TMT Breakout晨报复盘INTU、SMTC、ZM财报并汇总多条科技研究消息

Good morning. Futures slightly down as tech investors have all eyes on **NVDA** tonight (implied move 5%). **Crude -4%** while yields are flat following headlines Iranian Foreign Minister as Omani counterpart discussed “interim framework” aimed at helping ships pass through SoH.

Asia mainly up overnight: TPX +0.42%, NKY +0.62%, Hang Seng +0.56%, HSCEI +1.05%, SHCOMP +0.59%, Shenzhen +0.5%, Taiwan TAIEX +1.47%, Korea KOSPI +0.97%. **Softbank +2%; Hynix +3%; Samsung +2%**

In the US, **TMT Mo +1%** while **Software -1%** leading the way lower after **INTU’s** lowered topline guide fed into bear fears. **ZM -7%** was also mixed while **SMTC +5%** the bright spot on a nice DC driven beat and raise. We’ll cover those first, then move onto the usual.

Let’s get straight to it…


# **EARNINGS**

### **INTU: Clean Q4, but a deeper-than-feared TurboTax reset and no reacceleration timeline leave the stock narrative in “show me” mode and keeps bears in control**

FQ4 revenue, EPS and margins came in above Street, but FY27 revenue and especially TurboTax guidance were materially below expectations as INTU lowers DIY pricing and reinvests to rebuild the low-end customer funnel. Lots of fuel for bears: INTU formally acknowledged that customer acquisition is weak in both major franchises, identified price as the leading reason for TurboTax churn, reduced its medium-term growth frameworks and declined to provide a timetable for reacceleration. Add on AI disintermediation fears, a somewhat aggressive segment guide, and mgmt credibility taking another hit and this was a better day for bears. Wasn’t a favored long going in, now a less favored long leaving despite stock being cheap as these #s should help continue the INTU re-rate from premium compounder to a low-multiple secularly challenged stock. We’ll see how stock trades today — other “AI pressured” names have bounced decently after misses

**FQ4 #s**

**Revenue:** $4.354B, +13.7% y/y (last q +10.4% y/y) vs. Street $4.273B, +11.5%
**EPS:** $4.03, +46.5% y/y vs. Street $3.58
**Non-GAAP OM:** 33.3% vs. Street 30.7%

**FQ1’27 Guidance**

Revenue Guide: $4.294–4.313B, midpoint $4.304B, +10.8% y/y vs. Street $4.373B, +12.5%
Non-GAAP EPS Guide: $2.44–2.48, midpoint $2.46, +30–33% y/y on the recast SBC-inclusive basis. The legacy ex-SBC equivalent is $3.92–3.96 vs. Street $4.04.

**FY27 revenue** was guided to $23.279–23.512B, midpoint $23.396B, +9.1% y/y vs. Street $23.822B, +11.5%, while TurboTax growth of 2–3% was well below Street at 6.8% and below the roughly 3–4% investor expectation cited in the materials.
Non-GAAP Operating Income Guide**:** $902–915M, midpoint $909M, with 21.0–21.2% OM under the new definition. The legacy Street comparison was $1.480B and 33.8%; after adjusting Street for SBC, the guide is approximately 3% below Street.

The reset reflects weak new-customer formation, with online paying customers up only 3%, and lost DIY tax customers as price became the No. 1 reason for churn. INTU is deliberately sacrificing initial ARPC to rebuild volume and monetize customers later through assisted tax, Credit Karma and platform services.

The positives bulls will try to hang on to: Assisted Tax, Money and Mid-Market collectively grew 34% and reached 30% of revenue, Q4 profitability and cash flow were strong, and the FY27 profit guide was above Street after adjusting for the new treatment of SBC.

### **Key Takeaways:**

-

**Reset.** This was a genuine resetting of the growth algorithm rather than a routine conservative guide. FY27 revenue growth falls to 9–10% from 14% in FY26; GBS is guided to 13–14%; TurboTax falls to 2–3%; and the three-year frameworks were reduced to 10–15% for GBS from 15–20% previously and 4–8% for Consumer from 6–10%. The central question is no longer whether INTU can sustain mid-teens growth through price, mix and upsell. It is whether INTU can restart volume growth without creating a lasting ARPC and margin problem.

-

**Tax.** Price is now the leading reason customers leave TurboTax, and INTU acknowledged losing quality DIY customers to lower-cost providers. FY27 TurboTax growth of 2–3% compares with Street at 6.8% and an investor expectation of roughly 3–4%. TurboTax Live is expected to slow to the mid-teens from 37% in FY26, below the 20%+ investor bogey cited in the materials, because more than three-quarters of FY26 Live additions came from upgrades out of the existing DIY base rather than direct competitive takeaways. Mgmt did not quantify the DIY ARPC decline

-

**Economics.** The strategic argument is that lower DIY pricing should not be judged solely on tax revenue. Customers using both TurboTax and Credit Karma generate approximately 2x the revenue of single-product customers, while Credit Karma members filing through TurboTax increased more than 50%. DIY also remains the primary feeder into Live. The bull case therefore requires INTU to prove that lower acquisition ARPC produces better lifetime value through assisted tax, financial products and year-round engagement.

-

**Customers.** Total online paying customers reached 8.9M, up only 3%, roughly two points slower than FY25; U.S. QBO customers excluding self-employed grew 6%. INTU attributed the slowdown largely to its own prioritization of mid-market, Money and AI platform development over broad-based customer acquisition. QuickBooks Free and Lite are intended to reopen the funnel, with more than 20,000 customers already active or converted to paid offerings and early monetization coming through payments and upgrades.

-

**Timing.** Mgmt declined to say whether FY27 is definitively the growth trough or whether reacceleration could slip into FY28. Instead, it asked investors to judge progress quarter by quarter. This is important because the reset does not yet establish a clear earnings inflection date; the FY27 tax season and evidence of improving online customer additions are likely to remain the key proof points.

-

**Mid-market.** The upmarket thesis remains intact. FY26 mid-market revenue grew 39%, customers grew 28%, new-to-the-franchise customers grew more than 30%, and Intuit Enterprise Suite annualized revenue exceeded $145M, roughly 4x last year. The caveat is that approximately three-quarters of customer additions came from upgrades or desktop migrations. That proves product-market fit and cross-sell strength, but also highlights why the low-end funnel must be replenished.

