Meta Stock Rallies on Muse: Can It Truly Topple OpenAI and Anthropic?

 

Meta shares rallied on Monday, October 12. The Motley Fool's intraday record shows the stock up as much as 4.7% before settling at roughly $742.96, a gain of 3.38% at 2:56 p.m. Eastern, for a market capitalization near $1.9 trillion and a price close to the top of a 52-week range running from $520.26 to $779.82. The catalyst was specific: Meta released Muse Wonder and Muse Wonder Full that morning.

 

What repriced was not a product launch so much as a question that has hung over the stock for more than a year. Meta has pushed capital spending to a level few companies in any sector have attempted, and the market has never been able to confirm that the money buys genuinely frontier capability. According to Meta's announcement, the two models are the company's most capable to date, aimed at long-horizon tasks, memory and agentic work, which puts them directly against the flagship offerings from OpenAI and Anthropic. Monday's move was effectively a down payment on the idea that the capex has started to produce something.

 

 

 

The size of the move carries information. As Admiral Markets noted, Monday's 3.4% gain was about a third of the 11% single-day jump that followed the Muse agent launch in September, which says the market prices a model release far more cautiously than a consumer product release.

 

The models are built for agents rather than chat. Meta's announcement highlights Active Memory, which carries relevant information across conversations, and Active Focus, which keeps working toward a goal when a task gets hard, tries other approaches and recovers from errors.

 

Pricing already sits in the top tier. Per TechCrunch, Muse Wonder costs $4.00 per million input tokens and $20.00 per million output tokens, roughly comparable to GPT-6 Sol or Opus 5.5.

 

Benchmark claims remain vendor-reported. Meta's research post names Harvey's Legal Agent Benchmark, FocusBench and SWE-Marathon, but the text carries no comparable scores, and independent verification had not appeared at launch.

 

The spending leaves little room for error. Per Meta's second-quarter 2026 results, full-year capital expenditure guidance stands at $130 billion to $145 billion, quarterly capex reached $31.08 billion, and free cash flow compressed to $784 million.

 

The next checkpoint lands this month. Data providers expect third-quarter results on October 28 after the US close, which is when the pace of spending and the sign of free cash flow will answer the question directly.

 

 

 

Two features are worth separating. The gain peaked intraday at 4.7% and faded to 3.4%, which argues against one-way conviction buying. And the price never challenged the September high, with Admiral Markets noting the stock around $742, roughly 5% below the $779.82 peak. The launch repaired sentiment without extending the trend.

 

That restraint makes sense against the recent tape. invezz reported on September 25 that Meta ranked among the 50 worst performers in the S&P 500 as recently as August 18, then rebounded 43% from that trough, gaining 36% during September, after having been down 18% for the year in late July on a soft revenue forecast. A stock that has already recovered that much in two months responds less violently to any single release.

 

 

What trades here is not the model itself but whether a causal chain closes. Enormous compute spending has to convert into capability leadership that third parties can verify, that leadership has to convert into subscription and advertising revenue, and that revenue has to arrive fast enough to keep cash flow from staying negative while capex climbs. Monday's gain says the market raised its odds on the first link while reserving judgment on the rest. Meta's share price will track the verification of that chain over the coming weeks.

 

 

 

Per Meta's announcement, Muse Wonder is available to developers through the Meta Model API and Muse Code, while Muse Wonder Full is restricted to Muse Code subscribers and is not served through the API. Both will power the consumer Muse agent, reaching subscribers first. TechCrunch frames this as the first frontier release since Alexandr Wang restructured Meta's AI organization in August 2025, a period during which Meta deliberately avoided head-to-head comparisons with OpenAI and Anthropic.

 

Meta also disclosed where the models were trained. The research post states they were pretrained and reinforcement-trained at the gigawatt-scale Prometheus data center in central Ohio, under the company's Superintelligence Scaling Framework. For investors, that detail matters because it ties the capex line to a specific output for the first time.

 

 

The two capabilities Meta emphasizes address the places where agents most often fail. Active Memory handles retention across sessions, while Active Focus handles recovery after a task stalls, including switching approaches, checking in before consequential actions and asking for clarification. The announcement cites handling an insurance claim, understanding a lease and making sense of personal finances, alongside professional work in law, finance and health.

