30 October 2026 | Information cutoff: 06:50 GMT
Morning View
The overnight tape sharpened the distinction between AI demand and AI economics. Amazon Web Services accelerated to 39% growth and Amazon projected growth above 40% in both 2027 and 2028, yet the shares fell because a roughly $220bn 2026 capital program has pushed trailing free cash flow deeply negative. Intel delivered a much stronger quarter, but a manufacturing interruption, server-chip constraints and a weak fourth-quarter outlook exposed how little execution tolerance remains when fixed costs and capital needs are high.
The strongest read-through sits one layer below the cloud platforms. Monolithic Power Systems more than doubled enterprise-data revenue, Sandisk reported extraordinary NAND profitability and Entegris set a revenue record as leading-edge process complexity increased materials intensity. The demand signal across compute, memory, power and process materials is still constructive; the market is increasingly asking who can convert it into cash rather than merely backlog.
The morning is mixed for technology fundamentals and negative for indiscriminate positioning. Communications software and digital advertising are producing real operating leverage, but IT-services bookings remain uneven and inflation has put a December rate increase back into the market. The better setup is in companies with scarce content, pricing power or asset-light cash conversion; the more fragile setup is in capital-intensive infrastructure where financing, permitting or manufacturing execution can break an otherwise strong demand case.
1. Amazon’s cloud acceleration is real; the cash-flow reset is equally real
Amazon reported third-quarter sales of $201.9bn, up 12%, with Amazon Web Services revenue of $46.0bn, up 39.3% and the fastest growth rate in roughly five years. AWS operating income reached $17.8bn, while group operating income rose to $28.1bn and a record 13.9% margin. The tension is below the income statement: trailing operating cash flow increased 29% to $168.3bn, but trailing free cash flow swung to an outflow of $42.4bn as property and equipment purchases rose by $94.8bn, principally for AI infrastructure. Management retained roughly $220bn of 2026 cash capital expenditure and said AWS growth should exceed 40% in 2027 and 2028. The fourth-quarter sales midpoint was slightly below prevailing expectations and the shares fell about 7% after hours. The bull case is that accelerated capacity converts a visible demand backlog into durable cloud share gains, with retail efficiency funding the build; the bear case is that depreciation, power and financing costs arrive before utilization and pricing. Nvidia, Broadcom, Marvell Technology, Micron Technology, Arista Networks, Vertiv, Eaton and data-center landlords receive positive demand read-throughs, while Amazon’s valuation now depends more heavily on evidence that capital intensity peaks. The next test is fourth-quarter AWS growth, the cadence of cash capex and whether new agreements with OpenAI, Meta Platforms, Nvidia and other large customers translate into contracted revenue rather than additional speculative capacity. Sources: Amazon results; Reuters.
2. Intel’s quarter beat, but its guide turned the debate back to execution and capital intensity
Intel’s third-quarter revenue increased 21% to $16.5bn and non-GAAP earnings reached $0.52 per share, with Data Center and AI revenue up 78% to $7.3bn and Intel Foundry revenue up 58% to $6.7bn before $6.3bn of intersegment eliminations. That strength was overwhelmed by fourth-quarter guidance of $15.0bn–$16.0bn of revenue and $0.25 of non-GAAP earnings, materially below market expectations; the shares fell roughly 23% after hours. A September explosion at the Kiryat Gat factory is expected to reduce revenue by about $600m over two quarters, including roughly $300m in the fourth quarter, and cut fourth-quarter gross margin by approximately 200 basis points. Management also cited server-chip supply constraints, weak personal-computer demand into the first half of 2027, no positive cash flow this year and higher manufacturing investment next year. The prior narrative was that sovereign support, foundry milestones and AI-server demand had created a cleaner recovery path; the print instead shows demand being lost because supply and manufacturing execution cannot keep pace. Bulls can point to the 14A 0.9 process-design kit, a completed $23bn equity raise and unusually strong data-center demand. Bears will argue that dilution, rising capital requirements and a still largely internal foundry revenue base defer any proof of economic returns. Advanced Micro Devices gains an opening in server CPUs, while semiconductor-equipment and materials suppliers retain long-duration exposure to Intel’s build. The decisive catalysts are Kiryat Gat recovery, external 14A customers, gross-margin normalization and evidence that 2027 spending produces competitive yield rather than another cash drain. Sources: Intel results; Reuters.