-

**AI.** AI usage is producing tangible product outcomes: millions of customers are using AI-native experiences, customers are getting paid four days faster and reducing manual work by 30%, and more than 75% of Enterprise Suite customers use AI agents monthly. Mgmt did not characterize AI as the direct cause of the reset and argued that INTU will be the disruptor, not the disrupted. The sell-side interpretation is divided, however. This will feed bears view of low-cost AI-enabled tax alternatives as an emerging share threat, while the callback explicitly attributed TurboTax losses primarily to FreeTaxUSA and lower-end price competition rather than AI.

-

**Money.** Money remains one of the cleanest growth engines. Q4 online payment volume grew 32% including Bill Pay and 21% excluding it, while QuickBooks Capital loan volume increased 54% to $1.9B. Capital revenue will face a yield headwind as INTU sells a larger share of loans to forward-flow partners, but this also reduces balance-sheet risk. Q4 additionally benefited from a one-time launch service for a new partner, so not all of the quarter’s Money momentum should be extrapolated.

-

**Margins.** Q4 adjusted OM of 33.3% was 260bps above Street. FY27’s new SBC-inclusive non-GAAP OM is approximately 34.6%; on the legacy ex-SBC basis, guidance implies 43.3% versus Street at 41.8%. The apparent profit upside is therefore real, but definition changes make headline comparisons noisy. Workforce savings and efficiency gains are funding increased sales, marketing and product investment, and mgmt continues to target company-level margin expansion and at least high-teens annual EPS growth.

-

**Mailchimp.** Mailchimp revenue declined slightly in Q4 and is guided flat to down 1% in FY27, with higher effective pricing offset by churn. It will become a separate reportable segment, improving visibility but also separating a structurally weak asset from the healthier GBS franchise. Mgmt said it is focused on maximizing Mailchimp’s value and profitability, although no divestiture or other strategic action was announced.

-

**Capital.** Q4 operating cash flow was $1.331B versus Street at roughly $624M. INTU repurchased $2.1B of stock in Q4 and $5.5B in FY26, reducing weighted-average diluted shares by 2%; $7.9B remained authorized. Cash and investments were $7.2B against $7.7B of debt, and INTU issued $1.75B of senior notes to address FY27 maturities. Buybacks are expected to remain a significant part of the EPS algorithm.

-

**Callback.** The most important callback message was that both funnel problems were viewed as correctable execution issues. In QuickBooks, INTU said it overemphasized mid-market, Money and AI investments and underemphasized the broader customer base. In tax, the company identified FreeTaxUSA and low-end pricing, rather than AI disruption, as the primary source of losses. That supports the “self-help reset” thesis, but it also confirms that the weakness was not merely conservative guidance.

### **Bull vs. Bear Debate**

**The broader bull thesis** is that INTU still owns two highly valuable financial operating systems with brand, data and distribution advantages that lower-priced point solutions cannot easily replicate. QuickBooks sits inside core accounting, payroll, payments, bill pay, capital and workforce workflows, while TurboTax and Credit Karma give INTU both a tax-acquisition funnel and year-round consumer monetization engine. Proprietary financial data, the accountant network and regulated-domain expertise should make AI more valuable inside INTU’s platform than in a standalone generic application. Early product evidence supports that view: millions are using AI-native experiences, Enterprise Suite agent adoption exceeds 75%, and Accountant Suite already has more than 150,000 accountants on the platform.

This quarter also showed that the strategic growth assets are not broken. Assisted Tax, Money and Mid-Market grew 34% collectively; mid-market revenue grew 39%; online Money grew 31%; Enterprise Suite annualized revenue exceeded $145M; and Credit Karma grew 16.5% in Q4. These businesses provide multiple paths to offset lower entry pricing through services attach, mix and lifetime value. The reset may also prove conservative. TurboTax’s 2–3% guide is well below Street, Live drops from 37% to the mid-teens, and FY27 total growth is being set below 10% despite INTU not having grown below 10% since FY15. The Q4 margin and cash-flow results, coupled with an ex-SBC FY27 profit guide above Street, show that INTU retains meaningful expense and capital-return levers while it fixes customer acquisition.

Valuation is the other major bull argument. 12x one FY28 SBC-inclusive EPS estimate, an unusually low valuation for a company still targeting high-teens EPS growth. A modest stabilization in DIY units or an improvement in QBO customer growth could therefore produce both estimate upgrades and multiple expansion. The key catalysts are evidence that QuickBooks Free and Lite are adding incremental customers, better-than-guided tax volumes and a successful conversion from lower entry ARPC into payments, Live and Credit Karma monetization.

**The bear thesis** is that INTU’s low-end weakness is not merely a temporary allocation mistake. Online paying customer growth has decelerated to 3%, U.S. TurboTax units declined 2%, and price has become the main reason for tax churn. Low-cost and potentially AI-native products are reducing the value consumers place on traditional DIY software, while QuickBooks faces increasingly capable alternatives aimed at new and very small businesses. Lowering entry prices may stabilize volume, but it also risks exposing that prior revenue growth depended too heavily on pricing and upsell rather than durable customer expansion.

The Live story also looks less proven than bulls believed. More than three-quarters of FY26 Live additions came from converting existing DIY customers, and Live growth is now expected to slow from 37% to the mid-teens. That suggests INTU has been better at moving its own customers upmarket than taking meaningful share directly from tax professionals and stores. If DIY volumes continue shrinking, the future pool of customers available for Live conversion also shrinks. The TurboTax and Credit Karma lifetime-value argument is economically compelling, but it has not yet demonstrated that lower tax ARPC will create enough incremental engagement to offset the immediate revenue loss.

GBS has a similar funnel issue. Mid-market growth remains strong, but approximately three-quarters of additions came from upgrades or desktop migrations, total online customers grew only 3%, and Desktop represents nearly a quarter of GBS while entering structural decline. Mailchimp is flat to down, Capital is shifting toward lower-yield partner-funded loans, and mgmt would not commit to reacceleration by FY28. In that scenario, buybacks and SBC leverage may preserve EPS growth for a period, but investors could view the earnings algorithm as increasingly financial rather than operating and keep the multiple compressed.

