Agent of Record
In this edition of the Smart Investor newsletter, we spotlight the company controlling enterprise AI’s runways. But first, let’s review the latest Smart Portfolio developments.
1
Portfolio News and Updates
❖❖ Amphenol (APH) delivered a record quarter, with revenue rising 55% year-over-year to $8.8 billion and adjusted EPS climbing 67% to $1.35, beating expectations on both counts. The central story was accelerating AI infrastructure spending: IT datacom sales surged 89% year-over-year (63% organically) as demand broadened across Amphenol’s copper, fiber-optic, and power solutions. IT datacom represented 43% of total revenue. Companywide orders reached a record $10.7 billion, producing a strong book-to-bill ratio of 1.23.
Profitability rose sharply alongside revenue, with adjusted operating margin approaching 29% after excluding the tariff-recovery benefit, up from 25.6% a year earlier. The expansion reflected strong operating leverage as well as better-than-expected performance and integration at CommScope.
Cash generation remained robust, with operating cash flow of $1.6 billion and free cash flow of $1.2 billion. Amphenol ended Q2 with $5.4 billion in cash and short-term investments, $18.8 billion in debt, and net leverage of just 1.3x. The company also completed the small El.Com and Wilder Technologies acquisitions, adding capabilities in industrial, defense, aerospace, and high-speed datacom applications.
For Q3, management guided revenue to $9.3-9.4 billion, representing 50-52% year-over-year growth, and adjusted EPS to $1.40-1.42, implying 51-53% growth. IT datacom sales are expected to rise in the mid-teens sequentially, more than offsetting weakness in communications networks. APH also raised expected 2026 CommScope revenue to $4.6 billion and doubled projected EPS accretion to $0.30.
Several analysts raised their price targets following the report, adding to multiple increases ahead of earnings as Wall Street confidence in Amphenol’s AI exposure and execution continued to build.
1
❖❖ Vertiv (VRT) reported a mixed Q2, as strong demand and profitability collided with the growing complexity of large AI infrastructure projects. Revenue rose 24% year-over-year to $3.27 billion but missed expectations, while organic growth of 17.8% fell well short of consensus. Management attributed the shortfall to temporary supply-chain congestion and timing shifts in multiphase projects, which delayed revenue recognition without weakening demand. Backlog exceeded $15 billion, up more than 76% year-over-year, supported by hyperscalers, colocation providers, neoclouds, and enterprise customers.
Profitability remained excellent. Adjusted EPS surged 60% to $1.52, beating expectations, while adjusted operating margin expanded 410 basis points to 22.6%. Productivity gains, operational execution, and favorable price-cost dynamics more than offset tariff pressure.
Vertiv generated $1.1 billion in operating cash flow and $925 million in adjusted free cash flow, ending the quarter with a net cash position and negative 0.1x net leverage. This gives the company ample capacity to fund the manufacturing expansion needed to convert its massive backlog into sales.
Management raised its full-year outlook despite the miss, as strong July activity, additional capacity coming online, and substantial backlog coverage support the expected sharp ramp of delayed revenue in the second half. Vertiv now forecasts revenue of $13.8-14.2 billion, implying 37% year-over-year growth at the midpoint, adjusted EPS of $6.65-6.75, representing 60% growth, and adjusted free cash flow of roughly $2.5 billion, up about 32%.
The stock initially plunged as the miss landed amid a broader selloff in chips and AI infrastructure, but dip-buyers soon emerged, reflecting Vertiv’s essential position in the AI value chain. Several analysts trimmed their targets while largely maintaining Buy or Outperform ratings; even after those cuts, the consensus target still implies over 50% upside.
1
❖❖ EMCOR (EME) delivered a blowout Q2, with record results and a sharp guidance raise underscoring its increasingly important role in the AI infrastructure buildout.
Revenue rose 19.8% year-over-year, or 19.6% organically, to a record $5.15 billion, beating expectations. Diluted EPS surged 34.8% to $9.06, well above the consensus. Operating income climbed nearly 32% to $547.3 million. Operating margin expanded 100 basis points to 10.6%, helped by exceptional electrical-project execution, favorable mix, and stronger overhead absorption as revenue accelerated.
Growth remained broad-based, but data centers were the standout. Network and communications revenue increased 45% in Electrical Construction and more than doubled in Mechanical Construction, reflecting soaring demand for the power and cooling systems required by increasingly large AI facilities. Remaining performance obligations reached a record $17.14 billion, up 43.9% year-over-year, with 95% of the increase organic. Management said it has seen no change in data-center demand.
That strength supported a major 2026 outlook increase. EMCOR now expects revenue of $20.0-20.5 billion, up about 19% year-over-year at the midpoint, diluted EPS of $32.00-33.25, representing roughly 16% growth, and operating margin of 9.5-9.8%, all well above its previous forecasts. Five pending electrical-construction acquisitions should contribute $250-275 million of second-half revenue, although their near-term EPS impact will be limited by amortization and lower interest income.
Shares surged as investors rewarded the clean beat-and-raise. Five of the six analysts who publish price targets raised them after the report, reinforcing the strongly bullish response.
1
❖❖ Generac (GNRC) delivered a mixed Q2 on the surface, but the results materially strengthened its emerging data center growth narrative.
Net sales rose 11% year-over-year to $1.17 billion, narrowly missing expectations, while adjusted EPS surged 76% to $2.91 and exceeded consensus. However, approximately $71 million in tariff refunds boosted the quarter’s results.