3. Monolithic Power Systems shows that AI power content is compounding faster than accelerator volumes
Monolithic Power Systems reported third-quarter revenue of $1.202bn, up 63.1%, with enterprise-data revenue of $506.7m up 164.6% and communications revenue of $187.4m up 134.7%. Non-GAAP gross margin was 55.7%, operating margin 38.6% and earnings $8.19 per share, demonstrating that the current growth is carrying exceptional incremental profitability rather than merely buying share. Management said installed capacity can support $6bn of annual revenue today, $8bn by the second half of 2027 and a path toward $10bn, while enterprise-data demand should remain strong into 2027. The market had already accepted rising power content per accelerator, but the magnitude suggests the opportunity is broadening across racks, optical switching and higher-voltage architectures. The bull case is that power-management content grows faster than accelerator unit shipments as rack density and conversion complexity increase; the bear case is customer concentration, aggressive capacity ahead of demand and eventual insourcing by large silicon or systems customers. The read-through is favorable for Analog Devices, Texas Instruments, Infineon, Navitas Semiconductor, Vicor, optical-component vendors and cooling suppliers, but it also raises the bar for companies lacking high-voltage or data-center exposure. Investors should monitor fourth-quarter enterprise-data growth, gross margin as new capacity comes online and evidence of design diversity across accelerator, networking and custom-silicon platforms. Source: Monolithic Power Systems results.
4. Sandisk’s historic NAND margins make scarcity the earnings story—and the peak-cycle risk
Sandisk reported fiscal first-quarter revenue of $11.031bn, up 23% sequentially, non-GAAP earnings of $49.51 per share before its announced stock split and gross margin of approximately 85%. For the December quarter, management guided to revenue of $12.2bn–$12.6bn, non-GAAP gross margin of 84.5%–85.5% and earnings of $55–$57 per share, while authorizing $5bn of repurchases and an eight-for-one split. Data-center demand again exceeded supply, showing that AI infrastructure is transmitting into flash pricing and mix as well as DRAM and high-bandwidth memory. The result moves the debate beyond a routine memory recovery: current margins imply scarcity economics that will drive substantial estimate increases, but they also represent a level unlikely to be treated as a normalized terminal margin. Bulls see enterprise solid-state drives, inference storage and disciplined industry supply creating a structurally higher earnings floor; bears see peak pricing that will invite capacity and customer inventory substitution. Micron Technology, Samsung Electronics, SK Hynix and Kioxia receive positive near-term pricing read-throughs, while server and storage vendors face higher bill-of-materials costs. The next evidence is contract pricing, capital-expenditure discipline across the NAND industry and whether hyperscaler demand remains supply-constrained after the December quarter. Source: Sandisk results.