### **SMTC: Good numbers and Guide driven by 800G and Earlier 1.6T conversion and continued LoRa strength as GMs are inflecting**

### **The #s:**

F2Q27 revenue/EPS of $341.9M/$0.71 vs. Street $328.6–329.0M/$0.61 were followed by F3Q27 guidance of $405–415M/$1.02–1.08 vs. Street approximately $360M/$0.73, with the upside driven by faster 1.6T FiberEdge qualifications and much stronger mix leverage. F3Q adjusted GM/OM of 58.3%/31.0% compare with Street approximately 53.9–54.5%/22.5–23.6%; excluding cellular modules, GM is guided to 63.9%, and mgmt described approximately 64% as the post-divestiture starting point.

Data Center reached $100M, +91% y/y and +39% q/q vs. a roughly $97M buyside bar; F3Q is expected at approximately $145M, +160% y/y and +45% q/q vs. Street $115M, with 1.6T becoming more than 50% of the Data Center mix.

The existing 800G cycle is strengthening, while 1.6T FiberEdge qualified earlier than expected and SMTC now expects more than 50% 1.6T TIA share by fiscal year-end.

### **Key Takeaways**

-

**1.6T.** The most important incremental disclosure was that 1.6T FiberEdge qualifications completed earlier than SMTC had assumed. CopperEdge timing was largely on plan, while the faster FiberEdge conversion produced the major upside to F3Q expectations. SMTC expects 1.6T FiberEdge and CopperEdge to exceed 50% of Data Center revenue in F3Q, with the percentage continuing to rise and FiberEdge TIA share exceeding 50% by FY-end. ACC volume deployment is still expected to inflect in F4Q, providing a potential second leg after the near-term optical-led ramp.

-

**800G.** The 1.6T ramp is not coming at the expense of 800G. Industry 800G transceiver expectations have risen from approximately 50M units entering the year to 80–90M, while SMTC’s 800G TIA share has increased from approximately 18% two years ago to more than 50%. This creates simultaneous volume, share and product-generation tailwinds rather than a simple replacement cycle.

-

**Margins.** Adjusted GM increased to 54.5%, +150bp q/q, and adjusted OM reached 24.4%, +400bp q/q, both above Street. F3Q GM/OM are guided to 58.3%/31.0%, with approximately 380bp of sequential GM expansion coming from product mix even before the structural benefit from the cellular-module divestiture. Excluding modules, F2Q GM was 59.7% and F3Q is guided to 63.9%; mgmt views approximately 64% as a starting point rather than a ceiling as 1.6T, LoRa and photonics become larger contributors. F3Q OpEx is modestly above Street because SMTC is accelerating R&D on key Data Center programs, but EPS still grows more than twice as fast as revenue.

-

**Callback.** The post-earnings callback improved the capacity read-through: FY27 capacity, including potential upside, is secured and back-end testing/packaging capacity is largely in place. The more relevant constraint is front-end wafer availability for 2H FY28, with additional facilities being qualified across the foundry partner’s U.S., Israel and Japan footprint. Capacity, rather than end demand or qualification, has become the primary limiting variable.

-

**LoRa.** LoRa revenue reached a record $58M, +58% y/y and +31% q/q, and is expected to increase another approximately 15% q/q and 65% y/y in F3Q. Mgmt said growth above the prior 20% bogey is sustainable, supported by LoRaWAN industrial deployments, LoRa Plus in smart-home/security applications and longer-term Amazon Sidewalk adoption. LoRa Plus is already approximately 20–25% of the LoRa mix, while Sidewalk remains nominal and is expected to be only high-single-digit this year, meaning the current acceleration is not dependent on a large Amazon contribution.

-

**AI roadmap.** SMTC is broadening from TIAs and linear equalizers into drivers, gain chips, high-power CW lasers, SOAs and photodiode arrays. Mgmt is engaged in approximately 10–15 NPO programs, with some directly tied to end customers. Initial CW-laser revenue in transceivers is expected in 1H FY28, while the major content expansion is tied to 3.2T: design windows begin opening in approximately 12 months, early deployment could start in approximately 18 months and meaningful volume is expected in roughly two years. Mgmt clarified that “high double-digit” content means approximately $80–90 per transceiver, not $18–19, versus high-single-digit dollars today.

-

**Pricing.** Near-term pricing is unusually supportive. Mgmt said there is almost no expected erosion on booked orders, availability is currently more important to customers and modest cost increases are being passed through in most cases. Follow-up commentary also indicated 1.6T ASPs are slightly more than 2x 800G. The caveat is that current pricing partly reflects scarcity and rapid qualification, so investors should not automatically extrapolate the same pricing environment once competing capacity expands.

-

**Competition.** Mgmt describes CopperEdge linear equalizers as the de facto standard and believes it already has the lion’s share of the market, while FiberEdge is designed into every module provider in its target market, several on a sole-source basis. That is a strong competitive position, but competitors are also adding capacity across TIAs, drivers, equalizers, lasers and photodiodes. The next proof points are sustained production share, broader ACC deployment beyond the lead hyperscaler and attaching drivers/photonics without ceding pricing or sockets.

-

**Portfolio.** The $62M sale of the cellular-module business removes approximately $48.1M of quarterly revenue carrying only a 22.7% GM. The transaction is expected to close in F4Q27, increase consolidated GM by more than 500bp and remain non-GAAP EPS neutral. SMTC retains AirLink routers, gateways, device-management/security software and LoRa silicon. Mgmt currently likes the retained portfolio but said portfolio optimization remains an ongoing process.

-

**Customers.** SMTC is engaging all leading hyperscalers and is designed into every module provider in its target optical markets, which broadens the module-customer base. Several sockets are sole source. However, underlying AI infrastructure spending remains concentrated among a small number of hyperscalers, and the initial ACC volume ramp is still tied disproportionately to a lead customer. Mgmt did not provide a specific customer-concentration percentage on the call.

-

**Cash flow.** Operating cash flow was $69M and FCF was $61M, a 17.8% FCF margin, vs. Street approximately $45M. CapEx was only 2% of revenue and included gain-chip and CW-laser capacity; mgmt expects CapEx to remain generally below 5% of revenue, although equipment timing can cause individual quarters to run slightly higher. Cash was $204M, debt principal was $503M and net leverage was 1.1x, giving SMTC room to fund wafer commitments and photonics expansion.

