Light Within
In this edition of the Smart Investor newsletter, we spotlight the business moving AI at the speed of light. But first, let’s review the latest Smart Portfolio developments.
1
Portfolio News and Updates
❖❖ Palantir (PLTR) secured a foundational role in the U.S. Army’s Next Generation Command and Control program (NGC2), its highest-priority modernization effort and a core piece of the Pentagon’s broader Joint All-Domain Command and Control (JADC2) strategy. NGC2 is designed to fuse data across land, air, sea, space, and cyber for faster battlefield decisions.
As the program moves from prototype to scaling and production, the Army has selected Raft, Anduril, and Palantir as key architecture providers. Raft handles data registries, transformation tools, and federation, while Anduril’s Lattice platform serves as the tactical data and integration layer. Palantir’s Foundry – its enterprise AI and data platform – sits at the center of NGC2’s cloud data layer. In practice, future NGC2 applications, AI models, and battlefield systems built on this architecture will rely on Foundry.
The strategic read-through is significant. As NGC2 scales across more Army formations, Palantir gains a deeper platform role inside a long-cycle defense program. That strengthens its government AI franchise, expands recurring revenue potential, and reinforces its position as one of the few software platforms embedded in mission-critical military data infrastructure.
❖ In parallel, PLTR has deepened its partnership with NVIDIA (NVDA) through a Sovereign AI Operating System reference architecture designed for government and high-security commercial customers. The collaboration combines NVIDIA’s accelerated AI compute, Blackwell infrastructure, Spectrum-X networking, and Nemotron open models with Palantir’s AIP, Foundry, Ontology, and Apollo platforms. The goal is to let organizations train, customize, and deploy advanced AI agents on sensitive operational data while preserving control, security, and data sovereignty. This marks another step toward bringing frontier AI into high-stakes government and regulated enterprise environments where trust, compliance, and operational control are critical.
❖ In other news, Japan is reportedly evaluating Palantir’s Maven Smart System for the Self-Defense Forces’ command-and-control operations. According to media reports, Japan’s Defense Ministry is considering Maven as it prepares to revise key national security documents in 2026 to incorporate AI into military decision support.
❖ PLTR also continues to expand its enterprise footprint. In a recent development, Zeta Global – a leading AI-powered marketing cloud company – announced it is rearchitecting its entire Data Cloud on Palantir’s Foundry platform. The partnership creates a unified AI infrastructure that connects operational data, customer intelligence, and marketing execution, enabling faster deployment of AI-driven applications across the business.
Following the announcement, Wedbush analyst Dan Ives reiterated his Buy rating on Palantir and maintained a $230 price target, implying more than 100% upside from current levels. According to Ives, the deal demonstrates that Foundry is evolving into a foundational infrastructure layer for enterprise software platforms, reinforcing Palantir’s growing AI momentum beyond government and defense and strengthening its position in the broader enterprise AI buildout.
1
❖❖ Amazon’s (AMZN) four-day Prime Day event generated a record $26.4 billion in U.S. online sales, up 9.3% year-over-year. Discounts pushed consumers toward higher-ticket categories such as electronics, toys, appliances, and personal-care products, reinforcing Prime Day’s role as one of the biggest shopping events outside the holiday season.
The result highlights Amazon’s ability to anchor the broader e-commerce calendar while its business model continues to diversify across AWS, retail marketplace, logistics, advertising, and AI. That scale is driving bullish long-term forecasts, including expectations that AMZN could become the first company to cross $1 trillion in annual revenue by 2028.
1
❖❖ Leading U.S. financial institutions – including Smart Portfolio holdings JPMorgan (JPM), Citigroup (C), Morgan Stanley (MS), PNC Financial (PNC), and Bank of New York Mellon (BNY) – announced dividend increases and expanded capital-return plans after clearing the Federal Reserve’s 2026 stress tests. These moves are widely viewed as a sign of financial strength and have drawn analyst price-target increases across the sector.