5. Nscale’s reported financing strain shows why private AI backlog is not the same as funded capacity
The Financial Times reported that Nvidia-backed Nscale’s planned West Virginia Monarch campus, marketed at up to 8GW, is contending with construction problems, regulatory scrutiny, local opposition and stop-work orders related to flooding and stormwater. The report said Anthropic has agreed to pay $45bn for capacity from the first building, representing more than 40% of Nscale’s reported $103bn backlog, but financing for that facility has not been secured and Anthropic may withdraw if a funding deadline is missed. Nscale is also reported to be targeting a public listing that could value it near $35bn. These figures and terms remain reported rather than company-confirmed, but the investor implication is material: a large backlog can reflect conditional contracts attached to assets that are neither financed nor permitted. The prevailing narrative has treated hyperscaler and frontier-lab commitments as near-equivalent to bankable revenue; the Monarch case exposes concentration, completion and counterparty-option risk. Bulls will argue that scarce powered land and a marquee anchor customer can unlock project finance. Bears will argue that a single customer’s contractual flexibility, environmental enforcement and rising capital costs can impair both revenue visibility and equity value. CoreWeave, Crusoe, Lambda, Applied Digital, Nebius, private-credit lenders and infrastructure funds face the clearest read-through. The next catalyst is committed construction financing, resolution of stop-work orders and disclosure of termination rights before any listing proceeds. Source: Financial Times.
6. Twilio’s beat-and-raise was not enough to offset a visible fourth-quarter deceleration
Twilio increased third-quarter revenue 20% to $1.56bn, including 16% organic growth, while dollar-based net expansion improved to 114% from 109%. Non-GAAP operating income reached $326m at a 21% margin and free cash flow was $370m, a 24% margin. Management raised full-year guidance to 20% reported revenue growth, 15% organic growth and $1.19bn–$1.20bn of free cash flow, yet the shares fell about 15% after hours because the fourth-quarter organic guide of 11%–12% implies a material deceleration. The quarter supports the thesis that communications usage, including AI-driven messaging and voice workflows, can reaccelerate while cost discipline converts growth into cash. The reaction shows that investors had already capitalized much of that recovery and are unwilling to overlook the exit rate. Bulls see improving net retention, platform consolidation and high incremental margins; bears see normalization after favorable comparisons, carrier-fee distortions and competition from cloud communications and in-house tooling. Bandwidth, Sinch, RingCentral, Five9 and contact-center software vendors are exposed, while hyperscalers benefit if conversational applications increase communications and model consumption together. The key datapoints are fourth-quarter organic growth, 2027 guidance, net-retention durability and whether cash-flow margins hold through renewed investment. Sources: Twilio results; after-hours market reaction.
7. Reddit is converting differentiated human data into software-like margins
Reddit’s third-quarter revenue rose 53% to $895m, with advertising revenue up 54% to $847m and other revenue up 35% to $48m. Daily active uniques increased 12% to 130.3m and weekly active uniques 16% to 515.2m. Net income was $283m, adjusted EBITDA margin 44.9% and free-cash-flow margin 40.6%, while capital expenditure was only $1m. Fourth-quarter guidance of $1.00bn–$1.03bn of revenue and $480m–$510m of adjusted EBITDA reinforces the view that Reddit can monetize discovery and commercial intent without the infrastructure burden borne by model developers. The surprise is not only growth but the combination of growth and cash conversion. Bulls see an under-monetized, difficult-to-replicate corpus of current human discussion that gains value as search shifts toward AI answers; bears note that advertising still supplies nearly all revenue, user growth is slower than revenue growth and licensing remains small. Google, OpenAI and other model companies need authentic human data, while Pinterest, Meta Platforms and Snap compete for performance-ad budgets. The next catalysts are international monetization, licensing mix, traffic dependence on search and whether AI answer engines send sufficient referral traffic back to Reddit. Source: Reddit results.
8. Cognizant’s AI productivity is visible in margin, but not yet in bookings
Cognizant reported third-quarter revenue of $5.7bn, up 4.6% and 4.8% in constant currency, with adjusted operating margin up 30 basis points to 16.3% and adjusted earnings up 7.2% to $1.49 per share. Trailing-twelve-month bookings increased 4% to $28.7bn and book-to-bill was 1.3, but third-quarter bookings declined 7%. Management revised full-year constant-currency growth to 4.0%–4.4%, while Project Leap is expected to cost $180m–$250m and generate roughly $200m of annual savings; it also said half of code is now AI-assisted and 95,000 employees have earned AI badges. The result supports an internal-productivity case more clearly than a demand-acceleration case. Bulls see financial-services strength, better execution and AI lowering delivery cost without impairing growth. Bears see weak quarterly bookings, restructuring risk and customer efforts to use AI to reduce labor-based contract volumes. Accenture, Infosys, Tata Consultancy Services, Capgemini and Globant face similar pressure to turn AI productivity into margin while defending revenue. The next datapoints are fourth-quarter bookings, renewal scope, headcount and whether savings are retained or passed through to clients. Source: Cognizant results.