Commercial & Industrial (C&I) sales jumped 29% to $556 million, driven by more than $100 million in data center revenue. Data center backlog reached approximately $1.6 billion after Generac secured roughly $1 billion of new orders in 90 days. Meanwhile, Residential sales declined 2% as stronger home-standby demand failed to offset weaker portable-generator and energy-storage sales.
The first hyperscale customer committed nearly $700 million of 2027 volume, while GNRC signed a second hyperscaler whose expected orders are not yet included in backlog. To support the expansion, management plans to triple generator assembly and packaging capacity over the next 12 months.
For 2026, Generac raised its C&I growth outlook from the mid-to-high 20% range to the low 30% range and now expects nearly $450 million in data center revenue. Residential growth guidance was trimmed from roughly 10% to the high-single digits, while overall sales growth guidance remained in the mid-to-high teens. The adjusted EBITDA margin outlook rose to 20.0-21.0%, but excluding the 150-basis-point benefit from tariff refunds, the underlying forecast remains unchanged at 18.5-19.5%.
Shares surrendered their initial post-earnings gains. Several analysts cut their price targets, but all maintained their ratings, indicating caution over execution and business mix rather than rejection of the strengthening data center story.
1
❖❖ ASE Technology (ASX), Taiwan’s largest chip packaging and testing provider, delivered a powerful Q2 as AI demand pushed revenue, profit, and capacity investment sharply higher.
Revenue rose 27% year-over-year to approximately $6.0 billion, ahead of estimates, while diluted earnings per U.S.-listed ADS reached $0.29 and also beat consensus. Net income surged 180%. The core Assembly, Testing and Materials (ATM) business generated record revenue, up 36% year-over-year, while its gross margin expanded to 27.3% as higher utilization, operating leverage, and a richer advanced-packaging mix lifted profitability.
Demand now exceeds ASE’s available capacity. Leading-edge advanced packaging (LEAP) revenue is tracking above the previous $3.5 billion full-year forecast, and management aims to double it in 2027. Blended utilization stands at 80-85%, with near-term growth limited mainly by the speed of equipment installation and facility construction. The hyperscalers’ accelerating AI capex plans provide further support for sustained demand across the semiconductor supply chain.
The outlook reflects that momentum. Management expects Q3 consolidated revenue to rise 21-22% sequentially, with ATM revenue up 11-13% and ATM gross margin reaching 28-29%. It also expects that margin to exceed 30% in Q4.
ASE raised 2026 capex by $2 billion to approximately $10.5 billion and is managing 13 greenfield and eight brownfield expansion projects. That scale creates execution risk and will keep free cash flow under pressure, but the spending is directed at relieving a genuine capacity bottleneck. Investors welcomed the trade-off: the stock jumped after the report and extended its gains the following day.
1
❖❖ nVent (NVT) posted a blowout Q2 as accelerating data center demand drove exceptional growth across the business.
Sales surged 53% year-over-year to $1.47 billion, far exceeding expectations, while organic sales jumped 47%. Adjusted EPS rose 69% to $1.45, also beating consensus by a wide margin. Adjusted operating margin expanded by 110 basis points to 21.9%, despite more than $50 million in inflationary and tariff-related costs. Notably, adjusted results excluded a $25.8 million tariff reimbursement. Free cash flow increased 125% to $167 million.
Systems Protection organic sales soared 62%, while Electrical Connections grew 18%. Data center revenue is now expected to exceed $2 billion in 2026, more than double last year’s level, as demand for liquid cooling, cable management, engineered buildings, and power infrastructure accelerates. New products contributed more than 30 percentage points of quarterly growth.
To support this demand, nVent announced another 160,000-square-foot Minnesota liquid-cooling facility, expected to open in the first half of 2027. The expansion follows the recent opening of its Blaine facility and is intended to provide capacity into 2028.
For 2026, management raised organic growth guidance to 32-34% from 21-23% and adjusted EPS guidance to $5.00-5.10 from $4.45-4.55. Q3 organic growth is expected at 32-35%. The stellar results and the massive guidance raise further strengthened nVent’s AI infrastructure thesis.
1
❖❖ Microsoft (MSFT) delivered the clearest evidence yet that its enormous AI spending is producing equally enormous returns. Azure accelerated, enterprise demand broadened, and management expects growth to strengthen again – turning the quarter into a powerful answer to fears that Big Tech is building ahead of real demand.
Revenue rose 18% year-over-year to $90.0 billion, ahead of estimates, while operating income increased 18% to $40.6 billion. Adjusted EPS jumped almost 30% year-over year, reaching $4.74, although discrete items provided a net $0.27 benefit, led by a $3.2 billion gain on MSFT’s Anthropic investment. Excluding those items, underlying EPS was roughly $4.47 – still well above the $4.24 consensus estimate.
Azure was the main event. Revenue jumped 43%, pushing annual Azure sales above $100 billion for the first time, while Microsoft Cloud revenue rose 27% to $59.3 billion. Commercial remaining performance obligations surged 84% to $678 billion. OpenAI accounts for much of that increase, but the backlog grew 25% even without it, and every dollar of sequential growth came from customers outside the frontier-model companies. That is compelling evidence that enterprise AI demand is spreading far beyond a circular ecosystem of Microsoft-funded labs buying Azure capacity.
Monetization is also broadening across MSFT’s own products. Microsoft 365 Copilot surpassed 30 million paid seats, GitHub Copilot revenue accelerated more than 60% quarter-over-quarter, and Foundry revenue more than doubled. Management expects Azure growth to accelerate again to approximately 45% in constant currency this quarter, even as demand continues to exceed available capacity.
