The AI computing power narrative of 2026 is undergoing a quiet yet significant structural divergence. While the market remains fixated on GPU chip designers, capital flows have already shifted in a notable way. As of June 12, 2026, Applied Materials (AMAT) has surged approximately 121% year-to-date, while Lumentum Holdings (LITE) is up about 67%. In comparison, Nvidia (NVDA) has risen roughly 44% over the same period. These two AI infrastructure stocks have not only outperformed Nvidia but have also significantly outpaced the S&P 500 index during this timeframe.
This data points to a critical trend: the expansion of AI computing power is moving beyond the question of "who designs the chips" to "who provides the equipment for chip manufacturing" and "who solves the physical interconnect bottlenecks between chips." Equipment and photonics have emerged as the two most structurally promising segments in current AI infrastructure investment.
Equipment Segment: The Performance Logic Behind AMAT
Applied Materials is one of the world’s largest suppliers of semiconductor manufacturing equipment, holding leading market shares in key process areas such as thin film deposition, etching, ion implantation, and metrology. As of June 11, 2026, AMAT’s stock price closed at $552.64, with a 52-week range from $154.47 to a record high of $557.62. On June 11 alone, the stock jumped about 11.2%, reflecting strong market expectations for equipment demand.
This rally was directly fueled by the company’s record-breaking financial results for Q2 of fiscal year 2026. The quarter saw revenue reach $7.91 billion, with earnings per share also hitting all-time highs. The company raised its 2026 calendar year semiconductor equipment market growth forecast from over 20% to over 30%. CEO Gary Dickerson attributed the accelerated growth to "the rapid build-out of global AI computing infrastructure" and the company’s leadership in advanced logic chips, DRAM, and advanced packaging.
The robust demand for equipment is supported by broader macro data. The World Semiconductor Trade Statistics organization sharply revised its spring forecast, projecting the global semiconductor market to soar by 90% in 2026 to $1.51 trillion. Memory chip sales are expected to jump about 250% year-over-year, breaking $800 billion. This explosive growth is directly driving demand across the entire equipment supply chain, from etching and thin film deposition to advanced packaging and testing.
Wall Street analysts have echoed this logic. Cantor Fitzgerald analyst C.J. Muse raised AMAT’s price target from $575 to $650—the highest on Wall Street—citing a wafer fab equipment industry in a supply-constrained, multi-year upcycle, with order visibility extending to 2028. Barclays lifted its target from $500 to $590 and raised its 2027 wafer fab equipment market size estimate to $209.5 billion. UBS increased its target from $515 to $570, reiterating a buy rating.
On the capacity expansion front, the company invested $500 million in a new campus in Tampines, Singapore, more than doubling its advanced cleanroom capacity. Additionally, AMAT and Nvidia jointly participated in a $300 million Series C funding round for PhysicsX, an AI and physics simulation startup, signaling AMAT’s strategic push into the AI software ecosystem.
However, valuation pressures must also be considered. According to GuruFocus, AMAT’s current P/E ratio stands at about 51.9x, significantly above its five-year median of 20.4x—a premium of roughly 155%. The GF Value model indicates the stock is notably overvalued. Whether earnings growth can continue to justify this valuation remains a key variable to watch.
Photonic Chip Segment: The Supply-Demand Gap Story of Lumentum
If the equipment segment answers "how AI chips are made," the photonic chip segment focuses on "how AI chips communicate with each other." In AI clusters with tens of thousands or even hundreds of thousands of GPUs, interconnect efficiency is increasingly becoming the main bottleneck for system performance.
As of June 12, 2026, Lumentum Holdings traded at about $863, with a 52-week range from $80.39 to $1,085.68. Its trailing P/E ratio is around 156.82, and its price-to-book ratio is about 23.20. Among 24 analysts, the consensus rating is "Buy," with 19 recommending buy and none recommending sell. The 12-month average target price is $1,113.01, implying roughly 28.87% upside from current levels. On June 12, J.P. Morgan reiterated its overweight rating with a $1,130 target, noting that recent valuation pullbacks in optical communications stocks offer investors a safer entry point.
