Good morning. Fast growth is meeting a much tougher test. Amazon’s cloud is accelerating, Intel’s server sales are surging, and Reddit’s ads are growing. Cash demands, PC budgets and search traffic explain why the headlines are not enough.

Four deeper earnings stories lead today, followed by Coinbase and fresh Exxon results. Premarket changes compare with Thursday’s closes. Quotes are timestamped individually. The scoreboard separately shows Thursday’s regular-session closes.

Amazon’s cloud accelerates. Its cash deficit widens.

Archive image via Forbes Middle East.

AMZN: $225.93 · −7.33% premarket. 8:30 a.m. ET.

Amazon is demonstrating AI demand faster than it is funding the buildout from operating cash. AWS revenue grew 39.3% to $46.0 billion in Q3. Amazon’s overall operating margin reached a record 13.9%.

The cash comparison covers the trailing twelve months, not this quarter alone: operating cash flow rose 29% to $168.3 billion, while net purchases of property and equipment reached $210.7 billion. That left $42.4 billion of negative free cash flow, versus a $14.8 billion surplus a year earlier.

There is substantial operating progress behind the bill. AWS produced $17.8 billion of quarterly operating profit, up from $11.4 billion. Across Amazon, $28.1 billion of operating income exceeded the $26.5 billion top of its July guidance. Revenue of $201.9 billion landed near the upper end of the $197–$202 billion range.

Retail needs a calendar adjustment before declaring a demand collapse. Prime Day moved into Q2 in most large countries; Amazon says comparable Q3 sales growth would have been roughly four percentage points higher without that shift. The event changed quarters, rather than disappearing from customers’ shopping calendars.

On the call, management argued that customers prefer running AI inference close to their existing applications and data. That could give AWS an advantage over simply renting GPUs. It remains management’s explanation of the opportunity, rather than proof that every new data center will earn an attractive return.

Our reading: the quarter strengthens the demand case while increasing the financing burden. Cash spent on capacity arrives before the revenue that capacity may generate, leaving less internally generated money for other uses. The decisive improvement would be cash generation catching up with investment while cloud margins hold—not another revenue record by itself.

Intel’s server recovery meets a PC squeeze.

Coolcaesar / Wikimedia. Santa Clara, 2023. · Creative Commons license

INTC: $69.47 · −24.55% premarket. 8:30 a.m. ET.

Intel’s recovery is becoming more dependent on servers while its PC customers retreat. Q3 revenue rose 21% to $16.5 billion. Datacenter and AI sales jumped 78% to $7.3 billion, but client sales fell 5% to $8.1 billion. Those are very different businesses hiding inside one growth headline.

The quarter delivered against Intel’s own prior promises. Revenue finished inside July’s $15.8–$16.8 billion range, while adjusted EPS of $0.52 exceeded the $0.38 guide. That makes the latest outlook more revealing than calling the entire report a failure: Q4 revenue is forecast at $15–$16 billion, with adjusted EPS of $0.25.

At the midpoint, the revenue outlook implies roughly a 6% sequential decline. Management expects the PC slowdown to continue at least through the first half of 2027. On the call, it linked the pressure to memory prices, limited customer IT budgets and manufacturers reducing inventory to match weaker sales.

The AI investment cycle can help one part of Intel and hurt another. Infrastructure spending creates demand for server CPUs. Expensive memory raises the cost of a finished PC, leaving less room in a fixed purchasing budget. Product mix and Intel’s ability to supply processors determine the net effect.

Manufacturing economics also remain unfinished. Foundry’s operating loss narrowed to $1.85 billion from $2.32 billion a year earlier, which is real progress. But its reported revenue includes substantial internal business eliminated on consolidation. It cannot be read as $6.7 billion of outside customers validating the foundry strategy.

Stronger server demand and a smaller manufacturing loss support the recovery case. Sustainable earnings still require delivered server supply, improving foundry economics and a stabilizing PC outlook. Customer discussions and process-roadmap milestones do not, by themselves, establish profitable outside orders.

Reddit grows ads 54%. Search traffic raises a warning.

Reddit office photograph, via 01net. 2024 archive.

RDDT: $100.80 · −22.22% premarket. 8:29 a.m. ET.

Reddit’s monetization improved while its path to acquiring and re-engaging users became less dependable. Q3 revenue rose 53% to $895 million; advertising contributed $847 million, up 54%. But management warned that continuing search-referral declines could push Q4 daily active uniques down by a low-double-digit percentage from the current 130 million base.

That is a conditional warning, not a decline already recorded for the whole quarter. In October, referrals fell by more than one million on average each week. The largest impact was on logged-out web users, although search also brings existing users back to the app and logged-in experience.

The financial results otherwise show considerable strength. Revenue exceeded the $860–$870 million range issued in July. Adjusted EBITDA reached $402 million, about 45% of revenue, and operating cash flow was $364 million. Q4 revenue guidance is $1.00–$1.03 billion: a potentially larger business serving a potentially smaller daily audience.

The strongest counterevidence is the composition of that audience. Logged-in and app daily users grew sequentially; together, management says they generate more than 80% of monetization. Losing a casual search visitor does not have the same immediate revenue impact as losing a regular user whom advertisers can reach more effectively.

However, casual visitors are also a source of tomorrow’s regulars. Protecting current ad revenue does not make a shrinking discovery channel irrelevant. Reddit is investing in marketing, app conversion and product improvements, and plans a logged-in-only test in selected markets outside the U.S. during Q4. That is a test, not a completed global restriction.