 

That matters commercially because it decides whether a model remains a per-query tool or becomes a product that owns a workflow. The ceiling on the first is set by token economics. The ceiling on the second is set by the labor cost it displaces. Meta is clearly aiming at the second.

 

 

This is where restraint is warranted. Meta's research post names Harvey's Legal Agent Benchmark, Vals Finance Agent V2, MemoryBench, FocusBench, SWE-Marathon and DeepSWE, and claims state-of-the-art results across key benchmarks for agents, memory and coding, yet the text publishes no verifiable scores. TechCrunch, reading Meta's model card, reports scores comparable to Fable 5.1 and GPT-6 Astra, with Meta claiming an edge over Fable, Opus and Astra on MemoryBench.

 

Forkast's critique lands on exactly this point, calling the figures vendor-reported numbers with no independent verification and noting that buyers have nothing to check them against beyond Meta's own testing. For a release being treated as proof that spending bought capability, that gap is not a detail. The real scorecard will be whatever independent evaluators and developers publish over the next several weeks.

 

 

 

Meta's June-quarter numbers show the financial consequence of the current spending rate plainly. Revenue reached $60.80 billion, up 28%, while net income fell 14% to $15.85 billion, operating margin dropped to 31% from 43%, quarterly capital expenditure hit $31.08 billion, and free cash flow came in at $784 million against $8.55 billion a year earlier. Full-year capex guidance is $130 billion to $145 billion and total expense guidance is $165 billion to $169 billion.

 

Admiral Markets calculates that first-half capex of $50.92 billion implies $79 billion to $94 billion in the second half to land inside guidance, a pace fast enough to push free cash flow negative. The forecasts invezz cites go further, projecting free cash flow of negative $6.4 billion in 2026 and negative $29.2 billion in 2027, against positive $46 billion in 2025.

 

 

Meta has not guided 2027 capital expenditure, so every number in circulation is a sell-side estimate. Admiral Markets places analyst expectations near $185.6 billion, while invezz cites projections of $197 billion for 2027 and $215 billion for 2028. The spread itself measures the uncertainty. For the stock, whatever management says on October 28 about the 2027 spending path may matter more than the quarter's earnings.

 

There is also a subtle mismatch between capability and capital. Even if third parties confirm that Muse Wonder sits at the frontier, that proves the money was spent effectively, not that it was spent profitably. The first is a question of technical execution. The second is a question of return on capital, and the market ultimately prices the second.

 

 

 

This is the most direct one. The Motley Fool notes that Muse Wonder is available only on the $20 and $100 monthly tiers, which Meta hopes will expand its subscriber base. Admiral Markets cites Oppenheimer analyst Jason Helfstein's illustrative arithmetic: 115 million subscribers at $20 per month would produce roughly $28 billion a year. That figure is a scale illustration, not a disclosed business.

 

Set against capex of $130 billion to $145 billion for 2026, even that outcome covers only part of the bill. The real value of subscriptions may lie less in paying back the investment than in establishing a revenue line uncorrelated with the advertising cycle.

 

 

The larger prize remains the core business. 24/7 Wall St. reports that daily interactions with Meta AI rose 60% after Muse Spark was integrated, that more than a million businesses use Meta's business agents weekly, and that advertising revenue is still growing 27%. If better agentic capability improves targeting and conversion, the payoff never shows up in a subscriber count, yet it may carry the highest return of any path.

 

 

This is the variable the bullish narrative tends to skip. Per 24/7 Wall St., Amazon cut off Muse's shopping access on its site, warning that continued access by an unauthorized AI agent violates its conditions of use. Admiral Markets separately notes that OpenAI launched its rival agent, Dots, on September 29. An agent's commercial value depends on how many third-party environments it can complete transactions in, so if major commerce and service platforms keep closing their doors, capability gains will not translate into revenue one for one.

 

Between now and earnings, every revision to these expectations will show up on the tape first. Follow Meta's next repricing on the METASTOCK perpetual market

 

 

 

On valuation, the roughly 28 times earnings the Motley Fool cites rests on profits already compressed by spending, so the multiple loses reference value if depreciation and operating costs bite harder in 2027. On legal exposure, Meta's own second-quarter filing flags youth-related US trials that may result in a material loss, while Admiral Markets expects a pre-tax legal charge of roughly $10 billion in the third quarter, worth about $3.90 per share. On competition, neither OpenAI nor Anthropic has slowed its release cadence, so any lead can compress within a quarter. On verification, every capability claim currently traces back to Meta's own testing.

