Everything About AI
NextFin News - U.S. stocks closed higher on Wednesday, October 21, as the major indexes rebounded into the evening's most-watched earnings reports, with the S&P 500 gaining 0.59% to 7,757.85, the Dow Jones Industrial Average adding 0.69% to 52,856.09, and the Nasdaq Composite climbing 1.05% to 26,702.86. The rally set the stage for two verdicts on the artificial-intelligence spending boom: Meta Platforms reported third-quarter revenue of $64.6 billion, up 26% from a year earlier, and Tesla delivered an earnings-per-share beat even as its revenue stalled near $28.05 billion. The shared message from the two Musk-adjacent giants is that the AI era is rewarding operators who can spend heavily while still protecting margins and cash flow.
The Tape: A Broad Rebound Into Earnings Night
All three major averages finished the session well off their lows from the prior week, with the Nasdaq posting its strongest day in roughly two weeks. Technology and AI-linked names led the advance: Tesla rose 3.41% to $386.47 during regular trading, while Meta added 0.47% to $697.52, both holding steady ahead of reports scheduled after the 4:00 p.m. ET close. The calm was deliberate rather than complacent. Investors were positioned for a binary pair of prints that would test whether the market's tolerance for AI capital expenditure can survive contact with actual cash-flow statements.
The setup mattered because the AI trade had entered a skeptical phase. A week earlier, heavyweights in the chip complex had sold off on concern that data-center buildouts were outpacing monetization. By Wednesday morning, that pressure had eased enough for futures to turn positive, helped by SpaceX's report late Tuesday that revenue jumped 161% year over year to $14.58 billion and by Elon Musk's disclosure that SpaceX and Tesla had formed special committees to discuss a potential combination. The two companies together carry a market value of roughly $3.5 trillion, and the merger speculation added a second layer of drama to an earnings night that was already going to define the mood for the rest of the quarter.
The day's breadth was notable because it was not confined to the mega-caps. Energy stocks headed for a record close as crude climbed on Middle East headlines, and chip makers rebounded from a rough patch that had knocked the Nasdaq down 4.5% over the prior week. Oil's rise - Brent futures above $106 a barrel and West Texas Intermediate near $97 - added an inflation wrinkle that bond traders watched closely, with the 10-year Treasury yield around 5.12%. In that environment, a clean 1.05% Nasdaq gain was less a relief rally than a statement that earnings were about to matter more than macro noise.
Meta: The Core Business Is Paying for the AI Buildout
Meta's third-quarter results answered the spending question with a split verdict. Revenue of $64.6 billion beat the roughly $63.3 billion consensus and grew 26% year over year, with advertising revenue of $63.06 billion rising at the same pace. Diluted earnings per share came in at $3.83, up from $1.05 a year earlier, while net income reached $9.86 billion against $2.71 billion in the prior-year period. But the income statement showed what that growth is costing. Total costs and expenses climbed 68% to $51.6 billion, and operating income fell 37% to $12.99 billion, compressing the operating margin from 40% to 20%.
The headline distortion deserves parsing. A $9.89 billion charge tied to an agreement with 52 state attorneys general over teen mental-health litigation weighed on the quarter; excluding it, operating economics were far stronger. The prior-year tax provision of $18.95 billion, an effective rate of 87%, also depressed the year-ago comparison; this quarter's tax provision was $3.14 billion at a 24% effective rate. Even so, the cash-flow line cannot be adjusted away. Purchases of property and equipment reached $33.9 billion in the quarter, and after principal payments on finance leases, capital expenditure approached $34.9 billion. Free cash flow swung to negative $2.12 billion from positive $10.6 billion a year earlier.
"We hit major milestones this quarter toward delivering personal superintelligence for everyone - including launching Muse and Muse Wonder, our first frontier model," said Mark Zuckerberg, Meta's founder and chief executive. "Our investments are paying off, the core business is strong, and we're seeing velocity across the board."
That is the crux of the Meta bull case: the $35 billion-a-quarter infrastructure bill is being funded by a core advertising engine that is still growing at 26%, not by debt or dilution. The company guided fourth-quarter revenue to $73 billion to $76 billion and narrowed its 2026 capital-expenditure outlook to $135 billion to $145 billion, while flagging full-year expenses of $175 billion to $179 billion after the legal charge. Management also said it expects Reality Labs losses to narrow to around $15 billion next year as investment shifts toward AI priorities, and separately disclosed that the board has approved a new non-voting Class C share class effective in January 2027.
The strategic subtext is a capital-allocation triage. Reality Labs, the metaverse bet that has accumulated more than $80 billion in operating losses since 2021, is being deprioritized in favor of AI infrastructure that already shows a return. The company's own numbers make the logic explicit: Advantage+, its AI-powered ad suite, reached a more than $75 billion annual revenue run-rate during the quarter. When an AI product line prints revenue at that scale inside the existing business, it is easier to defend the next $35 billion data-center check than it was to defend the metaverse at its peak.
