Intelligence Architects
In this edition of the Smart Investor newsletter, we spotlight a company designing the digital infrastructure of the intelligence age.
We are not selling any stocks today as conflicting headlines on geopolitics, macro, and markets obscure our view. Each stock in the Portfolio has been carefully analyzed and vetted, and we see short-term turbulence amid jittery sentiment as not a good enough reason to sell. We have seen this pattern across industries, from banks to chips: stocks get pulled down by this or that development, and then they soar even more – or at least those that can counter the turbulence with robust fundamentals and strong operations. It is rarely advisable to try to time the market; time in the market is what works.
Before we dive in, let’s review the latest Smart Portfolio developments.
1
1
Portfolio News and Updates
❖❖ Vertiv Holdings (VRT) is slated to be added to the S&P 500 on March 23, as part of the index’s quarterly rebalance. The inclusion follows strong fourth-quarter results that supported another leg in the stock’s rally, with a trailing twelve months gain of over 200%. VRT’s stellar operational performance and strong fundamentals have prompted multiple analyst price-target upgrades. Now, Vertiv’s stock is expected to enjoy another level of institutional support due to the automatic inflows from passive index funds following the S&P 500 index.
Prior to the index inclusion announcement, Vertiv updated on the successful completion of two major financing transactions that represent a significant refinancing and restructuring of its debt profile, leveraging its recent upgrade to investment-grade credit status by all major agencies (Moody’s, Standard & Poor’s, and Fitch). The company issued its first-ever investment-grade senior unsecured notes, raising a total of $2.1 billion. The notes were structured in four tranches with varying maturities, ranging from 10 to 40 years, with this staggered maturity profile helping to spread out repayment obligations over time, reducing refinancing risk in any single period. In addition, VRT closed a new $2.5 billion senior unsecured revolving credit facility with a five-year term, which replaces the company’s previous $800 million asset-based lending program. The new revolver can be expanded by up to an additional $1 billion under certain conditions, providing scalability, while its pricing – tied to Vertiv’s credit ratings – could lead to even lower interest costs if ratings improve further. The new refinancing is a milestone that strengthens VRT’s financial profile and provides ample liquidity to support continued growth.
In other significant news, on March 4, Vertiv announced a strategic collaboration with an infrastructure investment firm Generate Capital on a “Bring Your Own Power & Cooling” (BYOP&C) initiative. The BYOP&C partnership is intended to provide a complete solution for data centers across the U.S., tackling one of the biggest bottlenecks in the current data center boom. The initiative offers integrated, on-site power generation and cooling system that data center operators can deploy without – or while – waiting for full grid upgrades or utility approvals. It allows operators to “bring their own” reliable power and thermal management to get facilities online faster, while keeping the door open to eventually connect to (or transition toward) traditional utility grid power when it becomes available. Vertiv will provide integrated power and cooling infrastructure, while Generate will offer financial and operational management and support. This collaboration is expected to expand VRT’s addressable markets, strengthen its positioning in the AI infrastructure ecosystem, and drive recurring revenue through services and potential volume growth in modular solutions.
1
❖❖ GE Aerospace (GE) announced an additional $1 billion investment in its U.S. manufacturing facilities and supplier network this year, aimed at boosting its jet-engine output and accelerating the production of engine parts. The spending builds on a $1 billion commitment announced last year, with the extended capacity needed to speed up the delivery amid surging demand across both commercial aviation and defense programs.
In parallel, GE said it plans to invest $100 million in its external supplier network during 2026. This investment is aimed at helping suppliers purchase tooling and equipment with the intent to stabilize production and reduce bottlenecks across the supply chain.
1
❖❖ On March 6, 2026, President Trump signed an executive order focused on combating cybercrime, fraud, and predatory schemes by transnational criminal organizations (TCOs), often operating scam centers abroad. This is part of a broader push to protect American families, businesses, and critical infrastructure from cyber-enabled crime under the White House’s National Cyber Strategy. The EO signals stronger U.S. government focus on disrupting cybercrime networks, which could drive increased demand for advanced detection, endpoint protection, threat intelligence, response tools, and network security solutions provided by Smart Portfolio holdings CrowdStrike (CRWD) and Palo Alto Networks (PANW).