### **Bull vs Bear Debate**

**The broad bull thesis** is that SMTC is rapidly transforming from a mixed analog and IoT portfolio with legacy Sierra Wireless complexity into a focused, high-margin AI connectivity supplier. Its position is not limited to one product or one connectivity architecture. FiberEdge participates in fully retimed optics and linear optics; CopperEdge participates in ACCs and on-board equalization; the photonics portfolio adds gain chips, lasers, SOAs and photodiodes; and LoRa provides a separate high-margin growth engine outside the Data Center. The cellular-module sale formalizes this strategic shift by removing a low-margin business and concentrating investment on Data Center and LoRa, the areas where SMTC believes it has the strongest growth and industry-leadership position. Mgmt also characterized bookings and backlog as extending well into the next fiscal year, while the divestiture announcement explicitly framed the transaction as a way to sharpen focus on Data Center and LoRa connectivity.

The competitive advantage bulls see is the combination of analog signal-integrity expertise, established module-provider relationships and an expanding content opportunity. SMTC’s 800G TIA share increased from approximately 18% to more than 50%, and mgmt believes it can follow the same path at 1.6T. Meanwhile, the 800G unit market itself has reset substantially higher, reducing the risk that 1.6T merely cannibalizes an existing revenue stream. At 1.6T, ASPs are more than 2x 800G, and at 3.2T the company can potentially sell a much broader combination of TIAs, drivers, lasers, gain chips, photodiodes and amplifiers. If content per transceiver can move from high-single-digit dollars toward $80–90, the long-term revenue opportunity becomes meaningfully larger than the current TIA market alone.

This quarter materially strengthens that case. The earlier FiberEdge qualification shows that the 1.6T thesis is converting into orders faster than expected. Data Center guidance of approximately $145M is around $30M above Street, the remaining FY27 plan is booked and more than 70% of FY28 demand is already represented in backlog. At the same time, the earnings conversion is stronger than the revenue upside: F3Q EPS is approximately 45% above Street, ex-module GM reaches 63.9% and adjusted OM reaches 31%. The upside is therefore not simply a temporary revenue surge; it is producing evidence that SMTC can become a mid-60s-plus GM and potentially high-30s-to-low-40s OM business. The wide upward revisions to out-year EPS across the attached models reflect this margin reset.

**The bear thesi**s is that the new estimates capitalize an unusually favorable combination of accelerated customer qualification, scarcity pricing, elevated backlog and simultaneous strength in both 800G and 1.6T. Backlog at this stage is not the same as non-cancellable revenue. Customer deployment schedules can change, orders can be pulled between quarters and AI infrastructure demand is concentrated among a small number of hyperscalers. SMTC may be designed into every major module provider, but the ultimate spending pool remains concentrated, and the first large ACC deployment still depends heavily on a lead customer and a new platform cycle.

Capacity is also a two-sided issue. It validates demand, but it can prevent backlog from converting into revenue and may require prepayments, wafer commitments and higher CapEx. Mgmt believes near-term and FY27 capacity is secured, but front-end wafer supply could constrain upside in 2H FY28. Qualification across additional facilities introduces yield, timing and execution risk. If capacity arrives late, SMTC could lose share precisely when competitors are expanding their own TIA, driver, equalizer, laser and photodiode capacity. If capacity arrives into a slower demand environment, SMTC could instead be left with underutilized commitments and weaker incremental returns.

The margin thesis also carries risk. The 64% ex-module GM starting point is being established during a period of favorable scarcity pricing and a rapid mix shift toward high-margin products. It assumes the module sale closes on time, stranded costs are removed, 1.6T yields are healthy, cost increases remain passable and normal optical price erosion does not return. The $62M sale price is only approximately 0.3x annualized module revenue, reflecting the asset’s low profitability but also illustrating that SMTC is surrendering a meaningful revenue stream. Outside AI Data Center and LoRa, High-End Consumer remains down y/y, Sidewalk is not yet a major contributor and mgmt did not provide evidence of a broad macro recovery.

### What Changed This Quarter

The bull case strengthened materially. The biggest changes were the earlier-than-expected 1.6T FiberEdge qualification, continued upward revision to the 800G unit market, unusually strong FY28 backlog visibility and proof that the product mix can generate far more operating leverage than the Street had modeled. The 64% post-divestiture GM starting point is especially important because it increases the value of each incremental dollar of Data Center and LoRa revenue. The quarter also made the story broader: near-term growth is primarily optical, ACC provides a potential F4Q and FY28 leg, and photonics/3.2T provides a longer-duration content opportunity.

The near-term bear case weakened because demand uncertainty and margin skepticism were the two easiest ways bears challenged the prior thesis, and the quarter produced strong evidence against both. However, the bear debate did not disappear; it moved outward. Investors now need to determine how much of the backlog is durable, whether front-end capacity arrives in time, whether 1.6T share sustains after the initial qualification wave and whether mid-60s GM survives normal pricing and competitive conditions.

The central debate has therefore changed materially. Before the quarter, the question was whether 1.6T and ACC would convert quickly enough to justify the AI-connectivity narrative. After the quarter, the near-term 1.6T conversion is largely proven. The debate is now how long the simultaneous 800G/1.6T cycle lasts, how much additional content SMTC captures and what multiple investors should pay for earnings that are rising rapidly but remain dependent on a concentrated and capacity-constrained AI spending cycle.

### **ZM: Enterprise and RPO acceleration look real, but Online top-of-funnel and AI costs keep the near-term guide from clearing the bar**

F2Q revenue and EPS were above Street, but F3Q revenue, operating-margin and EPS guidance came in below Street as Online customer acquisition softened and higher AI usage and R&D spending pressured margins.

### **The #s:**

**Revenue:** $1.277B, +4.9% y/y (last q +5.5%; +4.7% cc vs. +4.6% last q) vs. Street approximately $1.267B, +4.1%. **EPS:** $1.55 vs. Street $1.48. **Non-GAAP OM:** 40.0% vs. Street 40.4%.