Tesla: A Margin Beat Built on Cost Discipline, Not Demand
Tesla's report told a different version of the same story. Revenue of $28.05 billion came in essentially in line with the roughly $28.06 billion consensus and slightly below both the second quarter's $28.24 billion and the year-ago $28.1 billion. The beat lived in profitability. Non-GAAP earnings per share of $0.48 topped the consensus of about $0.45, helped by a total gross margin that rebounded to 19.6% from 16.8% in the second quarter and 18.0% a year earlier. Operating income of $718 million nearly doubled the prior quarter's $398 million, and free cash flow turned positive at $1.53 billion after a negative $1.09 billion in the second quarter.
The margin recovery is the number that matters most for a stock that has been punished for pricing pressure. But the composition of the beat matters too. Capital expenditure fell to $4.73 billion from $5.79 billion in the prior quarter, even as operating expenses rose 39% year over year to $4.78 billion. Research and development jumped 69% to $2.76 billion as the company channels spending into what it now calls "Super Intelligence," Optimus, and the Cybercab. Stock-based compensation more than doubled to $1.17 billion from $663 million, reflecting the 2025 CEO pay package.
In other words, Tesla bought its earnings beat partly through capex discipline and partly through accounting load that will weigh on GAAP results for quarters to come. GAAP net income of $790 million fell from $1.11 billion in the second quarter and $1.37 billion a year earlier. The company-compiled consensus on Tesla's own investor-relations page had called for average revenue of $28.34 billion, non-GAAP EPS of $0.45, GAAP EPS of $0.23, and free cash flow of negative $4.51 billion with capex of $7.77 billion. On the cash-flow line especially, the actual result was a clean positive surprise - positive $1.53 billion versus an expected negative $4.5 billion.
Volume, however, is not yet cooperating with the margin story. The company delivered 486,532 vehicles in the third quarter, up 1.3% sequentially but down 2.1% from the year-ago quarter, when a rush to capture the expiring U.S. federal EV tax credit pulled demand forward. Energy storage deployments of 13.7 gigawatt-hours rose 9.6% year over year but fell short of the roughly 15.9 gigawatt-hours analysts had expected. Tesla is earning better money on each car it sells, but it is not yet selling meaningfully more cars. That gap between unit growth and profit growth is the hinge on which the stock's next leg will turn.
The Mechanism: Why the Market Is Tolerating the AI Bill
Put the two reports side by side and the market's logic becomes visible. Investors are not rewarding AI spending in the abstract; they are rewarding spending that sits on top of a core business still generating growth and cash. Meta can justify $35 billion of quarterly capex because advertising revenue grew roughly $13.4 billion year over year in a single quarter. Tesla can justify its autonomy and robotics push because it restored a 19.6% gross margin and turned free cash flow positive while doing it.
The transmission channel runs through the discount rate applied to AI optionality. When the core business prints 26% revenue growth and double-digit margin, investors assign a low probability of distress to the moonshot bets layered on top. When the core business stalls, the same bets get re-rated as speculative. That is why Tesla's flat revenue mattered less than its margin print, and why Meta's negative free cash flow mattered less than its 26% ad growth. The market is pricing AI as an option on top of a cash machine, not as the cash machine itself.
This mechanism has a historical analog that is worth recalling. During the 2021 cloud buildout, Amazon Web Services funded a multi-year data-center expansion while the core e-commerce business generated the cash to pay for it; the stock was rewarded as long as AWS revenue growth stayed above 25% and operating cash flow held. When AWS growth decelerated into the mid-teens in 2022 and 2023, the same infrastructure spend was re-rated as a drag, and the multiple compressed even though the assets were identical. The lesson is that the market tolerates heavy capex only while the funding engine is visibly healthy. Meta today sits where Amazon did in 2021: strong core growth, heavy spend, and a patient tape.
This is where the cyclical and structural forces separate, and the distinction should be made explicitly. The margin rebound at Tesla and the tax-rate normalization at Meta are cyclical: they mean-revert as pricing competition returns and as one-time items drop out. Tesla's gross margin has oscillated between roughly 17% and 25% over the past three years; a return to the mid-18% range on renewed price competition would not be a regime change, it would be a reversion. Meta's 24% effective tax rate is not a permanent improvement over the 87% outlier of a year ago; it is a normalization.
The structural leg is the infrastructure buildout itself. Meta's $135 billion to $145 billion capex plan and Tesla's multi-year autonomy investment are regime shifts in how these companies allocate capital, and they will not self-correct until the assets are built. A cyclical wave and a structural shift are both present here, and they point in opposite directions across time horizons. In the short run, the cyclical margin recovery dominates the tape. In the long run, the structural question is whether the deployed compute and autonomy stack generate returns above the cost of capital. The risk is timing, not direction: if monetization lags the deployment curve, the cash-flow math turns hostile regardless of how strong the core business is today.