1
❖❖ Broadcom (AVGO) reported stellar sales and profit growth in its fiscal Q1 2026, and provided optimistic guidance. Total revenue jumped 29% year-over-year to a quarterly record of $19.3 billion, while adjusted EPS surged 28% to $2.05, both above analyst estimates. Adjusted EBITDA increased 30% year-over-year to a record $13.1 billion, representing 68% of revenue. Cash flow from operations soared 35% to $8.3 billion, and free cash flow was up 33% to $8.0 billion at quarter-end. The company also announced a new $10 billion share buyback program.
In FQ1, growth continued to be led by the Semiconductor Solutions segment with its sales up 52% and reaching 65% of total company revenue, while the Infrastructure Software segment, which makes up the rest, was broadly flat year-over-year. AI semiconductor revenue skyrocketed 106% year-over-year to $8.4 billion, and is expected to accelerate to $10.7 billion in FQ2, implying year-over-year growth of 140%. AI Networking revenue growth is slated to accelerate significantly in fiscal second quarter, reaching 40% of total AI revenue (up from 33% in FQ1), which implies year-over-year growth accelerating to broadly 145% from last quarter’s 60%.
This spectacular growth in Broadcom’s AI-related businesses underpins Broadcom’s total revenue guidance for FQ2, penciling in 47% year-over-year growth to approximately $22.0 billion. Meanwhile, the Infrastructure Software sales are expected to pick up modestly, rising about 9% year-over-year this quarter, led by VMware. The company expects its second-quarter EBITDA margin to remain at about 68%.
AVGO’s custom accelerator business grew 140% year-over-year in FQ1; the company expects this momentum to accelerate as its custom AI XPUs hit their next phase of deployment among the five existing customers – Alphabet’s (GOOGL) Google, Meta Platforms, ByteDance, Anthropic, and OpenAI – and as the sixth one (currently unnamed) comes on board. Additionally, Microsoft (MSFT) and Amazon (AMZN) reportedly work with the company to build custom AI chips tailored to their own models.
CEO Hock Tan said that Broadcom has secured line of sight to exceed $100 billion in AI chip revenue for fiscal 2027, reinforcing the company’s position as a major non-GPU AI supplier. At the same time, AVGO has secured the HBM and advanced-node manufacturing capacity at Taiwan Semiconductor (TSM) needed to support its AI chip production through 2028, easing concerns about potential supply shortages.
Multiple Wall Street firms responded by significantly raising price targets and reiterating Buy ratings, signaling growing institutional confidence in Broadcom’s industry position and growth prospects. AVGO’s new average PT implies an additional upside of over 35% even after an 80% rally over the past year.
1
❖❖ Taiwan Semiconductor, aka TSMC (TSM) announced that its combined revenue for January and February reached roughly $22.6 billion, up about 30% year-over-year, driven by surging global investments in AI infrastructure. For the coming full Q1, analyst estimates are anticipating growth of about 33% from the prior-year period. The chipmaker is accelerating construction of an AI-focused mega fab in southern Taiwan and is working to increase capacity and improve productivity to keep up with soaring demand, while benefiting from multi-year customer commitments. TSMC’s results are closely watched as an indicator of broader AI industry demand, as the company is the world’s primary foundry for the most advanced silicon. As such, its reports and announcements after the ongoing quarter will be under a magnifying glass, as analysts and investors will try to deduce the implications of the Iran war on AI-related investments and demand trends.
1
❖❖ ASE Technology (ASX), which stands hand-in-hand with TSMC in the AI-chip value chain, providing packaging and testing solutions for advanced silicon, has also reported strong revenue growth in February, with consolidated sales up 20.3% year-over-year. The company’s core ATM (assembly, testing, and material) segment saw particularly strong momentum with revenues surging 32.8% year-over-year, reflecting robust demand for advanced IC packaging tied to AI infrastructure expansion and increasing complexity in chip manufacturing.