The investment question has shifted from monetizing the community to renewing it as search changes. Direct visits and retained app users could reduce referral dependence; buying increasingly expensive traffic could weaken the economics. Sustained direct-user growth would provide more convincing evidence that Reddit controls its distribution.

Sandisk’s datacenter sales jump 61%. The split is secondary.

Sandisk / Wikimedia. Headquarters archive. · Creative Commons license

SNDK: $1,207.00 · −9.25% premarket. 8:30 a.m. ET.

Sandisk’s growth is increasingly coming from datacenter storage rather than consumer devices. Fiscal Q1 revenue rose 23% sequentially to $11.03 billion. Datacenter sales increased 61% to $4.80 billion, while consumer revenue fell 29% to $397 million. The company’s AI exposure has become much more visible in its actual sales mix.

Datacenter supplied about 44% of total revenue, versus 33% in the preceding quarter. Its $1.83 billion sequential increase accounts for roughly 88% of the company’s net revenue growth. That calculation identifies where the improvement occurred; it does not establish how much came from higher prices, additional units or a richer product mix.

The results cleared management’s own August outlook: revenue exceeded the $10.3–$10.8 billion range, and adjusted EPS of $49.51 beat the $44–$46 guide. Non-GAAP gross margin reached 85.1%. Next quarter’s revenue range is $12.2–$12.6 billion, with adjusted EPS of $55–$57 on the existing share basis.

On the call, management described persistent storage as increasingly important to AI inference: information can be retained and recalled instead of repeatedly regenerated. That supports a role for NAND alongside other memory technologies. It does not make NAND interchangeable with DRAM or the high-bandwidth memory attached to accelerators.

The company also announced an eight-for-one split. Extra shares are due after the close on November 19, with split-adjusted trading expected November 20. All earnings-per-share figures above are pre-split. Dividing each share into eight changes the unit price and share count, not the company’s total value or an investor’s ownership percentage.

The harder question is how much margin survives future supply additions and customer bargaining. Buyers respond to expensive storage, and suppliers have incentives to expand. Durable customer commitments and cash conversion would strengthen the case for lasting improvement. The split cannot answer a question about the cycle.

The Mag 7 + three: Thursday’s close.

October 29 versus October 28, 2026. Adjusted regular-session price changes, excluding dividends. C&C editorial watchlist, not an official index. These closes precede Amazon’s Thursday-evening earnings; its premarket quote appears above.

Coinbase’s $719 million profit comes with a large asterisk.

Coinbase signage, archive via Barron’s.

COIN: $150.75 · −6.13% premarket. 8:28 a.m. ET.

Coinbase returned to profit, but the headline overstates the improvement in its operating business. Q3 net income was $718.7 million, including a $715 million pre-tax gain on crypto assets held for investment. Operating income was $61.7 million; company-adjusted net income was $77.7 million.

Total revenue fell 35% year over year to $1.216 billion. Transaction revenue declined 44% to $590 million, while subscription and services revenue dropped 20% to $570 million. Sequentially, the latter rose 3% and transaction revenue slipped 1%, showing stabilization rather than a return to last year’s activity levels.

There is a genuine improvement beneath the investment gains: adjusted EBITDA rose 60% from Q2 to $333 million, helped by lower expenses. It remained 58% below the year-earlier quarter. Cost discipline improved the earnings base even though customer trading had not recovered comparably.

Q4 subscription-and-services guidance is $550–$630 million. USDC balances and newer transaction products can diversify revenue, but repeat customer fees are economically different from appreciated crypto holdings. The useful test is whether fee income and operating profit keep improving without another large investment gain.

Exxon produces $19 billion of quarterly free cash flow.

zodebala / iStock, via 10/12 Industry Report. Archive.

XOM: $186.74 · +1.88% premarket. 8:25 a.m. ET.

Exxon’s fresh Friday results provide a different picture of investment and cash returns. Q3 operating cash flow was $25.7 billion and free cash flow was $19.1 billion. Reported earnings were $15.4 billion; adjusted earnings were $17.5 billion. Those are three-month figures, unlike Amazon’s twelve-month cash comparison above.

The company distributed $9.4 billion through dividends and repurchases during the quarter—roughly half its free cash flow. It raised the quarterly dividend to $1.07 per share and announced $20 billion of planned 2027 buybacks, assuming reasonable market conditions. The plan describes future purchases, not cash already returned.

Exxon points to structural cost reductions and stronger refining operations as evidence of better execution. Favorable commodity prices and refining margins still matter, so one quarter’s cash generation should not be treated as a permanent annual run rate.

The distinction for shareholders is practical: the current cash surplus gives Exxon room to invest, reduce financial obligations or distribute money without relying entirely on new funding. Maintaining that flexibility through weaker energy markets would provide stronger evidence of business improvement than the payout announcement alone.

Labor costs ease slightly. Wage buying power falls.

Compensation costs rose 0.8% in Q3, after 0.9% in Q2. The 8:30 a.m. ET BLS release puts annual compensation growth at 3.4%. Civilian wages rose 3.1% over the year, but inflation-adjusted wages fell 0.6%. Employers face slower quarterly cost growth while workers still lose purchasing power—a difficult combination for consumer-facing companies.

Analysts: a rating cut versus target cuts.

KeyBanc downgraded Reddit to Sector Weight in its October 30 note. It also lowered price targets on Intel to $140 and Amazon to $345. Those target revisions are separate from a rating downgrade. See the dated reporting; targets are analyst opinions, not forecasts adopted by C&C.

Forwarded this email? Get Cash & Chaos in your inbox.

Independent market commentary. Forecasts and analyst opinions are not guarantees; C&C interpretations are analysis.