JPMorgan plans to raise its quarterly dividend by 10% and authorized a new $50 billion buyback program. Citi announced a 12% dividend increase while continuing its $30 billion repurchase program. Morgan Stanley plans to lift its quarterly dividend by 15% and reauthorized a multi-year buyback of up to $20 billion. BNY said it intends to raise its quarterly dividend by 19% while continuing repurchases under its existing plan. PNC plans an 18% dividend increase and expects to step up buybacks in the second half of 2026.
1
❖❖ Alphabet (GOOGL) entered the Dow Jones Industrial Average on Monday, replacing Verizon and joining Microsoft (MSFT), Apple (AAPL), Amazon (AMZN), and Nvidia (NVDA) in the blue-chip index. S&P Dow Jones Indices said Alphabet’s addition strengthens the Dow’s exposure to dynamic areas of the U.S. economy. For Alphabet, the inclusion adds prestige and visibility, though the Dow’s price-weighted structure means the resulting index-fund demand is more limited than it would be for an S&P 500 addition.
❖ In other news, Google’s AI infrastructure is becoming so heavily utilized that even Meta reportedly could not secure all the Gemini capacity it sought. According to media reports, Google limited Meta’s access to Gemini models, delaying some internal AI projects and pushing Meta employees to use AI tokens more efficiently.
The report underscores demand for Google’s AI platform and suggests Google Cloud’s record $20 billion in Q1 revenue could have been even higher without capacity constraints. That makes Alphabet’s aggressive investment in AI chips and data centers look less like optional spending and more like the infrastructure needed to unlock demand already waiting in the pipeline.
1
❖❖ CrowdStrike (CRWD) will perform a 4-for-1 forward stock split after the close on July 1, with split-adjusted trading on Nasdaq beginning July 2.
1
Portfolio Earnings and Dividend Calendar
❖ The Q1 2026 earnings season has ended, and there are no reports scheduled for the Smart Investor Portfolio holdings until the Q2 season begins in mid-July.
❖ The ex-dividend date for GE Aerospace (GE), Cisco Systems (CSCO), JPMorgan Chase (JPM), and ASE Technology (ASX) is July 6.
w
1
New Buy: Fabrinet (FN)
Fabrinet operates in one of the most technically demanding layers of the global technology supply chain – manufacturing the precision optical, electro-mechanical, and electronic products that power modern communications and computing infrastructure. The company partners with many of the world’s leading OEMs, helping transform complex engineering designs into high-volume, production-ready products where accuracy, consistency, and reliability are essential. Its expertise extends from optical communications supporting AI-driven data center connectivity to industrial lasers, automotive technologies, medical devices, and advanced sensing applications. As data traffic accelerates, AI clusters expand, and optical networking becomes increasingly central to digital infrastructure, Fabrinet occupies a strategic position inside the manufacturing ecosystem – enabling many of the industry’s most advanced technologies while remaining deeply embedded in customers’ product development and production cycles.
1
Optical Intelligence
Fabrinet’s history is the story of a manufacturing specialist that grew alongside the optical communications industry. Founded in 2000, the company was built to serve technology customers whose products demanded precision manufacturing, clean-room capabilities, optical expertise, and close engineering collaboration. Its core role has always been to take complex designs from advanced equipment makers and make them reliably manufacturable at scale.
That early focus became the foundation of Fabrinet’s identity. Rather than chasing broad contract-manufacturing volume, the company concentrated on technically demanding products where process knowledge, yield discipline, and deep customer trust were critical. Its 2010 public listing provided capital for capacity expansion while preserving a model centered on engineering support, manufacturing precision, and long-term partnerships.
Over the following decade, FN expanded beyond telecom optics into adjacent markets such as automotive, industrial lasers, medical devices, and sensing. Yet optical communications remained the heart of the business. That positioning became far more valuable as cloud infrastructure entered the AI era. High-speed optical transceivers, active optical cables, and advanced interconnects evolved from specialized networking components into critical enablers of large-scale GPU clusters and dense AI data centers.
The past several years transformed that long-built capability into a much larger strategic role. Surging demand for AI infrastructure dramatically lifted Fabrinet’s datacom business, with major customers such as NVIDIA and Cisco highlighting its importance to next-generation networking and compute. Customer concentration carries risk, but it also reflects Fabrinet’s deep embedding in the value chain for the industry’s most advanced optical hardware transitions.