1
❖❖ Morgan Stanley (MS) announced that it is laying off 3% of its global workforce, or about 2,500 employees, across all three of its primary divisions: Wealth Management, Investment Management, and Institutional Securities, which includes Investment Banking and Stock Trading segments. The cuts mostly affect back-office support staff worldwide and are communicated as a part of the effort to increase operational efficiencies amid shifting priorities, despite record 2025 revenues. This mirrors the ongoing trend across industries – including technology, finance, and others – where productivity gains from AI incorporation support plans to achieve the same or better level of output with far lower payroll. The investment bank’s stock has clocked in declines recently, but these have coincided with ongoing broad-market volatility and weakness in financials due to private-credit and macro concerns, obscuring the underlying sentiment.
Meanwhile, MS is actively pursuing expansion into secular high-growth areas like digital asset institutionalization and AI infrastructure buildout. These moves are aimed at expanding revenue streams in wealth management, investment management, and investment banking, while capitalizing on client demand for exposure to crypto and AI-enabling infrastructure. Two recent developments highlight this: an updated SEC filing for a Morgan Stanley-managed spot Bitcoin ETF – with Coinbase and BNY Mellon (BK) as the custodians – and a major financing deal of up to $1 billion with Core Scientific for AI data center expansion.
1
❖❖ Vertex Pharmaceuticals’ (VRTX) has seen a wave of analyst price-target hikes after the company announced overwhelmingly positive results from week-36 interim analysis of the Phase 3 RAINIER trial (a large, late-stage clinical trial) of povetacicept in IgA Nephropathy. IgA nephropathy is a common but serious chronic kidney disease, and the successful trial signals the probable addition of large revenue streams for the biotech giant; BofA models $2.9 billion in risk-adjusted peak sales. Vertex plans to complete its biologics license application by the end of March to seek accelerated FDA approval, potentially using a priority review voucher to shorten review time, so that the drug could be launched in the U.S. as soon as late 2026.
1
1
Portfolio Stocks Under Review
❖ We are keeping Oracle (ORCL) under review despite its stellar fiscal Q3 2026 results and strong outlook. The report delivered exactly the type of confirmation we had been waiting for – that Oracle is already monetizing the surge in AI infrastructure demand and that its enormous backlog of contracts is beginning to translate into tangible revenue growth. Equally important, the results clarified how Oracle is funding this expansion: through customer-funded infrastructure models that significantly reduce the company’s own cash burden. At the same time, given the still-volatile market backdrop, we believe it is prudent to allow sentiment around the stock to stabilize before making a final decision.
Operationally, the quarter was exceptionally strong. ORCL reported revenue of $17.19 billion, up 22% year-over-year, while adjusted EPS rose 21.8% to $1.79, both ahead of expectations. Most notably, this was the first quarter in more than 15 years in which both revenue and non-GAAP earnings grew by at least 20% simultaneously – a milestone that underscores how rapidly Oracle’s cloud transformation is accelerating.
The engine behind that growth is clearly AI-driven cloud demand. Total cloud revenue climbed 44% year-over-year to $8.9 billion, while Oracle Cloud Infrastructure (OCI) revenue surged 84% to $4.9 billion. RPOs reached an extraordinary $553 billion, highlighting the massive scale of contracted demand tied largely to AI workloads. Much of that demand is now being delivered under a “Bring Your Own Cloud” or customer-funded infrastructure model, in which clients provide upfront payments or supply hardware such as GPUs themselves. This structure materially reduces Oracle’s capital exposure while allowing it to continue expanding capacity aggressively.
Demand for AI computing also appears structurally stronger than supply. ORCL stated that cloud capacity for AI training and inferencing continues to be constrained globally, and management said the company expects to comfortably meet and likely exceed its fiscal 2027 revenue forecast of $90 billion. Several of the largest AI infrastructure customers have recently strengthened their own financial positions, which further reduces concerns about counterparty risk tied to these long-term contracts.