That demand makes Microsoft’s aggressive spending spree look more than justified. Fiscal-year capex reached $145.3 billion, including $41 billion in Q4, while Q1 spending will exceed $50 billion. MSFT added 31 data centers and 1GW of capacity during the quarter, while uncommenced data center lease commitments climbed to $329.1 billion. The costs are real: free cash flow fell 23% to $19.6 billion, and Microsoft Cloud gross margin narrowed to 65% as AI infrastructure and usage expenses rose. However, these investments support the long-term thesis and are already driving near-term profits, as this quarter showed.
MSFT also has unusual protection against picking the wrong AI winner. Azure offers models from OpenAI, Anthropic, Mistral, xAI, and Microsoft itself; its architecture lets customers swap models while keeping their data, context, and workflows on Microsoft’s platform. The company can profit from its investments, its own technology, and the infrastructure used by nearly every camp.
Investors responded by adding roughly $450 billion to Microsoft’s market value in one day, alongside a wave of analysts price target raises. The enthusiasm was justified: Microsoft has harnessed the AI wave across infrastructure, models, software, and investments – and is now monetizing every layer.
1
❖❖ Amazon (AMZN) shares surged after a spectacular Q2 release and a more than convincing earnings call showed that the company’s enormous AI investments are already producing accelerating revenue and profits.
Revenue rose 20% year-over-year to $200.6 billion, exceeding expectations, while operating income climbed 43% to $27.5 billion. Reported net income of $62.6 billion, or $5.75 per share, was heavily inflated by $53.4 billion of pre-tax non-operating income, primarily from Amazon’s Anthropic investment. Excluding this windfall, earnings per share would still have exceeded $1.90, well above consensus. The cleaner evidence lies in operations: even excluding two benefits totaling $1.2 billion, operating income grew approximately 37%.
AWS delivered the real fireworks. Revenue jumped 36.7% to $42.2 billion – its fastest growth in 18 quarters and the fifth consecutive quarterly acceleration. AWS operating income surged roughly 64% to $16.6 billion, producing a 39.4% margin. Excluding a $600 million energy-contract benefit, the margin was still approximately 38%, confirming that Amazon’s AI growth is already highly profitable.
AWS has now reached a $169 billion annualized revenue run rate, while its AI and proprietary-chip businesses have each exceeded $25 billion run rates and are growing at triple-digit rates. The figures likely overlap, but their scale is extraordinary. Trainium has secured multi-year, multi-gigawatt commitments from Anthropic and OpenAI, while Graviton is used by 98% of AWS’s 1,000 largest customers of EC2, its virtual-server and computing-capacity rental service.
Amazon also captures AI spending across the broader AWS stack, as workloads require CPUs, storage, databases, networking, and security alongside accelerators. Its position is reinforced by customers’ preference to run inference near the applications and proprietary data already hosted on AWS. Bedrock customer spending in Q2 exceeded all previous quarters combined.
The rest of AMZN also accelerated. North American sales rose 16%, International sales increased 15%, and advertising revenue jumped 26% to $19.8 billion. Faster delivery, grocery expansion, and AI-powered shopping tools are strengthening the retail platform while advertising adds another high-margin profit engine.
The cost is immense: 2026 capex forecast was raised from $200 billion to approximately $220 billion, partly because of higher memory costs. Moreover, trailing twelve-month free cash flow swung from positive $18.2 billion to negative $7.6 billion as property and equipment spending increased by $66.1 billion.
CEO Andy Jassy provided extremely strong and concrete defense for that spending. The company begins funding data centers roughly two years before revenue arrives, but those facilities can operate for more than 30 years. Servers and networking equipment generally break even in less than three years and remain productive for five to six years. Even after raising spending, AMZN expects capacity to remain insufficient through 2027. Most available 2027 capacity and meaningful 2028 capacity are already reserved, while AWS backlog jumped from $364 billion to $496 billion in one quarter.
That backlog carries a caveat: AMZN is helping finance major customers that are simultaneously financing its infrastructure buildout. The hyperscaler has invested $18 billion in Anthropic and could invest up to $33 billion in total, while Anthropic has committed more than $100 billion to AWS over the next decade. Amazon has also committed $50 billion to OpenAI, reportedly completing the remaining $35 billion tranche, while OpenAI has made $138 billion of AWS commitments. These arrangements can generate extraordinary returns for Amazon, but they also create concentration and funding-loop risk.
Q3 revenue guidance of $197-202 billion initially looked soft. However, Prime Day moved from Q3 last year into Q2 this year; excluding that timing shift, projected growth would be nearly four percentage points higher. Operating income guidance of $22.5-26.5 billion still implies powerful year-over-year growth.
Bottom line: Amazon is spending at a breathtaking rate, but AWS growth, margins, backlog, and capacity reservations show that the spending is already earning a return.
1
❖❖ Energy Transfer (ET) delivered a strong Q2, with core operating growth, record volumes, and favorable commodity conditions combining to produce a substantial earnings beat and guidance raise.
Net income attributable to partners reached $2.09 billion, while GAAP earnings came in at $0.59 per unit versus the $0.38 consensus. However, the headline EPS included unusually favorable market-sensitive benefits: roughly $387 million of incremental realized optimization and spread earnings, plus a $296 million year-over-year increase in unrealized commodity gains. Excluding those benefits, we estimate EPS would have been roughly $0.39 – still slightly above consensus. That distinction prevents the headline result from overstating ET’s repeatable earnings power without erasing the underlying beat.