That positioning is now evolving into a broader opportunity. The 2025 agreement with Amazon Web Services added a major hyperscale customer, widening FN’s customer base while keeping the company firmly in its high-growth optical manufacturing lane. For Fabrinet, the win is significant: AWS is one of the defining buyers of AI infrastructure, and hyperscalers increasingly need partners capable of turning complex optical designs into reliable, high-volume production.
Fabrinet spent two decades building rare manufacturing depth in precision optics. Today, the industry’s biggest bottleneck has shifted directly into the area where the company is strongest.
1
The Swiss of AI
Fabrinet sits at one of the most critical junctions of the AI infrastructure buildout: manufacturing the optical hardware that allows increasingly powerful AI computing clusters to communicate at the speeds modern workloads demand.
The company does not design chips, switches, or networking equipment. Instead, it manufactures the highly complex optical products behind them for many of the industry’s leading technology companies, including Cisco, NVIDIA, Lumentum, and AWS.
One of Fabrinet’s biggest competitive advantages is its neutrality. The company manufactures its customers’ technologies, allowing leading networking and optical companies to outsource sensitive manufacturing without strengthening a potential competitor. This independence has become increasingly valuable as AI infrastructure grows more strategic and customers prioritize IP protection, dedicated capacity, and supply-chain resilience.
However, FN’s moat extends well beyond neutrality. Nearly 70% of its manufacturing footprint consists of clean-room facilities, reflecting the precision required for advanced optical products at commercial scale. For many programs, Fabrinet produces 60-70% of the bill of materials internally, giving it superior control over quality, yields, lead times, and costs. Long qualification cycles – often 18-24 months – further deepen customer relationships, turning manufacturing expertise itself into a durable competitive moat. The result is a business that competes on engineering execution, process consistency, and the ability to reliably manufacture products few others can scale.
Those capabilities are becoming dramatically more valuable. Industry researchers recently raised their Datacom market outlook, now expecting roughly 35% annual growth through 2028, driven primarily by 800G and emerging 1.6T optical networking. This demand flows directly into FN’s core optical communications business via optical transceivers, Data Center Interconnect (DCI) products, and sophisticated photonic assemblies. While DCI powers the networking that connects AI systems, high-performance computing (HPC) is emerging as a second AI-driven growth engine through increasingly complex compute-system manufacturing programs. New direct hyperscaler and merchant transceiver programs are further broadening the customer base across both opportunities.
The opportunity extends further. Silicon photonics, optical circuit switching, co-packaged optics, and advanced semiconductor packaging represent successive steps in the industry’s effort to move more data with greater bandwidth and lower power consumption as AI clusters become larger and denser. Fabrinet is positioning itself across these architectures through manufacturing investments, expanded packaging capabilities, and targeted partnerships – allowing it to participate regardless of which specific optical solutions win.
Supporting this strategy is a manufacturing footprint concentrated in Thailand, where FN has built deep engineering expertise, supplier relationships, and one of the industry’s largest pools of optical clean-room capacity. Ongoing capacity expansions provide multi-year runway, while its debt-free balance sheet enables disciplined investment funded largely by internal cash flow. That financial strength is itself a massive competitive advantage: customers planning multi-year AI deployments increasingly favor manufacturing partners with the financial capacity to expand alongside them.
Component shortages, customer concentration, and the pace of new capacity ramps remain the primary risks, alongside potential cyclicality in optical spending. Even so, as AI shifts the industry’s primary bottleneck from raw compute power toward efficiently moving massive volumes of data, Fabrinet is becoming an increasingly vital manufacturing partner in the optical infrastructure layer that makes large-scale AI possible.
1
Photonic Yields
Fabrinet’s financials increasingly reflect a company entering a new phase of scale. Revenue is accelerating, profitability is expanding, and management is financing one of the optical industry’s largest manufacturing expansions largely through internally generated cash. That combination is unusual even among AI infrastructure beneficiaries, allowing the company to invest aggressively without diluting shareholders or taking on debt while preserving one of the strongest balance sheets in the sector.