Enterprise revenue accelerated to $788M, +7.8% y/y (last q +7.2%) vs. Street approximately $776M, +6.3%, while RPO growth accelerated to +14.1% from +10.9% and cRPO improved to +6.9% from +5.6%.

**F3Q revenue guidance** midpoint of $1.278B, +3.9% y/y, was below Street at approximately $1.282B, +4.2%, with operating margin of 40.1% vs. Street 40.9% and EPS of $1.47 vs. Street $1.50–1.51.
**
FY27 revenue midpoint** of $5.090B, +4.5%, was essentially in line with Street; EPS and FCF were raised above Street, but Online was reduced to flat and AI costs plus Common Room kept operating-income guidance unchanged.

### **Key Takeaways**

-

**Enterprise.** Enterprise revenue was $788M, +7.8% y/y (last q +7.2%) vs. Street approximately $776M, +6.3%, the strongest growth in three years despite an approximately 60bps white-label churn headwind. Enterprise reached 62% of revenue, NDR stayed at 99%, and customers contributing more than $100K rose 8.2% to 4,625. This is the quarter’s strongest evidence that the reacceleration thesis is working.

-

**Forward KPIs.** RPO reached $4.536B, +14.1% y/y (last q +10.9%), cRPO accelerated to +6.9% from +5.6%, non-current RPO grew 25%, and deferred revenue grew approximately 6% vs. ZM’s prior 2–3% guide. No specific Street bogeys were cited for these metrics. Mgmt tied the strength to larger, longer-duration and multi-product deals, although it cautioned that the formal revenue guide remains the best indicator of near-term revenue.

-

**Online.** Online revenue was $490M, +0.6% y/y (last q +2.8%) vs. Street approximately $492M, +1.1%. Mgmt reduced the FY27 outlook from slight growth to flat, which implies a modest H2 decline after positive H1 growth. The issue was new-customer acquisition rather than retention: churn stayed at 2.9%, and an aggregate approximately 6% price increase produced little to no churn response. The new risk is that AI-based discovery structurally disintermediates the traditional search funnel.

-

**Callback.** The most important callback message was that mgmt did not view the Online softness as a broader macro or spending problem. The H2 guide carries the F2Q top-of-funnel environment forward rather than assuming further deterioration. Mgmt is responding through answer-engine optimization, better buy flows, improved PLG conversion, pricing and packaging, and land-and-expand motions across Phone, Webinar and AI.

-

**AI.** Licensed Workplace AI-feature MAUs grew 125% y/y, ZVA customers increased more than 250%, ZRA paid customers grew 41%, and paid AI appeared in nine of the top 10 ZCX deals. Workvivo also surpassed $100M of ARR. Contact Center remains the most material source of direct AI monetization today; ZoomMate, ZRA/Common Room and BrightHire are earlier-stage opportunities.

-

**CX/Phone.** ZCX ARR continued to grow at a high-double-digit rate and delivered a record number of seven-figure ARR deals. Phone ARR growth moderated to the teens from mid-teens last quarter, but the competitive evidence remained constructive: all 10 largest Phone deals were takeouts, nine involved on-premise replacements, and five of the top 10 Phone deals included Contact Center. This supports the UCaaS-plus-CCaaS consolidation thesis, although the Phone growth deceleration is worth monitoring.

-

**Margins.** Non-GAAP gross margin was 79.1%, down approximately 70bps y/y and below Street at 79.8%; operating margin was 40.0%, down 139bps y/y and below Street at 40.4%. R&D increased approximately 24% y/y. The pressure came from elevated usage and trials for newer AI products, infrastructure and token spending, product reinvestment and lower-margin acquired businesses. Mgmt expects AI-cost optimization to help in H2 and reiterated its long-term approximately 80% gross-margin target.

-

**Pricing.** Online’s approximately 6% aggregate price increase had no meaningful impact on churn, while Enterprise saw record-low discounting, longer contract durations and stronger deal shaping. That is encouraging evidence of pricing discipline. For newer AI offerings, ZM is increasingly supporting consumption-based and outcome-based structures, but mgmt characterized the revenue contribution from those models as early.

-

**Guide.** F3Q revenue, operating margin and EPS were all below Street. FY27 reported revenue guidance was essentially in line, while constant-currency revenue guidance was slightly above Street. Mgmt passed approximately $9M of the F2Q constant-currency outperformance into the FY guide, but Online weakness, lower FX tailwinds and approximately 40bps of FY white-label churn limit the acceleration. Positive revisions against this new baseline will likely be needed for a more durable rerating

-

**M&A.** Common Room, acquired for approximately $250M, was included in guidance but described by mgmt as de minimis to ZM’s approximately $5B revenue base and dilutive to margins in the near term. Mgmt did not quantify the revenue contribution.

### **Bull vs. Bear Debate**

The broad bull thesis is that ZM is successfully transitioning from a mature video-meetings vendor into a high-margin, multi-product enterprise communications and AI platform. The most important proof point this quarter was Enterprise revenue accelerating to +7.8% y/y vs. Street around +6.3%, despite an approximately 60bps white-label headwind. RPO growth of +14.1%, cRPO acceleration, record-low discounting and larger, longer-duration contracts provide forward evidence that the improvement is not simply a one-quarter renewal anomaly. Bulls will argue the next major milestone is Enterprise NDR moving above 100%, particularly once the temporary white-label churn laps.

The longer-term drivers are Phone, Contact Center, ZVA, Workvivo and direct AI monetization across the installed Workplace base. ZM has a potentially durable advantage in combining UCaaS, CCaaS, voice, video and conversational context within one platform. This quarter’s competitive evidence was particularly constructive: all 10 largest Phone deals were takeouts, nine involved on-premise systems, paid AI featured in nine of the top 10 ZCX deals, and ZVA customers increased more than 250%. Bulls see a multiyear replacement cycle as enterprises reassess legacy communications platforms for an AI and agentic-workflow world.

Margin pressure can also be interpreted positively if it reflects genuine product usage. Mgmt attributed the gross-margin shortfall largely to an AI-usage spike and free trials, rather than pricing pressure or customer losses. Federated routing and migration of high-volume tasks onto ZM’s SLM provide a plausible path back toward approximately 80% gross margins. If Enterprise remains near double-digit growth, Online stabilizes, NDR rises above 100%, and AI costs moderate, total revenue could move toward mid-to-high-single-digit growth while FCF margins expand.