Fundamentally, Lumentum delivered results far exceeding expectations in Q3 of fiscal 2026. Quarterly net revenue reached about $808.4 million, up about 90% year-over-year. GAAP EPS was $1.50, and non-GAAP EPS was $2.37, beating the market’s $2.27 estimate. Component business revenue was around $533.3 million, up 77.3% year-over-year, while systems revenue hit $275.1 million, up 121.1%. Its 1.6T transceivers have entered mass production.
Supply-demand imbalance is the key to understanding Lumentum’s current market position. Industry estimates suggest a data center laser chip supply gap exceeding 30%. Narrow linewidth laser shipments grew about 120% year-over-year, and pump laser shipments rose about 80%. Market data shows that leading global photonic chip makers have already sold out their capacity through the end of 2027. Nvidia has made strategic investments of about $2 billion each in Coherent and Lumentum, locking in long-term supply agreements for critical components. These signals all point to a shift: high-end photonic chips are moving from optional to essential, and from abundant to scarce.
Lumentum’s mid- to long-term growth narrative is advancing on two fronts. CPO (co-packaged optics) technology is seen as the core for next-gen AI cluster interconnects, with the company expecting CPO to unlock over $5 billion in incremental revenue opportunities. CEO statements suggest the market potential for NPO (near-packaged optics) could be even larger, as NPO deployments require more optical channels per rack—creating a favorable product mix for Lumentum as a laser supplier. The company also recently acquired its fifth indium phosphide wafer fab, aiming for mass production by 2028. This vertical integration strategy helps establish supply-side barriers in core photonic components.
Industrial policies in China and the EU are also providing macro-level structural support. On June 10, 2026, China’s Ministry of Industry and Information Technology issued the "AI + Information and Communication" Innovation Development Implementation Guidelines (2026–2028), calling for enhanced R&D of high-end optoelectronic chips and devices, including high-speed optoelectronic chips, all-optical switching devices, and optoelectronic co-packaged devices. The EU’s CHIPS Act 2.0 also recognizes photonic technology as a structural policy focus, explicitly supporting the development of photonic integrated circuits and advanced design capabilities. These policies offer sustained momentum for the long-term growth of the photonic chip sector.
From GPUs to Bottleneck Assets: The Logic Behind Institutional Allocation Shifts
The performance of these stocks is not an isolated phenomenon, but rather a reflection of a broader shift in institutional AI investment strategies. Recent 13F filings show that among 327 institutions with over $10 billion in assets, AI computing power allocations have moved from "buying the most visible AI heavyweights" to "reprioritizing based on bottlenecks." New incremental allocations are increasingly focused on power infrastructure, cooling, Ethernet switches, optical modules, HBM, advanced equipment, and yield management. NVDA and AVGO remain core holdings, but the marginal increase in allocation is clearly targeting the physical constraints most likely to limit AI cluster expansion.
Morgan Stanley and Bank of America recently released a joint 2026 semiconductor investment outlook, highlighting semiconductor equipment spending and the data center optical interconnect supply chain as their top investment themes. Bank of America’s report states that the global AI arms race is still in the "early to mid-stage," with wafer fab equipment sales expected to see high single- to double-digit year-over-year growth. Morgan Stanley projects the global chip testing and sorting equipment market will grow from $400 million in 2023 to $6.6 billion in 2027—a more than tenfold increase.
Technological advancements are also accelerating. TSMC is actively advancing its silicon photonics advanced packaging platform, COUPE, which is set to enter mass production in 2026. According to TSMC’s advanced packaging integration team, COUPE has evolved from a lab concept into a fully scalable semiconductor process technology. As the industry’s three- to five-year technology roadmap and direction become clearer, the inflection point for mass adoption of silicon photonics is approaching.
Global CPO (co-packaged optics) technology for optical interconnects is also expected to see large-scale deployment by 2027. Industry research firm LightCounting has sharply raised its shipment forecast for 1.6T CPO products, boosting the 2029 estimate from about 2 million units to around 9 million. In terms of market size, the CPO market is expected to exceed $5 billion by 2027 and continue growing through 2030.
In the foreseeable future, every order-of-magnitude increase in AI cluster scale will drive superlinear growth in demand for optical interconnects and advanced equipment. As clusters scale from tens of thousands to hundreds of thousands of GPUs, equipment demand rises by 5–7 times, while total optical interconnect bandwidth demand can grow by over 10 times. This elasticity gives equipment and optical interconnect segments a structural advantage in AI capital expenditures.