That profile was fully evident in fiscal Q3 2026. Revenue climbed 39% year-over-year to a record $1.21 billion, exceeding guidance and extending FN’s long-running pattern of top-line outperformance. Non-GAAP diluted EPS surged 47.6% year-over-year to a record $3.72, beating expectations for the fourteenth consecutive quarter, while gross profit, operating income, and net income all reached new highs. Trailing twelve-month revenue now exceeds $4.2 billion, illustrating the rapid expansion tied to AI infrastructure investment.
Profitability is scaling faster than revenue, driven by a favorable business mix. Optical Communications, which generates roughly three-quarters of revenue, remains the core. Within that segment, Telecom provides a stable revenue base, while Datacom – including high-speed optical transceivers and DCI products – has become one of the company’s fastest-growing businesses as hyperscalers continue expanding AI clusters. Outside Optical Communications, HPC is contributing a growing share of business through complex electronic assemblies for AI compute platforms, helping offset softer automotive demand.
Operating leverage remains one of FN’s defining financial characteristics: operating expenses represented just 1.4% of revenue during the quarter, allowing earnings to compound faster than sales despite continued investment in manufacturing capacity. Non-GAAP gross margin remained above 12%, with only modest pressure from forex movements and rapid program ramps. Fabrinet continues prioritizing long-term customer relationships and cost competitiveness over short-term margin maximization.
The balance sheet is a competitive advantage in its own right. Fabrinet finished the quarter with nearly $950 million in cash and short-term investments and no long-term debt, providing ample flexibility to fund expansion without interest expense, refinancing risk, or shareholder dilution.
Cash flow reflects FN’s current investment phase. Operating cash flow remained healthy, while free cash flow turned modestly negative during the quarter as capex accelerated and working capital grew to support larger production ramps. Inventory and receivables both expanded alongside shipments, illustrating that cash is being deployed to capture future growth.
The centerpiece of investment is Building 10, the new flagship facility in Thailand. Already partially operational and expected to ramp to full capacity in the coming quarters, it is projected to deliver roughly 40% return on invested capital once fully utilized. This expansion represents the physical embodiment of Fabrinet’s next growth phase in optical communications and AI-related manufacturing. In an industry where manufacturing capacity has become a key constraint, Building 10 materially expands one of Fabrinet’s most valuable competitive assets.
Guidance signals continued momentum. Management expects fiscal Q4 revenue of $1.25-1.29 billion (roughly 40% year-over-year growth at the midpoint), with non-GAAP EPS guided to $3.72-3.87 (up about 43% at the midpoint); both are above Wall Street expectations. Management anticipates growth across all major categories, with new hyperscale and merchant Datacom programs contributing more meaningfully in fiscal 2027. The principal challenge now is execution – converting new capacity, improving component availability, and expanding customer programs into the next phase of profitable growth.
1
Multiple Optics
Fabrinet operates within the EMS industry, but its business differs materially from most traditional contract manufacturers. FN is a specialist, excelling at the high-precision optical side of the supply chain. Jabil – a Smart Portfolio holding – provides the broad-scale benchmark as the diversified industry leader, serving end markets ranging from cloud infrastructure to healthcare and automotive. Celestica provides another strong AI data-center manufacturing comparison, while Sanmina and TTM Technologies offer useful benchmarks for complex, high-reliability electronics production.
The entire peer group has benefited from the AI infrastructure investment cycle over the past year, but the magnitude of returns reflects more than AI exposure alone. Jabil delivered the most measured performance, up about 70% – consistent with its diversified business model and lower sensitivity to any single technology trend. Celestica and Sanmina enjoyed much sharper gains as investors increasingly recognized their roles in AI data-center manufacturing, while TTM Technologies emerged as the standout performer after the market aggressively rerated its advanced PCB business, which sits at another critical bottleneck in AI hardware.
While Fabrinet’s stock gained an impressive 80%, its performance has largely tracked the company’s accelerating earnings, as investors remained focused on temporary supply constraints and an unusually heavy capital investment cycle. As new capacity comes online and hyperscaler programs scale during fiscal 2027, FN appears well-positioned to enter a second phase of its AI story – one driven by another step-up in earnings growth.