Another encouraging signal is that the AI infrastructure business itself is already profitable. Oracle reiterated expected gross margins of roughly 30-40% on AI accelerators, with additional profitability coming from adjacent cloud services such as storage, networking, and security. These services typically represent 10-20% of total customer spending and often carry higher margins. Meanwhile, the company’s multicloud database offerings – which grew rapidly in the quarter – deliver margins of 60-80%, strengthening the overall profitability profile of OCI.
For now, the primary factor weighing on margins is simply the pace of construction. Oracle is building multiple AI data centers simultaneously to keep up with demand, which means some costs are recognized before revenue from that capacity begins. However, many of these facilities are already backed by customer contracts and funding commitments, meaning the current negative free cash flow profile is largely a timing issue rather than a structural weakness in the economics of the AI infrastructure business. Management stressed that most of the infrastructure currently under construction is already contracted at profitable rates and should generate increasing returns as facilities come online.
The Street’s reaction to the report has been broadly positive. Analysts described the results as a “huge relief” after months of investor anxiety around Oracle’s aggressive AI spending. Several firms raised price targets, and JPMorgan upgraded the stock to Buy from Hold, arguing that the drop from last year’s highs has pushed sentiment from “blind faith to widespread pessimism,” notably improving ORCL’s risk/reward setup.
The fiscal Q3 report alleviated our concerns around Oracle’s debt levels and negative free cash flow during the AI infrastructure buildout, clearly supporting our bullish long-term thesis. The emerging customer-funded infrastructure model further strengthens that conclusion by shifting a meaningful portion of capital requirements away from ORCL’s own balance sheet.
However, markets remain highly volatile amid macro and geopolitical turbulence, and in such an environment, even fundamentally strong companies can see renewed volatility if risk-off sentiment returns. Our conviction that ORCL is emerging as one of the central infrastructure providers for the AI economy has strengthened with the latest data, which showed that the monetization phase of the company’s AI strategy is already underway. However, we prefer to observe how sentiment toward the stock evolves in the coming weeks, and whether market perception shifts decisively back in a positive direction after a prolonged period of skepticism.
1
❖ We are removing CrowdStrike (CRWD) from our “Under Review” bracket and moving it back into the regular Smart Portfolio holdings list. Recent developments have again stressed the indispensability of cyber defense that can withstand attacks by various hostile actors, from scams and phishing to state-sponsored cyber warfare targeting governments, critical infrastructure, energy, finance, and healthcare.
The war with Iran’s Islamist regime has underscored what for a while has been buried under heaps of advanced-tech rave: like generations before us, we aren’t living in a peaceful utopia – far from it. While kinetic combat may be successfully conducted thousands of miles away, cyber warfare is constantly ongoing on all digital levels around us. Iran has been widely recognized as a capable, persistent nation-state actor with a long history of disruptive attacks, espionage, and proxy operations – and yet, it is far weaker in its destructive capabilities than China or even Russia.
No AI agents, no matter how advanced, are able to replace endpoint security, hybrid-cloud protection, threat intelligence, zero trust systems, mission-critical systems security, and other components of defense protecting the U.S. and its allies from actors trying to sow chaos and inflict damage on all levels of society. On the contrary: the advance of artificial intelligence has accelerated the expansion of attack surfaces, amplifying threats – and making established and trusted cybersecurity platforms like CRWD all the more essential. As the realization of the hostile reality dawns on the markets, sentiment towards cyber experts aligns with it, supporting our decision to continue holding CrowdStrike in our Portfolio.
1
1
Portfolio Earnings and Dividend Calendar
❖ The earnings season is drawing to a close, with only one Smart Portfolio holding, Jabil (JBL), scheduled to report this week, releasing its fiscal Q2 2026 on March 18.