**The broad bear thesis** is that ZM remains a roughly 4–5% grower whose mature, highly profitable Online and Meetings businesses are being offset, rather than transformed, by newer products. F2Q revenue was above Street, but F3Q revenue growth of +3.9% at the midpoint was below Street around +4.2%, and the FY27 revenue midpoint was merely in line. More importantly, the Online outlook was cut from slight growth to flat, implying a modest H2 decline. The shift from traditional search toward AI platforms may be a structural customer-acquisition problem rather than a temporary macro issue. Stable churn protects the installed base, but it does not solve a weakening new-customer funnel.

The bear case also questions whether AI monetization will generate attractive incremental economics. Gross margin of 79.1% was below Street at 79.8%, operating margin of 40.0% was below Street at 40.4%, and R&D grew approximately 24%. If rising usage, tokens, model costs, free trials and consumption-based products require structurally higher COGS, the approximately 80% gross-margin target could become difficult to sustain. Common Room adds another lower-margin asset, while its precise organic contribution was not disclosed. Bears can also point to Phone slowing from mid-teens to teens, Enterprise NDR remaining below 100%, and intensifying competition from Microsoft, CRM vendors, established contact-center providers and AI-native entrants.

The FCF raise is less clean than the headline suggests because approximately $40M came from extending a data-center asset’s useful life rather than a fundamental improvement in unit economics. Meanwhile, the Anthropic stake is illiquid and could face taxes, dilution or valuation volatility, so bears will not give full credit until monetization becomes visible. If the Enterprise reacceleration settles back into the mid-single digits, Online declines and AI spending prevents margin expansion, ZM risks remaining a low-growth value stock rather than earning a broader software multiple.

# **TECH RESEARCH/NEWS**

### **ORCL: Citi opens 90-day Positive Catalyst Watch, sees historic selloff as opportunity**

Citi says Oracle’s >30% three-month underperformance and >50% peak-to-trough drawdown reflect investor capitulation, technical selling and credit/ATM concerns more than a deterioration in AI demand. The key read-through is from neoclouds, where GPU pricing is up ~25–30%, newer contracts carry ~5–10pts better contribution margins and shorter durations improve economics, suggesting ORCL could see similar upside on recently signed AI infrastructure deals. Citi expects positive estimate revisions into earnings and the October investor day, raises FY28–30 revenue/EPS, and reiterates Buy/$330 PT.

### **TEAM: RBC initiates Outperform, $215 PT, sees AI as a tailwind rather than a seat-compression threat**

RBC argues Atlassian can sustain 20%+ subscription ARR growth as AI increases software creation and coordination complexity, making Jira/Confluence more valuable rather than less. The key differentiator is the Teamwork Graph, which spans 200B+ objects and can act as a control plane for agents; RBC cites early users of AI coding tools showing ~5% more Jira tasks, higher MAU and faster seat expansion. Collections/Rovo adoption and improving enterprise GTM provide additional upside, while the end of the multi-year platform R&D cycle should support meaningful margin expansion.

### **META: Meta, States Have Discussed Settling Teen Social Media Case - Bloomberg**

Bloomberg reports Meta and attorneys general from 29 states have discussed a possible mid-trial settlement in the teen social-media addiction case, potentially reducing what Meta itself estimates could be up to $1.4T of theoretical trial exposure. A settlement would likely be far smaller, though the states are also seeking mandatory changes to Facebook and Instagram, making the case important for both financial liability and future product design.

### **SAP: UBS downgrades to Neutral as cloud growth slows and AI monetization lags**

UBS sees H2 cloud backlog growth decelerating from 24.6% in Q2 toward ~23%, while cloud gross margins have begun to feel higher AI costs and FCF upside looks less likely this year. AI remains more promise than earnings driver, with only 17 out-of-the-box agents launched versus a 200 year-end target and gradual Joule adoption. PT moves to €201 from €164.

### **SE: 3p saying e-comm GMV tracking below street driven by softer trends in Indonesia**

### **Anthropic**: **Anthropic Expected to Tell Investors It Sees Over $30 Trillion in Potential Revenue - [WSJ](https://www.wsj.com/tech/ai/anthropic-expected-to-tell-investors-it-sees-over-30-trillion-in-potential-revenue-a611efea)**

**

WSJ reports Anthropic is likely to pitch a $30T+ TAM in its IPO, above SpaceX’s $28.5T estimate, by framing the opportunity around the full scope of work AI models could eventually perform. The company could seek to raise as much as $100B at roughly a $2T valuation, both above SpaceX’s recent offering, with IPO filings expected in coming weeks and a potential debut as soon as September or early October. Anthropic’s revenue more than doubled to $11.6B in 2Q, though WSJ notes the enormous TAM assumptions are likely to draw scrutiny given how speculative AI market sizing has become.

### **OpenAI**: **OpenAI’s Head of Data Centers Has Left the Company / next pre-train “Bel” finished - [WSJ](https://www.wsj.com/tech/ai/openais-head-of-data-centers-has-left-company-6d24fd83?mod=newsviewer_click&adobe_mc=MCMID%3D26622675878327304541006841196593856970%7CMCORGID%3DCB68E4BA55144CAA0A4C98A5%2540AdobeOrg%7CTS%3D1787685744&_gl=1*sbdtio*_gcl_au*MzE0ODgyNDgxLjE3ODc2ODU3NDIuLS4tLjE3ODc2ODU3NDIuNjcxNjQ5NzA1LjE3ODc2ODU3NDIuMTc4NzY4NTc0Mg..*_ga*MTU4ODI3OTQ5Ni4xNzc5NTA1MzE4*_ga_K2H7B9JRSS*czE3ODc2ODIxMzckbzI3NyRnMSR0MTc4NzY4NTc0MiRqNjAkbDEkaDcwODEyNDI5MA..)**





### **AI Infra: William Blair says competition is shifting from chips to systems, with custom ASICs increasingly credible**

William Blair’s main Hot Chips takeaway is that differentiation is moving beyond raw FLOPS toward full-system architecture across compute, memory, networking, software and reliability. Nvidia still has the clear lead, but AMD’s Helios architecture shows a much more mature open-Ethernet scale-up stack, positive for ALAB/AVGO, while OpenAI’s Jalapeño demonstrated that tightly co-designed custom ASICs can deliver 50–90% better performance/W and materially lower latency than Blackwell-class systems on targeted inference workloads. Google’s TPUv8 similarly reinforces the custom-silicon threat, while Blair views HBF as more of a niche lower-cost memory tier than a broad HBM replacement.