Wall Street appears to share that view, with the consensus price target implying more than 35% upside from current levels. Valuation also looks more balanced than it did several months ago. Fabrinet still trades at a premium to most EMS peers, changing hands at roughly 38x forward non-GAAP earnings for fiscal 2027. However, consensus expects earnings to grow rapidly enough that the multiple falls to about 31x the following year and below 25x by fiscal 2029, illustrating how much of the company’s anticipated earnings expansion is already embedded in current expectations.
Relative to peers, FN’s valuation looks elevated but defensible. While cheaper than TTM Technologies across the board, it trades above Jabil, Sanmina, and Celestica on most forward earnings and EBITDA multiples, though the gap narrows sharply by FY3 as consensus estimates increasingly factor in accelerating earnings growth. The premium is supported by FN’s higher EBIT and net margins than every peer in the group, its debt-free balance sheet, and strong forward revenue and earnings growth. That growth also translates into a reasonable forward PEG ratio of about 1.5, which looks increasingly attractive for a company combining accelerating earnings, durable competitive advantages, and direct exposure to one of the fastest-growing layers of AI infrastructure.
1
Investing Takeaway
Fabrinet has become one of the most important enablers of the AI infrastructure buildout. As computing clusters grow larger and more powerful, moving data efficiently is becoming as critical as generating it, placing advanced optical manufacturing at the center of the next phase of AI investment. Few companies combine Fabrinet’s manufacturing expertise, customer trust, financial discipline, and ability to scale alongside the industry’s largest technology players. The current investment cycle carries execution risks, particularly around supply availability and capacity ramps, but those risks appear tied more to timing than demand. With major expansion projects nearing completion and new hyperscaler programs beginning to ramp, Fabrinet looks well-positioned to translate its manufacturing leadership into another phase of earnings growth, making the stock an attractive long-term way to participate in the expanding optical backbone of AI.
1
New Sell: Philip Morris (PM)
We are selling Philip Morris not because the business has weakened, but because the stock appears to have reached a valuation ceiling that even excellent execution is struggling to overcome. The investment thesis has largely played out. PM remains the clear global leader in smoke-free nicotine, with IQOS and ZYN continuing to gain market share, margins expanding, and earnings consistently beating expectations. First-quarter results reinforced that strength, with adjusted EPS rising 16%, smoke-free revenue and profit growing at double-digit rates, and full-year guidance reaffirmed.
The problem lies elsewhere. Despite another quarter of outstanding execution, the shares have advanced only about 16% since we added them seven months ago. More importantly, the stock has repeatedly failed to respond to news that, under different circumstances, would likely have driven a meaningful rerating.
The clearest example came this week, when the FDA granted the first-ever Modified Risk Tobacco Product orders for nicotine pouches, allowing ZYN to market FDA-authorized claims that switching from cigarettes reduces the risk of several smoking-related diseases. For Philip Morris, whose long-term strategy revolves around smoke-free products, this represents a significant regulatory milestone and further validates years of investment in scientific research. Yet the shares barely reacted, instead inching down alongside the broader Consumer Staples sector.
That price action reflects the market’s current perception. PM already trades at a substantial premium to traditional tobacco peers and at elevated multiples relative to the broader Consumer Staples sector. While the company deserves a premium thanks to its superior growth prospects and leadership in reduced-risk products, investors appear increasingly unwilling to expand that premium further. At more than 25x forward earnings and roughly 7.7x forward EV/Sales, much of the good news already seems priced in.
Philip Morris remains an outstanding company, and we continue to believe its long-term strategy is succeeding. But when exceptional fundamentals no longer translate into meaningful stock appreciation, the risk-reward balance changes. We believe capital can be deployed more effectively in companies where both business performance and investor sentiment offer greater upside potential.
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.
The markets were extremely volatile, but the Club member count remained steady with 28: GE, AVGO, TSM, ANET, HWM, APH, EME, VRT, IBKR, ASX, MTZ, STRL, PH, CRWD, PANW, GOOGL, CSCO, KEYS, ATI, JBL, ORCL, BNY, MS, CRDO, NVT, RTX, C, and JPM.
The first runner-up is now LLY with a 21.29% gain since its purchase just two months ago. Will it break into the winners’ circle, or will another stock outrun it to the finish line?
1
|