❖ The ex-dividend date for TSMC (TSM) and Vertiv Holdings (VRT) is March 17.
w
1
New Buy: NVIDIA (NVDA)
NVIDIA stands at the center of the modern computing era, supplying the advanced silicon and software platforms that power artificial intelligence, accelerated computing, and the world’s largest data centers. Its graphics processing units – originally designed for parallel workloads – have evolved into the industry’s dominant architecture for training and running AI models, making NVIDIA a foundational supplier to hyperscale cloud providers, enterprise developers, and research institutions. Around that hardware core, the company has built a broad ecosystem that spans networking, AI frameworks, simulation platforms, and full-stack data center solutions, embedding its technology deeply across the global compute infrastructure.
When not only financial markets but even the world map becomes volatile, capital often gravitates toward the industries that form the backbone of economic progress. A century ago it was railroads, fifty years ago oil, and today the strategic infrastructure of the digital economy is advanced silicon. In that landscape, NVIDIA functions less as a single chipmaker and more as a central architect of the AI computing stack that increasingly underpins modern industry.
1
The Silicon Age
NVIDIA’s story begins in 1993 in a modest Silicon Valley office, where three engineers – Jensen Huang, Chris Malachowsky, and Curtis Priem – set out to solve what looked like a narrow technical challenge: how to render complex graphics fast enough for the next generation of computer games. The solution was the graphics processing unit, or GPU – a chip built for massive parallel computation. Gaming became the proving ground, but the architecture would prove far more important than its founders initially imagined.
Through the 2000s, NVIDIA steadily expanded the reach of that technology beyond entertainment. The launch of the CUDA programming platform in 2006 allowed developers to harness GPUs for general-purpose computing – quietly laying the groundwork for a future dominated by data-intensive workloads. When machine learning researchers discovered that the same chips could dramatically accelerate neural networks, NVIDIA suddenly found itself supplying the essential engine of the emerging AI revolution.
Under the long-standing leadership of CEO Jensen Huang, the company moved decisively to capitalize on that moment. Rather than remaining a component supplier, NVIDIA built an integrated computing ecosystem around its chips – combining GPUs with high-speed networking, developer software, and full-stack platforms used by hyperscale cloud providers, enterprises, and research institutions. The result was a transformation from graphics specialist into the backbone of the global AI infrastructure buildout.
Strategic acquisitions and investments reinforced that position. NVIDIA integrated Mellanox networking to strengthen data center interconnects and acquired AI software platform Run:AI to improve resource orchestration across massive GPU clusters. Along the broader AI supply chain, the company has also invested in technologies aimed at preventing future bottlenecks. Recent positions linked to optical and photonics specialists such as Lumentum and Coherent reflect a push into advanced interconnects and optical networking – critical components for scaling next-generation AI systems. Parallel efforts in robotics and physical AI are expanding NVIDIA’s reach into autonomous machines and industrial automation, opening entirely new markets beyond the data center.
The pace of change has been extraordinary. As demand for AI infrastructure exploded, NVIDIA surged to become the world’s most valuable publicly traded company, generating enormous profits from the global compute cycle now underway. Strategic decisions increasingly reflect the shifting geopolitical landscape surrounding that technology. The company has recently curtailed shipments of H200 AI chips to China and redirected manufacturing capacity at TSMC toward its next-generation Vera Rubin architecture – aligning production with surging Western demand while navigating the escalating Cold War 2.0 between the United States and China over control of advanced technology.
From its gaming roots to its central role in powering artificial intelligence, NVIDIA’s rise illustrates a recurring pattern in economic history: when a new industrial era begins, the companies that supply its core infrastructure often become the architects of that era itself.
1
Emperor of Compute
NVIDIA operates today as the computational backbone of the artificial intelligence economy – the layer of infrastructure upon which much of the modern digital world increasingly depends. What began as a graphics chip designer has evolved into a full-stack accelerated computing platform that integrates processors, networking, software frameworks, and complete data-center systems into what the company now calls “AI factories.” These systems form the physical infrastructure behind generative AI, scientific computing, robotics development, and the emerging class of autonomous software agents.