Adjusted EBITDA rose 31% year-over-year to $5.07 billion, while adjusted distributable cash flow attributable to partners climbed 32% to $2.59 billion, producing excellent distribution coverage of roughly 2.2x. Unlike GAAP EPS, adjusted EBITDA excludes unrealized commodity gains, although it still includes the realized optimization benefits.
The underlying business was also strong. Excluding Sunoco and USA Compression, core segment EBITDA increased approximately 19%. NGL transportation and exports rose 13% and 25%, respectively, while crude-oil transportation and gathered gas volumes each reached new records. Approximately 90% of ET’s EBITDA remains fee-based, limiting its structural commodity exposure even though market volatility provided an extra boost this quarter.
Management raised full-year adjusted EBITDA guidance to $18.8-19.1 billion, increasing the midpoint by $550 million and implying approximately 18% year-over-year growth. Importantly, the new outlook assumes little additional volatility benefit in the second half, leaving room for upside if favorable conditions persist.
Energy Transfer also continues to build visible future growth. The Hugh Brinson Pipeline entered commercial service, while fully contracted ethane-export capacity is expanding at the Nederland terminal in Texas. Meanwhile, ET’s data center and power plant connections are targeting attractive 5-6x EBITDA build multiples.
The main caveat is capital intensity. ET now expects $5.6-5.9 billion of 2026 growth capex and anticipates annual spending above $5 billion through 2029. Management says project-return thresholds remain disciplined, but the heavy program limits near-term financial flexibility.
The stock surrendered most of its premarket surge and closed with a small gain, suggesting investors recognized the strong quarter but discounted the temporary optimization boost and long capital-spending runway. Still, the report supports the thesis: ET’s core assets are growing, coverage remains excellent, and today’s investments are backed by visible demand.
1
❖❖ Sterling Infrastructure (STRL) delivered an exceptional Q2, but the market punished the one imperfection in the report: the company’s fastest-growing acquired business carries substantially lower margins than its legacy site-development operations.
Revenue surged 90% year-over-year to $1.17 billion, including approximately 50% organic growth, while adjusted EPS more than doubled to $5.80 and came in well above consensus. Adjusted EBITDA rose 104% to $256.7 million, producing a strong 22% margin. First-half operating cash flow nearly doubled to $328 million, and Sterling ended June with $464.5 million in cash against $285.3 million of debt.
The core growth engine remained emphatically healthy. E-Infrastructure revenue jumped 192%, powered by data centers, semiconductor facilities, and other mission-critical projects. Legacy site-development revenue rose 111%, while CEC’s electrical-services revenue increased 140% from its pre-acquisition level. Signed backlog climbed 116% to $4.33 billion, including 50% organic growth, while 92% of E-Infrastructure backlog came from mission-critical work. More than $1.4 billion of high-probability future phases sits outside reported backlog, and management said some customer campuses could keep Sterling working for five to 12 years as they expand.
That visibility supported another substantial guidance raise. Sterling now expects 2026 revenue of $4.00-4.15 billion, adjusted EPS of $19.70-20.30, and adjusted EBITDA of $891-916 million. The new midpoints are meaningfully higher than the outlook issued after Q1 and imply year-over-year growth of 64%, 84%, and 79%, respectively.
The selloff centered on margins. Sterling lowered its full-year gross-margin outlook by 50 basis points to 25.0-25.5% and raised capex guidance from $100-110 million to $130-140 million. CEC produced an adjusted margin of roughly 11.4% in Q2, versus the high-20% range for legacy E-Infrastructure site development. Because CEC is growing much faster, it diluted the consolidated segment margin even though margins improved within both businesses. This is a genuine mix constraint, but it is not evidence of deteriorating execution. Management expects CEC’s margins to improve as it exits lower-margin legacy work and scales larger data-center projects, while the additional capex will expand capacity to meet demand Sterling already sees.
There are two items to monitor. First, $1.24 billion of the $1.28 billion in unsigned awards came from CEC and Stone Ridge. Therefore, almost all the difference between signed and combined backlog originated in the acquired businesses rather than legacy Sterling. Second, management expects a temporary lull in Q3 awards before activity strengthens in Q4 and Q1 2027. Actual conversion will need to confirm that this is merely a matter of customer timing.
STRL shares sank about 11% in regular trading and were only modestly higher after hours, showing that dip buyers have not yet stepped in. We will watch closely for signs that negative sentiment is becoming entrenched, particularly if the Q3 award lull or margin dilution persists longer than management expects. Still, the results do not warrant placing STRL Under Review. Organic growth, signed backlog, cash generation, operating execution, and guidance all strengthened sharply. The market identified a legitimate margin-mix issue, not a broken business; the Smart Investor Buy case remains firmly intact.
1
❖❖ Arista (ANET) reported a blowout Q2 that showed AI-networking demand accelerating faster than even bullish expectations, while profitability remained extraordinarily strong despite ongoing supply-chain pressures.
Revenue surged 37.7% year-over-year and 12.1% sequentially to a record $3.04 billion, comfortably above the $2.83 billion consensus and marking Arista’s first $3 billion quarter. Adjusted EPS jumped 40% to $1.02, versus $0.89 expected.
The earnings quality was equally impressive. Non-GAAP gross margin reached 63.4%, improving from 62.4% in Q1 and exceeding management’s outlook despite higher memory, silicon, and manufacturing costs. Adjusted operating margin expanded to 49.9%, crushing the guided 46-47% range and demonstrating exceptional operating leverage. Operating cash flow reached approximately $1.1 billion, while cash and marketable securities rose to $13.3 billion.