The Counter-Thesis: This Is a Capex Bubble Wearing a Quality Mask
The strongest argument against the bullish read is that both companies are being credited for discipline they do not fully control. Meta's negative $2.1 billion free cash flow is not a rounding error; it is the direction of travel. If advertising growth slows as the macro weakens, the company faces a brutal choice between cutting the AI buildout mid-stream or funding it externally. Bears point to the valuation embedded in the AI narrative: Meta's market capitalization of roughly $1.78 trillion and Tesla's $1.53 trillion already assume that the autonomy and model bets pay off handsomely.
The peer comparison cuts the same way. Across the mega-cap technology complex, capital expenditure has become a shared arms race rather than a differentiated strategy. When every major platform is spending double-digit percentages of revenue on AI infrastructure, the spend ceases to be a competitive advantage and becomes the price of admission. The company that blinks first loses the race; the company that does not blink risks destroying capital on assets with no clear monetization path. This is the classic capex-supercycle trap: individually rational decisions by each player produce collectively excessive capacity, and the survivors are those who can outlast the downturn, not those who built the most.
There is also a concentration risk baked into the day's action. Two companies controlled by Elon Musk reported within 24 hours of each other, with merger speculation linking them, and a third Musk company, SpaceX, setting the tone. When a single individual's capital-allocation decisions drive the marginal sentiment for nearly $3.5 trillion of combined market value, the market's diversification is thinner than the index weights suggest. A change in Musk's priorities - a larger share of Tesla's balance sheet redirected toward a SpaceX combination, or a shift in Meta's AI timeline - would move both stocks in the same direction at the same time.
The falsifying signal for the constructive read is specific and observable: if Meta's 2027 capex guidance exceeds $180 billion while advertising growth decelerates below 15% year over year, or if Tesla's gross margin falls back below 17% while revenue remains flat for two consecutive quarters, the "quality inside the AI spend" thesis breaks. Either outcome would show that the infrastructure bill is growing faster than the engine that pays for it.
The Second-Order Consequence: The Burden Shifts to the Suppliers
The underappreciated implication of Wednesday's prints is where the risk migrates next. If Meta and Tesla have earned another quarter of patience, the semiconductor and data-center equipment complex that supplies them gets a reprieve as well. Chip makers, advanced packaging firms, and power-infrastructure providers have been priced for a continuation of the buildout; a pause in hyperscaler spending would hit them harder, on a percentage basis, than it would hit the hyperscalers themselves. The second-order trade is therefore asymmetric: the upside from one more quarter of capex growth is already embedded in supplier multiples, while the downside from a cut is not.
That asymmetry is why the next two earnings seasons matter more than this one. Wednesday's reports bought time, but they did not answer the terminal question. The terminal question is not whether AI infrastructure is being built - it clearly is. It is whether the revenue generated per dollar of deployed compute is rising fast enough to justify the next round of spending. Until that ratio is visible in reported earnings rather than in management commentary, the market will remain in the position it occupied on Wednesday: willing to believe, but requiring quarterly proof.
What Comes Next
The immediate catalyst is Thursday's regular-session reaction, the first full trading day for both stocks after the after-hours prints. Meta's guidance envelope - fourth-quarter revenue of $73 billion to $76 billion and 2026 expenses of $175 billion to $179 billion - sets a high bar that the stock must now clear on cash-flow discipline. Tesla's bar is different: investors will watch whether the 19.6% gross margin holds into the fourth quarter and whether capital expenditure stays near the $4.7 billion level rather than re-accelerating toward the $7.8 billion consensus forecast.
Beyond the single names, the rest of the Magnificent Seven earnings queue will determine whether Wednesday's rebound generalizes. The base case is that the market grants AI spenders another quarter of patience as long as core-business growth stays above 20% and free cash flow does not deteriorate further. The upside case is that Meta's 26% revenue growth and Tesla's margin recovery pull the entire complex higher, with chip suppliers and cloud infrastructure names re-rating on evidence that demand is real. The downside case is that investors begin demanding proof of AI monetization rather than proof of deployment, which would compress multiples across the semiconductor and data-center equipment complex.
Short term, sentiment is positive and the tape confirmed it. Medium term, the burden of proof has shifted from the revenue line to the cash-flow line. Long term, the question is whether today's infrastructure spending becomes tomorrow's earnings power or tomorrow's impairment. The answer will not arrive in one quarter; it will arrive in the quarterly gap between capital expenditure and the revenue that capital expenditure eventually generates.
The market spent Wednesday deciding that quality operators can carry an AI bill. The next few quarters will decide whether the bill was an investment or an anchor.
Explore more exclusive insights at nextfin.ai.