The core of this ecosystem remains the GPU, but NVIDIA’s role extends far beyond silicon. Modern AI deployments increasingly rely on tightly integrated systems built around platforms such as Hopper, Blackwell, and the upcoming Vera Rubin architecture. These platforms combine GPUs with high-bandwidth memory, specialized CPUs, and high-speed interconnect technologies like NVLink and InfiniBand to create rack-scale computing systems capable of training and running the largest AI models. In practical terms, NVIDIA is no longer selling chips – it is selling entire computing architectures designed specifically for the age of artificial intelligence.
A critical part of this dominance lies in software. CUDA, the programming framework introduced in 2006, allows developers to harness GPU parallel processing for AI, simulation, and scientific workloads. Over time, it has grown into a massive software ecosystem used by millions of developers worldwide. This developer base represents one of NVIDIA’s most durable competitive advantages: applications built on CUDA naturally favor NVIDIA hardware, creating a powerful feedback loop between software adoption and hardware demand.
The result is an ecosystem that reaches across the entire AI value chain. Hyperscale cloud providers build massive GPU clusters powered by NVIDIA platforms. Enterprises deploy NVIDIA-accelerated systems to run AI workloads internally. Governments are building sovereign AI infrastructure around the same architecture to secure domestic data and technological capabilities.
In effect, NVIDIA sits beneath the modern AI stack in much the same way operating systems once underpinned personal computing – an invisible but indispensable layer powering a rapidly expanding digital economy.
1
The AI Coalition
If the first phase of the AI revolution established NVIDIA as the dominant infrastructure provider, the next phase is expanding the company’s reach into adjacent industries and entirely new technological frontiers. The rapid global build-out of AI infrastructure is creating opportunities that extend well beyond traditional semiconductor markets, from telecommunications networks and robotics to sovereign national computing systems.
One of the most important structural developments shaping this expansion is the emerging geopolitical split in global technology ecosystems. U.S. export restrictions have effectively closed China as a meaningful market for NVIDIA’s most advanced chips. Rather than slowing growth, the company has recalibrated around the Western AI ecosystem – a network of hyperscale cloud providers, enterprise AI deployments, and government-funded “sovereign AI” projects across North America, Europe, and allied economies. In practice, NVIDIA’s growth engine no longer depends on Chinese demand, reducing one of the biggest strategic uncertainties that once surrounded the company.
At the same time, the broader AI infrastructure market is expanding rapidly. Industry forecasts suggest that accelerator chips alone could grow into a market exceeding $600 billion over the next decade as computing architectures shift away from traditional CPUs toward specialized AI processors. Within this landscape, NVIDIA’s GPUs remain the dominant technology for training complex AI models, while demand for inference computing – running AI models at scale – is growing even faster.
Competition is intensifying, particularly from hyperscale cloud providers developing their own custom chips. Google’s TPUs and Amazon’s Trainium processors are designed to optimize internal AI workloads and reduce long-term computing costs. Yet the competitive landscape is increasingly viewed as complementary rather than adversarial. Most large AI systems today rely on a combination of silicon architectures, with NVIDIA GPUs continuing to serve as the most versatile and widely supported platform for training frontier models and running heterogeneous workloads.
Beyond cloud computing, NVIDIA is pushing its architecture into entirely new domains. Telecom operators are experimenting with GPU-powered AI-native radio networks, robotics developers are building physical AI systems using NVIDIA platforms, and emerging “agentic AI” software systems are creating demand for enormous inference capacity. In this environment, accelerated computing is spreading into industries that previously had little connection to high-performance silicon.
Taken together, these developments suggest that the AI infrastructure boom is unlikely to be a zero-sum contest between chip designers. Instead, the market itself is expanding so rapidly that multiple architectures will coexist – with NVIDIA positioned at the center of the largest and most mature ecosystem as artificial intelligence becomes embedded across the global economy.