### **NVDA/AMD/AMZN: JPM OpenRouter work shows token demand exploding, but pricing and B200 rentals starting to soften**

JPM’s August OpenRouter data show token volume +47% m/m and 28x y/y, driving spend +7% m/m and 12x y/y despite volume-weighted pricing falling 28% m/m and 56% y/y. The bigger yellow flag is infrastructure pricing: B200 rentals declined sequentially for the first time in 6+ months and the premium versus H100 narrowed, while H100 pricing actually firmed, hinting at some rotation toward older capacity. Net, AI usage remains extremely strong, but rapidly falling model pricing and the first cracks in leading-edge GPU rental economics are worth watching for NVDA/neocloud monetization.

### **NBIS/CRWV: Susquehanna sees GPU rental pricing strengthening**

Susquehanna’s August cloud GPU work shows pricing up ~7.9% q/q across leading-edge NVIDIA instances, with B200 the standout at +15% and near the top of its historical range, directly contradicting JPM’s indication of initial B200 rental softness. At the same time, frontier LLM token pricing fell ~40% to ~$2.84/million tokens versus only ~4% for open models, suggesting model-layer commoditization is accelerating even as infrastructure pricing remains firm. Susquehanna also raises its 2030 AI accelerator TAM to $1.1T from $1.0T, with AWS/GCP continuing to broaden Blackwell availability.

### **AVGO: RBC expects slight beat/raise, sees FY27 AI guide moving toward ~$120B**

RBC expects a modest beat/raise driven mainly by networking, with strong demand across Tomahawk/Jericho, PCIe and optical helping offset lingering TPU share concerns. The key debate is FY27 AI revenue: RBC thinks management could lift the prior $100B+ framework toward Street at ~$120B, while MediaTek/MRVL create some competitive noise around Google TPU sockets but are unlikely to meaningfully disrupt AVGO’s long-term position. RBC models ~$77B of FY27 custom ASIC revenue and sees additional upside from networking, though lower-margin TPU mix remains a GM headwind; maintains Sector Perform/$400 PT.

### **TTWO: Benchmark says “buy the leak,” sees GTA VI footage strengthening the fundamental case**

Benchmark says the latest unauthorized GTA VI clips are a marketing headache but a positive product signal, showing a build that looks increasingly polished, stable and launch-ready with no evidence of weaker demand or another delay beyond Nov. 19. Engagement is also rising sharply, with Google searches +169% w/w and strong social/video traction, while the footage appears to show GTA VI already running on PC, adding support for a future PC release and another monetization leg. Benchmark sees the Aug. 27 Netflix presentation as Rockstar’s opportunity to retake control of the narrative with polished gameplay.

### **PANW: BTIG checks improve materially, sees room for FY27 guide above Street**

BTIG’s partner work strengthened versus April, with core PANW practice growth now commonly tracking around 15%+ versus low-teens previously, helped by larger platform deals, firewall refresh, Prisma SASE and XSIAM. It sees FQ4 revenue of ~$3.38B, +33% y/y/+15% organic, modestly above Street, and NGS ARR of ~$8.94B vs. ~$8.91B consensus. More importantly, BTIG thinks Street’s FY27 revenue implies only ~13% pro forma growth including CyberArk/Chronosphere versus a potential ~16%+, leaving room for guidance above consensus and further upside through the year; reiterates Buy/$380 PT.

### **ZS: Evercore adds Tactical Underperform, sees FY27 ARR guide skewing to low end as competition/GTM pressure builds**

Evercore’s channel checks point to an in-line F4Q with limited upside, but growing concerns around FY27 as new-logo momentum softens, competitive intensity rises across SASE, and recent sales leadership turnover weighs on execution. It sees FY27 ARR growth closer to the low end of the initial 16–17% guide versus Street/buyside around ~16.6–16.9%, with PANW, NTSK and Cato increasingly pressuring win rates. Evercore stresses the broader security backdrop remains healthy; the issue is more ZS-specific, making the stock a “show-me” story into FY27.

### **Memory: China’s YMTC aims to overtake Samsung and SK Hynix in flash memory race - [FinancialTimes](https://www.ft.com/content/1b1c357f-6a1c-4ce3-bb47-980c3de4d3d2?syn-25a6b1a6=1)**

**

*Financial Times r*eports YMTC is targeting the global NAND #1 position by as soon as end-2027, overtaking Samsung and SK hynix, as it prepares a Rmb33bn ($5bn) IPO largely to fund capacity upgrades and R&D. YMTC is already the world’s #3 NAND producer and generated Rmb47bn revenue / Rmb33bn net income in 1Q, with an expected post-listing valuation above Rmb1tn ($150bn), though IPO pricing may be conservative. The bigger read-through for memory investors is supply: YMTC and CXMT are using aggressive capacity expansion and state support to narrow the technology gap, raising concerns that Chinese supply could pressure global NAND/DRAM pricing.

SK Hynix, which remain the industry’s two largest suppliers by shipment volume.


### **China AI: DeepSeek nears pre-IPO funding round as 2027 market debut takes shape. Landmark round to lift Chinese start-up’s valuation to US$74 billion and set the stage for a Shanghai Star Market listing, sources say - [SMTC](https://www.scmp.com/tech/big-tech/article/3365280/deepseek-nears-pre-ipo-funding-round-2027-market-debut-takes-shape-sources?utm_source=feedly_feed)**

### **SoftBank: Exploring a $10B-$20B bond offering to refinance bridge financing tied to its OpenAI investment – [Bloomberg](https://www.bloomberg.com/news/articles/2026-08-26/softbank-mulls-up-to-20-billion-bond-sale-for-openai-financing)**

**

Bloomberg reports SoftBank is discussing a $10B-$20B dollar/euro bond sale as soon as September, with proceeds partly used to refinance the $40B bridge loan backing its OpenAI investment and fund additional AI bets. SoftBank is expected to invest nearly $65B in OpenAI by October, and the potential deal would significantly narrow a $20B+ funding gap, but credit markets are showing some concern, with its bonds falling and CDS spreads widening on the news.