9. Entegris broadens the AI semiconductor trade beyond equipment into process materials
Entegris reported record third-quarter sales of $1.006bn and non-GAAP earnings of $1.23 per share, ahead of its guidance, then guided fourth-quarter revenue to $1.025bn–$1.055bn, earnings to $1.32–$1.40 and adjusted EBITDA margin to 30.7%–31.2%. The print confirms that advanced-node complexity, stricter contamination control and greater materials intensity are producing growth outside the most visible lithography and deposition suppliers. The market had expected leading-edge logic, high-bandwidth memory and advanced packaging to support demand; the upside suggests that consumables and specialty materials can compound even when wafer-fab equipment orders are uneven. Bulls see recurring process content, high qualification barriers and exposure to every incremental layer. Bears see customer concentration, China restrictions, leverage and the risk that an equipment pause eventually reaches consumables. Applied Materials, Lam Research, KLA, ASML, MKS Instruments and specialty-chemical suppliers receive constructive read-throughs, while foundries and memory producers absorb more process cost. The next catalysts are fourth-quarter margin delivery, leading-edge fab utilization and any change in export controls or customer inventory. Source: Entegris results.
10. Hotter inflation raises the hurdle rate for the most capital-intensive AI equities
The Bureau of Economic Analysis reported September personal-consumption-expenditure inflation of 0.5% month over month and 3.6% year over year, while core inflation was 0.3% and 3.1%. Third-quarter gross domestic product grew at a 3.4% annualized rate. The combination of resilient activity and sticky inflation pushed market pricing toward a December rate increase, with one widely cited measure near 85%. This matters for technology because AI investment is simultaneously supporting economic growth and increasing the discount rate applied to the assets financing that build. Hyperscalers with large operating cash flows can continue spending, but neoclouds, data-center developers, power projects and long-duration software valuations face a higher cost of capital. Bulls will argue that nominal growth, productivity and AI demand support revenue enough to absorb a modestly higher rate path. Bears will argue that leveraged infrastructure economics and terminal-value-heavy software multiples have not fully adjusted. CoreWeave, Nebius, Applied Digital, Equinix, Digital Realty, utilities and private data-center developers are most exposed, while cash-rich Microsoft, Alphabet, Amazon and Meta Platforms have relative funding advantages. The next tests are the Federal Reserve’s December communication, labor data and whether long yields rise enough to alter project finance or hyperscaler hurdle rates. Sources: Bureau of Economic Analysis; Reuters.
What to Watch
Over the next seven days, focus on fourth-quarter guidance revisions following Amazon and Intel; any committed financing or regulatory resolution for Nscale’s West Virginia project; NAND contract-price and supply commentary after Sandisk’s results; the first evidence that a December Federal Reserve rate increase is changing data-center financing terms; and 2027 growth and capital-intensity commentary from the next wave of cloud, semiconductor and infrastructure-software earnings.
Bottom line
AI demand remains stronger than the bearish macro narrative, but the market is withdrawing its blanket premium for capacity announcements and revenue backlog. Amazon, Monolithic Power Systems, Sandisk, Reddit and Entegris show that growth and operating leverage can coexist; Intel and Nscale show that manufacturing, permitting and financing can prevent demand from becoming cash flow. The next leg of sector performance should be driven less by exposure to AI in the abstract and more by evidence of funded demand, scarce content or components, and credible returns on capital.