1
Neural Net Worth
NVIDIA enters 2026 with financial results that reflect the extraordinary scale of the global AI infrastructure buildout. For fiscal 2026, which ended in January, the company generated $215.9 billion in revenue, representing 65% year-over-year growth and marking one of the fastest expansions ever recorded by a company of its size. The overwhelming driver was the Data Center segment, which produced roughly $193.7 billion in revenue, underscoring how central accelerated computing has become to modern AI development.
Profitability remains equally remarkable. Net income reached approximately $120 billion, pushing net margins above 50% – a level rarely seen in large-scale hardware businesses. Gross margins stabilized near 75%, supported by premium pricing for the company’s Blackwell-based AI systems and the growing contribution of higher-margin software and networking components embedded within NVIDIA’s platforms. Adjusted EPS surged to $4.77 in fiscal 2026, up 60% from the previous year.
Cash generation has reached historic levels as well. NVIDIA produced more than $90 billion in free cash flow during fiscal 2026, providing enormous financial flexibility to fund research, expand production capacity, and pursue strategic initiatives across emerging AI markets. The scale of these cash flows allows the company to invest aggressively in next-generation architectures while maintaining one of the strongest balance sheets in the semiconductor industry.
The underlying demand environment remains exceptional. Hyperscale cloud providers continue to deploy massive GPU clusters to support generative AI models, with some installations now spanning tens or even hundreds of thousands of GPUs. These deployments are part of a broader wave of global investment in AI infrastructure that analysts expect to reach multiple trillions of dollars by the end of the decade as computing architectures shift away from traditional CPUs toward specialized accelerated systems.
At the same time, NVIDIA’s revenue model is evolving beyond discrete chip sales. Increasingly, customers are purchasing rack-scale computing platforms that integrate GPUs, networking technologies such as NVLink and InfiniBand, and AI software frameworks into complete data-center systems. These deployments significantly increase the economic value of each installation while embedding NVIDIA more deeply into the infrastructure layer of AI.
The company’s near-term outlook reinforces how powerful that demand cycle has become. NVIDIA guided for approximately $78 billion in revenue for fiscal Q1 2027 (implying ~77% year-over-year growth), even while assuming no data-center compute revenue from China. In practical terms, management’s guidance suggests that growth is now being driven primarily by Western hyperscale cloud providers, enterprise AI deployments, and sovereign AI infrastructure programs.
For years, headlines about U.S. export restrictions on advanced AI chips to China frequently weighed on NVIDIA’s outlook. The latest guidance signals a structural shift: China is no longer a core pillar of the company’s financial model. Instead, the scale of AI investment across the U.S. and allied economies appears more than sufficient to sustain NVIDIA’s expansion on its own, highlighting both the magnitude of the current infrastructure cycle and the company’s central position within it.
1
Pricing the Kingdom
NVIDIA’s most relevant comparables are not traditional semiconductor peers alone but a small group of large-cap technology leaders positioned across the AI infrastructure stack. Advanced Micro Devices represents the closest direct silicon competitor, offering an alternative GPU platform for large-scale AI training and inference workloads. Broadcom reflects the rise of custom AI silicon and high-speed networking, supplying specialized accelerators and interconnect technologies that hyperscalers use to scale massive computing clusters. Amazon and Alphabet complete the comparison set as both customers and competitors – operators of the world’s largest cloud platforms that deploy vast fleets of NVIDIA GPUs while simultaneously developing their own AI processors such as Trainium and TPUs. Together, these companies define the technological ecosystem surrounding NVIDIA: hyperscale cloud infrastructure above it, specialized silicon beside it, and accelerated computing at its core.