### **SPCX: BofA says wireless ambitions are more complementary than disruptive, positive for TMUS/towers/spectrum**

BofA says SpaceX would still need much more spectrum, years of permitting/buildout and a terrestrial coverage layer to compete meaningfully in U.S. wireless, making the oft-discussed 1.5B-femtocell approach unrealistic. Its calls with TMUS and CCI suggest the more likely path is partnership and/or macro-tower deployment, with direct-to-device remaining a rural/complementary layer rather than a nationwide substitute. The read-through is positive for TMUS, tower demand and spectrum values, while implying a much higher execution/capital burden for SpaceX if it wants to build a full terrestrial network.

### **DOCN: Baird names Top Pick, sees AI mix and smaller footprint as underappreciated**

Baird sees 2027 revenue growth reaching ~54% versus ~50% guidance with ~40% EBITDA margins, supported by AI ARR already at ~$234M, ~21% of total. The differentiated angle is DOCN’s much smaller ~155MW footprint versus multi-GW neocloud peers, limiting power/capacity execution risk, while the recent ~30% GPU price increase points to healthy inference economics.

### **APP: Needham cuts outer-year estimates as Consumer ramp and new AI products slip**

Needham cuts 2027/28 revenue by ~$472M/~$1.2B as Consumer penetration is rebased materially lower and Games growth expectations come down, while 2026/27 EBITDA falls ~4.5%/~3.3%. Pixel additions have also slowed and the delayed gen-AI creative/lead-gen products remove two near-term TAM catalysts, leaving Axon improvements as the main growth lever. Needham reiterates Buy but cuts PT to $475 from $500.

### **SPCX: Wolfe says Louisiana Starbase could materially relax launch bottleneck, Flight 14 next catalyst**

Wolfe views SpaceX’s Louisiana purchase plus recent White House regulatory changes as meaningful steps toward easing one of the biggest constraints on Starship: launch capacity. The planned site could ultimately support >12 launch towers and, at full buildout, potentially ~30 launches/day, versus current licenses of ~145 launches/year and only ~25/year of Florida pad capacity. That matters directly for orbital AI ambitions, where Wolfe estimates ~190 Starship launches would be needed to deploy 1GW of compute; a successful orbital Flight 14 in early September is the next key proof point.

### **CIEN: Citi sees FY27 guide above Street as key catalyst, optical momentum remains strong**

Citi expects Jul-Q sales of ~$1.64B, +35% y/y, with Optical Networking posting a third straight quarter of >40% growth, and EPS of ~$1.75 vs. Street $1.72. The bigger focus is preliminary FY27 guidance, where Citi sees ~$8.3B revenue, +30.5%, versus Street ~$8.0B/+25.7%, supported by sustained optical demand, improving supply visibility and continued share gains. Dell’Oro data also remain supportive, with pluggables growing >70% and CIEN optical transport sales up >40% y/y.

### **CIEN: Needham sees telecom optical becoming the next AI bottleneck, expects beat/raise**

Needham’s checks point to very strong demand across both cloud and telco, with AI scale-across and MOFN projects accelerating and backlog expected to rise materially from ~$7.7B. It sees 3Q revenue ~4% above Street and 4Q guidance ~5% above consensus, with FY27 growth of at least ~36% achievable despite management likely guiding prudently. The more interesting read-through is broader hyperscaler traction: META is already >12% of YTD revenue, a second hyperscaler is ramping, and Needham expects Amazon to shift its in-house transponder toward Ciena silicon in 2H26; reiterates Buy/$600 PT.

### **CDW: Evercore ISI CFO meeting reinforces healthy IT demand, sees services/margin leverage improving in 2H**

Evercore ISI says its meeting with CDW CFO Al Miralles and IR reinforced a healthy demand backdrop, with enterprise hardware, server/storage and AI-related activity stronger than expected, though the mix toward lower-margin hardware remains a GM headwind. The more interesting setup is 2H: recent hardware deployments should begin pulling through higher-margin software/services attach, while Geared for Growth is tracking at or above plan and should add operating leverage. PC units remain down LSD but higher ASPs are offsetting the decline; Evercore maintains Outperform and a $160 PT.

### **FSLR: Deutsche Bank says Section 232 strengthens moat, but pricing benefit will be gradual**

Deutsche Bank reiterates Buy/$272 PT, arguing FSLR is one of the clearest beneficiaries of the new Section 232 framework given its fully integrated U.S. manufacturing, CdTe technology and ~$14B backlog mostly committed through 2028. The policy should structurally raise the U.S. module cost floor and improve FSLR’s relative economics, but DB cautions the existing ~95GW module stockpile means ASPs likely rise gradually rather than immediately. Longer term, DB sees domestic integrated producers potentially becoming the lowest-cost suppliers after 45X credits, supporting higher revenue/EBITDA from 2027 onward and making FSLR’s current valuation discount look too punitive.

### **AI: [Bill Gates cautioned](https://www.gatesnotes.com/a-turbulent-ai-era-and-critical-choices-to-make) that AI development is advancing with insufficient oversight and preparation, warning the transition could be one of the most disruptive periods in history**

**

Bill Gates argues AI could become either the greatest equalizer in history or a major driver of inequality, and believes society is badly underprepared for the transition. Unlike past technology shifts, he expects AI and robotics to rapidly replace both white- and blue-collar work, potentially leaving structurally fewer jobs rather than simply creating new categories of employment. He highlights three major risks: permanent job displacement, dramatically greater cyber, biological and political harms, and damage to children’s development, relationships and critical thinking. At the same time, Gates sees enormous upside across healthcare, scientific discovery, agriculture, education and government services, particularly if AI capabilities are broadly distributed. His policy prescription is aggressive: create new national and global AI institutions, designate some work as “Human Reserved,” and tax AI tokens and robots to slow labor displacement and fund retraining and stronger safety nets.