 

 

If verification holds, independent evaluations confirm frontier-level performance, Muse subscription conversion accelerates visibly, and the October 28 guidance on 2027 spending lands inside expectations. The stock would have a path back toward the September high at $779.82, with targets converging toward the upper end of the range invezz compiled, including JPMorgan at $920 and KeyBanc at $900.

 

If the verdict is neutral, the models prove broadly comparable without a clear edge, subscription data comes in unremarkable, and capex keeps climbing while free cash flow turns negative. That points to range trading between roughly $713 and $780 on the levels Admiral Markets maps, with the decision deferred into the first half of 2027.

 

If verification fails, independent tests show a gap to the leaders, or distribution blocks spread and constrain where agents can operate, while 2027 capex guidance comes in above expectations. The narrative-driven rally since September would then face retracement risk, and the gap zone between roughly $666 and $681 that Admiral Markets identifies becomes the first reference.

 

 

The October 28 third-quarter report is the pivot. Consensus revenue sits near $63.25 billion, inside Meta's own $61 billion to $64 billion guidance, but the focus belongs on the spending pace, the sign of free cash flow and any language about 2027. Next comes the timing and content of independent benchmark results. Then Muse retention, where Sensor Tower counted roughly 5 million downloads in 22 days while retention remains unpublished, and the gap between downloads and daily actives is what subscription revenue actually rests on. Last is distribution, where whether platforms beyond Amazon follow with restrictions will set the ceiling on the agent business model.

 

 

For James Mitchell, the significance of this release is not the model but the fact that it finally places Meta's capital spending and an externally testable output on the same page. The disagreement about Meta has been hard to resolve because the spending was certain while the output was narrative. Now the output has a name, a price and a list of benchmarks, which means the debate can be settled with evidence rather than conviction. Seen that way, a 3.4% move looks conservative rather than excessive.

 

Two readings look vulnerable. The first equates reaching the frontier with spending well. The model card conclusion TechCrunch relays is parity with Fable 5.1 and GPT-6 Astra, and parity means catching up, not leading, bought with $130 billion to $145 billion of 2026 capex and a quarterly free cash flow of $784 million. For a company long defined by cash generation, that is a change in the valuation framework itself. The second treats subscriptions as the payback route. Oppenheimer's illustration of 115 million subscribers producing roughly $28 billion a year would, even if fully realized, offset only part of a single year's spending. Subscriptions validate willingness to pay. They do not close the gap.

 

Three datasets deserve tracking, and the order matters. First, independent reproduction on long-horizon benchmarks such as MemoryBench and SWE-Marathon, which determines whether the capability story survives contact with outside testing. Second, the free cash flow sign and the 2027 capex language on October 28, which determines how tight the financial constraint becomes. Only third, subscription and advertising conversion, which determines the return. Markets habitually trade the third dataset's optionality first, but the first two are what force a valuation to re-anchor.

 

The cross-asset lesson runs wider than one stock. The AI infrastructure cycle is moving from the storytelling phase into the grading phase, and Meta is simply the first large platform to hand in its work. When a company's capital spending exceeds a fifth of revenue and free cash flow is pressed toward zero, its shares stop behaving purely like a technology holding and start resembling a heavy project investment, priced off project returns. For investors holding both equities and crypto, that transition is worth watching, because it changes how patient capital across the whole risk-asset complex is willing to be about distant payoffs, and patience usually shifts before prices do.

 

 

 

The immediate driver was the October 12 release of Muse Wonder and Muse Wonder Full, which Meta calls its most capable models to date, built for agentic work, memory and coding. The Motley Fool's record shows the stock up as much as 4.7% intraday and settling near $742.96, a 3.38% gain, for a market value around $1.9 trillion. The deeper driver is that investors began re-rating the assumption that Meta's enormous AI capital spending can produce genuinely frontier capability.