The outlook was even stronger. Arista guided Q3 revenue to approximately $3.3 billion, almost 12% above consensus and implying another 8.7% sequential increase. Adjusted EPS guidance of $1.06-1.08 also came in far above the roughly $0.92 expectation, while the projected 48-49% operating margin indicates that rapid growth is not coming at the expense of profitability.
Management raised full-year revenue guidance from $11.5 billion to approximately $12.6 billion, implying 40% year-over-year growth and representing a remarkable $2.1 billion increase from the company’s original Analyst Day target. The full-year adjusted operating-margin target was also raised to 48-49%, while gross-margin guidance remained at 62-64%.
AI remains the central driver, but growth is broadening. Arista now has more than 100 AI fabric customers, versus only four or five discussed in 2024, and continues to target at least $3.5 billion of AI fabric revenue this year. Its new 1.6-terabit platforms cover scale-up, scale-out, and scale-across networking, positioning the company across increasingly complex AI clusters. Campus revenue is expected to reach at least $1.25 billion, while core data-center, enterprise, and routing demand also contributed to the higher outlook.
Supply constraints remain the main risk. Purchase commitments have nearly tripled year-over-year to $9.7 billion, inventory reached $2.54 billion, and management expects industrywide component tightness to persist into 2028. However, Arista has secured its 2026 memory requirements, improved visibility into 2027, and expanded manufacturing and distribution capacity.
The market’s verdict speaks volumes. ANET sank roughly 13% after its Q1 beat as investors focused on supply costs and margin pressure. This time, the shares surged over 12% after hours to a record high, indicating that the combination of accelerating growth, better supply visibility, and near-50% operating margins decisively changed the perception of those risks. The gain may moderate in regular trading, but the initial response is unmistakable: investors see Arista strengthening its position at the center of the AI-infrastructure buildout.
1
Under Review
❖❖ We are removing Palantir Technologies (PLTR) from our “Under Review” bracket after the company’s earnings report proved that it deserves its place in the Smart Investor portfolio.
PLTR was placed under review because the stock had been underperforming despite continued operational strength. The Q2 report answered both sides of that concern: Palantir delivered an exceptional beat-and-raise quarter, and investors responded by sending the shares almost 30% higher in Tuesday trading. That surge more than repaired the portfolio position’s recent weakness, moving it from a loss back into a gain.
Revenue jumped 93% year-over-year and 19% sequentially to $1.94 billion, well above the consensus. Adjusted EPS reached $0.41 versus $0.34 expected, soaring more than 150% year-over-year for the second straight quarter. More importantly, the businesses at the center of our thesis accelerated sharply. U.S. commercial revenue surged 149% to $764 million, while U.S. government revenue rose 90% to $809 million. Palantir closed $3.37 billion of total contract value, including a record $2.13 billion in U.S. commercial TCV, showing that AIP demand is converting into large production commitments rather than remaining trapped in pilot projects.
Growth also came with extraordinary profitability. Adjusted operating income reached $1.19 billion, producing a 62% margin, while adjusted free cash flow rose to $1.22 billion – a 63% margin. That directly resolves one of our main review questions: Palantir is not merely buying growth or relying on distant promises to justify its valuation; its AI platform is already producing exceptional operating leverage and cash generation.
Management raised full-year revenue guidance by roughly $500 million to $8.150-8.158 billion, implying about 82% growth. U.S. commercial revenue is now expected to exceed $3.424 billion, up at least 134%, while adjusted free cash flow guidance increased to $4.5-4.7 billion. Q3 revenue guidance of $2.160-2.164 billion also came in comfortably above the roughly $2.0 billion consensus.
The valuation remains demanding, and international growth continues to lag the U.S. business. Those risks have not disappeared, nor does one huge rally guarantee a straight-line stock recovery. However, “Under Review” status was never meant to punish volatility; it was meant to determine whether operational momentum and market recognition still supported the holding. Q2 answered both questions decisively.
Palantir remains one of the clearest software beneficiaries of the AI cycle, with AIP adoption, government demand, agentic-AI deployment, expanding margins, and surging free cash flow all strengthening the thesis.
1
Portfolio Earnings and Dividend Calendar
❖ The Q2 2026 earnings season is past its peak, but several Smart Investor portfolio holdings are scheduled to reveal their quarterly results this week. Eli Lilly (LLY) and AppLovin (APP) will reveal their quarterly results today, while ATI (ATI), Nova (NVMI), Howmet Aerospace (HWM), and Parker Hannifin (PH) are scheduled for tomorrow, August 6.
❖ The ex-dividend date for Howmet Aerospace (HWM), Energy Transfer (ET), and Baker Hughes (BKR) is August 7.
w
1
New Buy: ServiceNow (NOW)
ServiceNow is positioning its cloud platform as the control layer for a new era of enterprise AI that moves beyond answering questions to taking action. That shift raises a harder problem – determining what autonomous agents may access, which systems they can change, and how their work is governed. Its technology already connects people, data, applications, and processes across complex organizations, turning requests into structured workflows spanning IT, employee services, customer operations, security, and risk. That installed position gives ServiceNow something newer AI entrants must still build – operational context and permissioned access to the systems where work happens. As companies deploy fleets of agents across fragmented technology estates, ServiceNow is expanding beyond workflow automation into the infrastructure that discovers, coordinates, secures, measures, and audits digital labor.