Performance over the past year reflects both NVIDIA’s dominance and the extraordinary expectations surrounding the AI trade. The stock has risen more than 70% over the past 12 months, an exceptional gain for a company already valued in the trillions, yet it has actually lagged several of its closest AI infrastructure peers. Broadcom and Alphabet, in particular, delivered even stronger returns as investors rotated into different parts of the AI value chain. NVIDIA’s relative underperformance was driven by a steady stream of macro and geopolitical concerns that periodically weighed on sentiment. Over the past year alone, investors navigated fears of an AI bubble, questions about whether NVIDIA’s customers could sustain the capital required for massive GPU clusters, escalating U.S.–China technology tensions, and more recently a mix of geopolitical risks, inflation worries, and broader market volatility. Each headline created bouts of uncertainty around the sustainability of the AI infrastructure cycle. As most of those concerns have eased, however, attention has shifted back to the company’s underlying demand strength. With record revenue growth, accelerating global AI infrastructure investment, and China no longer central to the company’s outlook, analysts now see average upside potential approaching 47%, suggesting that NVIDIA’s stock may still be catching up to the scale of the technological shift it is helping to power.
Despite its dominant position in AI infrastructure and industry-leading profitability, NVIDIA’s stock now trades at multiples that look surprisingly reasonable, especially on a forward basis. Based on current estimates, NVIDIA trades at roughly 22x forward earnings on both GAAP and non-GAAP estimates, below every company in its comparison set. Even on trailing figures – where rapid growth naturally inflates multiples – NVIDIA’s 38x P/E is lower than AMD and Broadcom, despite its superior growth profile. NVIDIA’s past and projected revenue growth far outpaces all peers, while its profitability metrics remain unmatched – with return on equity exceeding 100%, placing the company in a league of its own even among large technology leaders.
On a sales basis, the company does command a premium – roughly 20x price-to-sales – reflecting its central role in the AI infrastructure cycle. Yet that premium appears far more reasonable when viewed alongside its growth profile and capital efficiency. Few companies combine NVIDIA’s revenue expansion, cash generation, and platform economics at similar scale. Meanwhile, growth-adjusted valuation reinforces the argument that the stock remains undervalued relative to its growth profile. NVIDIA’s forward PEG ratio near 0.6 sits well below peers, reflecting a rare combination of hyper-scale growth and extraordinary margins.
NVIDIA’s capital allocation policy reinforces the same picture of financial strength amid rapid growth. Its capital return model follows a clear hierarchy: reinvest heavily in AI infrastructure, research, and platform development, while returning excess cash primarily through large-scale share repurchases, with a small dividend (yielding roughly 0.02%) instituted to signal stability and confidence. In practice, buybacks dominate this strategy. In fiscal 2026 alone, NVIDIA repurchased approximately $40.1 billion of its own shares, representing about 1.7% of its massive total float. The scale of the program remains substantial, with $58.5 billion still available under existing repurchase authorization following a major expansion of the program in 2025.
Taken together, the picture that emerges is unusual for a company at the center of a historic technology buildout: extraordinary growth, industry-leading profitability, and valuation multiples that remain grounded relative to both peers and forward expectations. With capital returns accelerating alongside continued reinvestment in AI infrastructure, NVIDIA’s financial structure increasingly reflects a company scaling not only technologically but economically at the core of the global AI expansion.
1
Investing Takeaway
NVIDIA now underpins what is rapidly becoming one of the largest technological buildouts in human history. Artificial intelligence is no longer a speculative frontier – it is already hardening into global infrastructure, and NVIDIA’s platforms form one of its key pillars. From hyperscale data centers to sovereign AI projects and enterprise deployments, the computing backbone of this new economy is increasingly built on NVIDIA architecture. Competition is rising, but the broader trend points to expansion rather than displacement: the world simply needs far more compute than any single vendor can provide. In that environment, NVIDIA is not merely participating in the AI era – it is helping define it. For long-term investors, that role remains extraordinarily powerful.
1
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.
Markets have been tumultuous, but our Club has held up well, with the return of CRWD expanding the ranks of Winners to 20 stocks: GE, AVGO, HWM, TSM, ANET, EME, APH, VRT, PH, IBKR, MTZ, ORCL, GOOGL, KEYS, RTX, ATI, ASX, BK, CRWD, and CSCO.
The first in line to reenter the Club is still STRL with a 26.98% gain since purchase, followed by MS with 24.92%. Will they come back, or will another stock outrun them to the finish line?
1
1
|