 

 

Muse Spark was a lightweight, low-latency model, while Muse Wonder is Meta's first frontier release, larger and aimed at complex work. Meta highlights Active Memory, which carries information across conversations, and Active Focus, which keeps pursuing a goal through difficulty and recovers from errors. Muse Wonder is available through the Meta Model API and Muse Code, while Muse Wonder Full is limited to Muse Code subscribers and is not served through the API.

 

 

Every current claim traces back to Meta's own testing. The research post names Harvey's Legal Agent Benchmark, MemoryBench, FocusBench and SWE-Marathon among others but publishes no verifiable scores in its text. TechCrunch, reading the model card, reports results comparable to Fable 5.1 and GPT-6 Astra. Forkast points out that nothing was independently verified at launch. Until third-party evaluations appear, these should be treated as vendor claims rather than settled facts.

 

 

Per the second-quarter results, full-year 2026 capex guidance is $130 billion to $145 billion, quarterly capital expenditure reached $31.08 billion, free cash flow fell to $784 million, and operating margin dropped to 31% from 43% a year earlier. Admiral Markets calculates first-half capex of $50.92 billion, implying $79 billion to $94 billion in the second half to stay within guidance. Meta has not guided 2027, where sell-side estimates cluster near $185.6 billion.

 

 

Not in the near term. Muse Wonder is available on the $20 and $100 monthly tiers, and Oppenheimer analyst Jason Helfstein's illustrative math suggests 115 million subscribers at $20 a month would generate roughly $28 billion annually. That is a scale illustration rather than a reported business, and it would still represent a fraction of a single year's capital spending. Subscriptions matter more as proof of willingness to pay and as a revenue line uncorrelated with advertising.

 

 

Third-quarter results, which data providers expect after the US close on October 28. Consensus revenue sits near $63.25 billion, within Meta's own guidance of $61 billion to $64 billion. More important than earnings are the spending pace, the sign of free cash flow and management's language on 2027. Admiral Markets also flags a likely pre-tax legal charge of roughly $10 billion in the quarter, which will push GAAP earnings per share well below the adjusted figure.

 

 

It adds a competitor for whom neither capital nor compute is a binding constraint. Muse Wonder's API pricing of $4.00 per million input tokens and $20.00 per million output tokens places it alongside GPT-6 Sol and Opus 5.5, which signals that Meta intends to compete on capability rather than undercut on price. That said, OpenAI launched its rival agent Dots on September 29, release cadence across the field has not slowed, and any lead can be compressed within a single quarter.

 

 

The information above is provided for general market information and analysis only and does not constitute investment advice, financial advice, legal advice, tax advice or a recommendation to trade. Prices of crypto assets, equities and other related financial assets can fluctuate sharply, and past performance, technical indicators and on-chain data do not guarantee future results. The share prices, financial figures, analyst price targets, benchmark claims and market expectations cited here reflect publicly available information at the time of publication, and some of them are vendor-reported or sell-side estimates that have not been independently verified. All of it is subject to change, so the latest official disclosures from the relevant companies and institutions should be treated as authoritative. Readers should conduct their own research and make decisions based on their own financial circumstances, investment objectives and risk tolerance, consulting a qualified professional where appropriate. The MEXC Crypto Pulse team accepts no liability for any direct or indirect loss arising from the use of this information.

 

 

James Mitchell specializes in technical analysis, market trends, and trading strategies for both Bitcoin and altcoins. Based in London, he has over 10 years of experience in financial markets. Before joining MEXC Learn, James worked as a senior analyst at a leading European investment firm, where he developed expertise in risk management and quantitative trading. His transition to cryptocurrency markets began in 2017, and he has since become recognized for his data-driven approach. He holds a Master's degree in Financial Economics from the London School of Economics. His analytical approach combines traditional technical analysis with on-chain metrics to provide readers with actionable insights.

 

Areas of Expertise: Technical Analysis, Market Trends and Cycles, Trading Strategies, Bitcoin and Altcoin Analysis, Risk Management.

 

 

 

Want the fastest access to MEXC's latest updates? Join our official Telegram group now!

MEXC Platform: Download App | Zero Fee Trading | Buy USDT Now

Join MEXC Community: X (Twitter) | Telegram | Discord

Get Insights: MEXC Learn | Terms of Service | MEXC Blog | MEXC News

Account Verification: Understand KYC | How to Complete KYC

External Content Platforms: Substack | Medium | Paragraph | LinkedIn | X(News)