1
Once Upon IT
ServiceNow began with a then-unusual premise: enterprise software should be delivered through a web browser and be intuitive to use. Founded in 2004, the company initially developed a browser-based platform that it brought to market through IT service-management applications. NOW’s product replaced emails, spreadsheets, and disconnected ticketing tools with structured workflows, giving technology teams a shared system for handling incidents and requests.
The company went public in 2012 and gradually carried the same workflow logic beyond the IT help desk. Human resources, customer service, security operations, application development, and other corporate functions joined the platform, widening ServiceNow’s role from a departmental tool into an enterprise operating layer. From 2019 onward, NOW accelerated its push into larger customer relationships and industry-specific products, while positioning its platform as a common interface across legacy systems.
The next transformation was built around intelligence and visibility. The 2020 acquisition of Element AI added research talent and practical AI capabilities, while Lightstep brought cloud observability in 2021. The rise of generative AI then opened a new chapter. ServiceNow introduced Now Assist across its products, using AI to summarize information, generate content, and help employees complete tasks within existing workflows. As autonomous AI agents emerged, the company began extending that foundation toward systems capable of performing multi-step work on their own – marking the start of its agentic era.
Acquisitions in 2025 and 2026 added the capabilities needed to support that broader role. Moveworks added a conversational front door and enterprise search, while Logik.ai extended ServiceNow’s reach into sales and order management. Pyramid Analytics brought analytics directly into workflows. Veza added identity security, and the much larger Armis acquisition expanded visibility and protection across connected devices and cyber-physical assets.
In 2026, ServiceNow deepened its AI ecosystem through partnerships with OpenAI and Anthropic, bringing frontier models and specialized capabilities for multimodal interaction, application development, and industry workflows into the ServiceNow AI Platform. These agreements complemented its longstanding collaboration with NVIDIA on enterprise AI models and applications. Together, they expanded ServiceNow beyond its own embedded AI, creating an open, multi-model ecosystem in which leading external models could operate within governed enterprise workflows.
The company is now integrating these extended capabilities into the agentic strategy it has been building since 2024. NOW’s platform is expanding into a management layer through which companies can deploy and oversee AI agents, connect them with enterprise data and third-party systems, and control what they can access and do. The company that began by routing IT tickets is evolving into infrastructure for coordinating, authorizing, and auditing work performed by humans and machines across the enterprise.
1
Workflow in Progress
ServiceNow’s core advantage is that its platform is already embedded in enterprise work processes. Its configuration database maps technology asset relationships; its workflows encode approvals, responsibilities, and service histories; and its integrations connect systems that were never designed to cooperate. That context is essential for autonomous agents, which need more than a capable model to complete sensitive tasks safely.
The business begins with IT service and operations management, where NOW is deeply entrenched, but the same architecture now supports employee and customer service, CRM, security and risk, and custom application development. That role becomes more powerful in the age of agentic AI. LLMs can generate answers, but enterprises still need permissions, identity controls, system access, audit trails, compliance, workflow logic, and human oversight before agents can act inside core operations. NOW is building that governed execution layer – where work can be secured, routed, completed, and recorded. Jensen Huang captured the ambition clearly, calling ServiceNow “the operating system of enterprise AI agents.”
The moat begins with scale and architecture. Customers expand by adding workflows, users, and capabilities, with almost all recent large deals including at least eight products. NOW supports more than 100 billion workflows and 6.5 trillion annual transactions across thousands of organizations, including about 90% of the Fortune 500 and government agencies across nearly every U.S. state. This creates a deep base of process history, operational context, and customer trust. Its single-platform structure – one data model, one security model, and one user experience – gives AI agents more consistent rails than a patchwork of point solutions. Its open architecture lets customers use different models, clouds, data systems, and interfaces while ServiceNow governs and executes the work.
ServiceNow’s AI strategy follows the same logic: govern the work, then automate more of it. AI Control Tower gives customers visibility and control over NOW’s AI systems and third-party tools, while Action Fabric helps govern swarms of agents and enables automation with tighter oversight. Otto and EmployeeWorks bring these capabilities closer to employees, turning enterprise search, AI assistance, and internal requests into completed workflows. Together, these products move the platform from recording work to coordinating digital labor.
Data is becoming a second major growth pillar because agents are only as useful as the enterprise context behind them. Large companies today hold critical data across Snowflake, Databricks, and many other systems. ServiceNow turns those connections into context for action: what a request means, who has authority, which assets are affected, and which approvals or exceptions apply. This creates a richer decision layer for agents and expands NOW’s monetization opportunity as customers prepare fragmented data estates for automation.
Security and risk are a natural extension of this opportunity. As companies deploy more agents, bots, and autonomous workflows, they must govern nonhuman actors that can access data and make operational changes. Security therefore becomes a prerequisite for AI at scale, while Veza and Armis deepen NOW’s capabilities in identity, connected-device visibility, and cyber-physical protection.
ServiceNow’s Security and Risk business already exceeds $1 billion in ACV, while management said its security products appeared in 80% of the company’s largest Q2 deals. The August launch of six unified Autonomous Security offerings – most already available – also shows that Armis and Veza are being integrated into sellable products rather than remaining an acquisition story on paper.
As NOW expands beyond IT workflows, its addressable market is widening. Competition and execution risks remain, but the company is well positioned to benefit from the growth of AI governance. ServiceNow was the only vendor named a Leader in Gartner’s 2025 Magic Quadrant for AI Applications in IT Service Management, while its embedded analytics capabilities ranked first across all use cases in Gartner’s 2026 Critical Capabilities report.
