global rate shocks is already entrenched.”

Next Catalyst: Micron Earnings

Goldman Sachs TMT’s note made the market’s next focus clear at the end:

Away from conferences and geopolitics, the next catalyst on my radar is Micron (MU) earnings next week.

As a core supplier of AI storage chips, Micron’s earnings will directly test the real strength of AI infrastructure demand. Notably, amid this chip rally, the storage segment has clearly lagged—SanDisk dropped about 3% that day, Western Digital and Seagate Technology were flat—forming a stark contrast with the CPU chip rally.

Whether Micron’s results can bridge this gap will be an important reference for judging the sustainability of a “return to AI trades.”