1
NOW It Adds Up
ServiceNow’s Q2 results shattered the bearish narrative that AI is eroding demand for major enterprise platforms. Revenue rose 24% year-over-year to $3.99 billion, well above the consensus, while subscription revenue increased 24.5% to $3.88 billion, marking a fifth consecutive quarter of acceleration. The results exceeded the top of management’s guidance, although strong U.S. federal demand pulled some on-premise revenue forward from Q3 and made the quarter look slightly better than the underlying run rate. Some analysts also warned that the July 1 transition to AI-native pricing may have encouraged customers to renew early, potentially benefiting Q2 bookings at the expense of subsequent quarters. Q3 performance should help show how much of the apparent acceleration reflected timing.
Still, the backlog showed that demand extended beyond that timing benefit. Current remaining performance obligations reached $13.2 billion, up 21% year-over-year, while total RPO also rose 21% to $29 billion. ServiceNow signed 123 deals worth more than $1 million in net new annual contract value, nearly 40% more than a year earlier, and finished with 658 customers generating over $5 million in ACV, up 23%. Its 98% renewal rate remained intact.
AI is becoming commercially material inside those contracts. AI annual contract value crossed $1 billion, the number of customers with agentic AI in production increased ninefold in nine months, and the share of renewing customers buying agentic AI for the first time doubled both sequentially and year-over-year. Real deployments are producing measurable outcomes: Raleigh says ServiceNow agents now autonomously resolve nearly half of its IT requests, while Leidos is targeting up to 60% autonomous resolution across a much larger and highly regulated environment. These examples show the platform moving beyond pilots into day-to-day operations.
Profitability held up well despite heavy investment. Non-GAAP operating margin remained 29.5%, while adjusted EPS rose about 11% to $0.90, beating the consensus. Non-GAAP free cash flow increased 19% to $634 million after $161 million of business-combination and related costs were added back. Subscription gross margin slipped to 80.5% from 83% as increased customer use of NOW’s hyperscaler partnerships raised the third-party cloud costs NOW incurs, reflecting the near-term infrastructure cost of scaling AI adoption.
Management raised the lower end of full-year subscription revenue guidance to $15.76-15.78 billion, implying growth of 22.5%. It maintained targets for a 31.5% non-GAAP operating margin and 35% free-cash-flow margin. Q3 subscription guidance of $3.975-3.980 billion implies 20.5% growth and sits below consensus, partly reflecting the federal pull-forward. Meanwhile, cRPO is expected to grow 19.5% year-over-year despite a $35 million currency headwind, pointing to durable demand.
The 2030 targets outline a much larger version of ServiceNow’s existing model: over $30 billion in subscription revenue, 30% of ACV generated by AI, and combined subscription growth and free-cash-flow margin above 60%. Current AI momentum provides a credible starting point. Reaching the targets will require AI adoption to spread far beyond initial deployments, deepen customer relationships, and lift contract values while the platform continues producing substantial cash.
Armis should be viewed within that strategy. The $7.6 billion acquisition has increased leverage while giving ServiceNow a much deeper position in cybersecurity, connected-asset intelligence, and governance – capabilities that become increasingly valuable as AI agents gain access to enterprise systems. NOW’s recurring cash generation makes the financing manageable. The central question is whether ServiceNow can integrate Armis effectively, cross-sell its capabilities across the installed base, and turn the acquisition into a meaningful contributor to AI and security growth.
1
Change Request
Salesforce, Workday, Atlassian, and Adobe frame ServiceNow across the dimensions that matter most. Salesforce is the closest scaled enterprise-platform peer, with increasing overlap in customer workflows and agentic AI. Workday offers a clean comparison through mission-critical back-office applications and recurring subscriptions, while Atlassian competes more directly in IT service management and developer workflows. Adobe has less product overlap, but its comparable size, subscription model, margins, and exposure to AI-disruption fears make it a useful financial benchmark. Together, the group helps test whether NOW’s premium reflects superior growth and strategic importance or expectations that remain too demanding.
The software sell-off has been broad, though ServiceNow’s former premium made its reset especially deep. NOW fell about 35% over the past year, underperforming Salesforce, Workday, and Adobe, while holding up better than Atlassian. The year-to-date picture is less severe: NOW’s decline of about 23% is smaller than the drops in Salesforce, Atlassian, and Adobe, though Workday has performed slightly better. Investors initially treated AI as a threat to established software economics and added acquisition, leverage, and integration concerns to ServiceNow’s risk premium. Q2 results strongly challenged that view, reflecting accelerated subscription growth and a surge in AI contract value. Those results provide concrete support for continued recovery as the market distinguishes strategic platforms from vulnerable point solutions.
The stock has staged a strong rally over the past two weeks, but the prior reset has made NOW much more approachable, lowering its valuations by over half versus its 5-year averages, though it is still trading at a premium versus peers that saw similar stock declines. This premium is supported by a combination no peer matches: expected revenue and EPS growth of over 20% and a roughly 35% levered free-cash-flow margin. Basically, NOW’s fast expansion comes in parallel to converting roughly one-third of revenue into free cash flow – an unusual profile at its scale. That strength explains why the stock commands the group’s highest headline multiples, including 28.07x expected non-GAAP earnings and 7.28x forward sales. These remain demanding levels, though the reset has brought them much closer to the company’s operating reality. Even its group-high forward PEG ratio of 1.14 is moderate in absolute terms and well below the Technology sector median, showing that NOW’s valuation looks far less demanding when adjusted for growth.
ServiceNow’s superior growth, cash generation, and expanding role in enterprise AI provide a solid foundation for its premium over peers and leave room for a meaningful rerating. Wall Street sees considerable room for that thesis to play out, with the average price target implying a potential upside of over 22% for the Strong Buy-rated stock.
Management took advantage of the stock’s decline amid strong cash generation, repurchasing approximately $2.23 billion in Q1 and retiring as much as 1.3% of shares outstanding during the quarter. With the company subsequently directing capital toward the Armis acquisition, the remaining $4.2 billion authorization provides flexibility for future buybacks.
1
Investing Takeaway
ServiceNow is positioning itself as a control layer for enterprise AI – the platform through which intelligent systems can move from generating answers to completing real work securely, reliably, and accountably. Its workflows, permissions, data relationships, and audit trails are embedded across large organizations, giving NOW the operational context AI agents need to act inside critical systems. As enterprises move from AI pilots into deployment, ServiceNow’s opportunity should widen across automation, data, security, and governance, deepening customer relationships and driving earnings growth. Integrating acquisitions and absorbing the infrastructure costs of scaling AI will test execution, but adoption, retention, and cash generation provide room to invest. With the selloff resetting expectations just as AI monetization accelerates, the stock offers an attractive entry into a platform positioned to govern digital labor.
1
New Sell: MasTec (MTZ)
We are selling MasTec and locking in a gain of roughly 75%. This is a tactical exit from a company whose long-term investment thesis remains intact. The recent rebound alongside the broader risk-on rally and strength across infrastructure peers provides a better exit point, but no company-specific catalyst has emerged and the post-earnings estimate reset remains intact. After a rapid rally followed by an approximately 30% decline over the past month, MTZ remains in a market setup in which good fundamentals may no longer be enough to durably reverse entrenched negative sentiment.
The latest results were strong. Q2 revenue rose 23% year-over-year to a quarterly record of $4.37 billion, adjusted EBITDA increased 40% to $384 million, and adjusted EPS jumped 49% to $2.22. Adjusted EBITDA margin expanded by 100 basis points, while 18-month backlog climbed 30% to a record $21.4 billion, including $1.1 billion of sequential organic growth. Revenue, EBITDA, and EPS all exceeded management’s guidance.
The problem was the outlook beneath the headline raise. MasTec increased its full-year revenue forecast to $18.2 billion from $17.5 billion and adjusted EPS guidance to $9.30 from $8.79, but both remained below consensus estimates of $18.46 billion and $9.52. More importantly, management cut its Communications revenue outlook by approximately $400 million to $3.25 billion and now expects high-single-digit margins, about 100 basis points below last year. Wireless activity is slowing ahead of next year’s spectrum deployments, RDOF projects are rolling off, replacement wireline projects have been delayed, and some current projects suffered execution and cost pressures.
Those issues are manageable for the company, but damaging for the stock. MTZ had rallied too far and too fast, leaving no room for a segment-level disappointment. JPMorgan subsequently slashed its price target from $549 to $369 as it reduced estimates and peer multiples, despite retaining an Buy rating and calling MasTec’s valuation discount historically wide. JPMorgan was far from alone: seven other firms also cut their price targets following the report, most by meaningful amounts, confirming that the Communications shortfall triggered a broad reset in estimates and valuation assumptions even as all eight retained bullish ratings. The stock is now inexpensive, but cheapness alone is not a catalyst when estimates are falling and investor confidence has cracked.
Crucially, the original thesis is holding up. MasTec’s diversification allowed strength in Power Delivery, Pipeline Infrastructure, and Clean Energy & Infrastructure to absorb the Communications weakness. Clean Energy revenue surged 43%, Pipeline EBITDA nearly doubled, Power Delivery backlog reached another record, and most of the company’s $2.5 billion backlog growth over the past two quarters is expected to benefit 2027. The Superior acquisition also materially expands MasTec’s electrical and mission-critical data-center capabilities. Communications represents only about 15% of the company’s earnings power, according to JPMorgan.
There are credible future catalysts. MasTec expects additional large mission-critical awards during 2026, while Superior’s roughly $1.4 billion backlog will enter reported results in Q3. Management expects Superior cross-selling opportunities to become evident before year-end, and deferred Communications work could begin restarting as new spectrum equipment becomes available in 2027. Delivery of more than $1 billion in operating cash flow and leverage below 2x by year-end would also reduce acquisition-related concerns.
However, the next hard checkpoint is not expected until late-October earnings, and MTZ now needs a genuinely narrative-changing catalyst rather than another respectable report. Meanwhile, the Smart Investor portfolio retains substantial exposure to the same infrastructure cycle through several other stocks.
We will watch MasTec closely from the sidelines and remain ready to snap it back into the portfolio if a major mission-critical award, successful Superior integration, a Communications recovery, or a decisive earnings beat puts fundamentals back in charge of the stock performance.
1
Smart Investor’s Winners Club
The Winners Club represents stocks from the Smart Investor Portfolio that have risen at least 30% since their purchase dates.
Strong tech earnings and geopolitical optimism have propelled the markets to outstanding gains, and the Club ranks expanded again. With MTZ sold while NVT and CRDO are back, the Winners now count 25 stocks:
GE, AVGO, ANET, TSM, HWM, EME, APH, IBKR, CRWD, PH, ASX, VRT, PANW, GOOGL, CSCO, ATI, KEYS, BNY, RTX, MS, JBL, STRL, JPM, NVT, and CRDO.
The first runner-up is now NVDA with a 14.7% gain since purchase. Will it break into the Winners’ circle, or will another stock outrun it to the finish line?
1
|