Key risks to monitor include several dimensions. The semiconductor equipment industry is historically cyclical, so today’s strong demand must be matched by future capacity absorption. On valuations, AMAT’s current P/E is well above its historical average, and a slowdown in earnings growth could trigger valuation pressure. In optical interconnects, both CPO and NPO technologies are still transitioning from introduction to mass adoption, and uncertainties in technology roadmaps and customer adoption pace remain variables to watch closely.
This is a dedicated section to be embedded above, focusing on Gate’s stock trading features, core advantages, and specific operational steps.
How Gate’s Real US Stock Trading Extends Allocation Pathways
On June 1, 2026, Gate officially launched its real stock trading service, providing users with a bridge to allocate both cryptocurrencies and US stocks within a single account system. As of June 2026, Gate offers over 10,000 real stocks and ETFs, covering all five major exchanges including the NYSE and NASDAQ. This means that users can directly purchase Applied Materials (AMAT) and Lumentum (LITE) using USDT from their Gate accounts, with no need to open an additional brokerage account or go through the traditional processes of selling crypto, withdrawing funds, cross-border remittance, or currency conversion.
Gate’s real stock trading platform stands out in three key areas.
First, asset authenticity and regulatory compliance. Unlike common stock tokenization or RWA-mapped products, Gate partners directly with Alpaca, a US broker-dealer licensed for clearing, ensuring that every share purchased on Gate is backed by real assets independently custodied within the DTC system. Alpaca is also a member of the Securities Investor Protection Corporation (SIPC), providing eligible protection for client securities under applicable conditions. This regulatory framework means that holding real stocks like AMAT, NVDA, or Lumentum on Gate grants users the same rights as with a traditional brokerage account—including cash dividends, stock dividends, and participation in corporate actions such as rights issues, stock splits, and bonus shares.
Second, seamless USDT settlement eliminates funding barriers. On Gate, users can buy real US stocks directly with USDT liquidity in their accounts, and the entire transaction can be completed in seconds. For crypto users who hold USDT long-term, this transforms US stock allocation from a process requiring extra account setup into a one-click, diversified portfolio option.
Third, fractional share trading lowers the entry threshold. With AMAT trading above $557 in 2026 and NVDA and Lumentum also at high price levels, traditional whole-share trading rules create a significant barrier for users with limited capital. Gate’s fractional share feature solves this: users can start with as little as 0.01 shares—just $1—to gain exposure to leading US stocks like AMAT, NVDA, and Lumentum. This allows users to flexibly build diversified AI computing portfolios across equipment, optical interconnect, and GPU design stocks according to their risk preferences, without waiting to accumulate enough funds for whole shares.
The process for buying US stocks with USDT on Gate involves four steps: hold or acquire USDT in your Gate account; navigate to the stock trading section via the Gate interface; transfer USDT from your spot or unified account to your stock account; and during US market hours, search for your target stock and place a buy order. Note that Gate Stocks spot products currently do not support short selling or margin trading. For leveraged exposure, users can consider Gate’s US stock contract products, but leverage amplifies both gains and losses, so users should carefully assess their risk tolerance before trading.
For users focused on long-term allocation, Gate’s real stock spot trading does not involve swap fees or overnight holding costs typical of CFD trading. The cost structure of holding real assets is more transparent, making it better suited for strategies centered on AI infrastructure stocks like those discussed above.
In June 2026, Gate also launched the "Stock Express" campaign for new users and the invitation-only "US Stock Glory Gala," further lowering the initial barrier to US stock allocation. As crypto platforms evolve from "single-asset exchanges" to "multi-asset allocation platforms," structural opportunities in AI infrastructure are moving from theoretical analysis to real, actionable investment options.
Conclusion
The AI investment narrative in 2026 is shifting from "chip designers" to "chip manufacturers" and "chip interconnect providers." AMAT’s 121% gain and Lumentum’s 67% surge reflect a market that is repricing the structural bottlenecks in AI computing power expansion. Semiconductor equipment and photonic chips share a common trait: they don’t compete directly in GPU design but provide the essential manufacturing and interconnect capabilities that underpin the entire AI computing ecosystem. In the era of clusters with tens or hundreds of thousands of GPUs, these are precisely the segments with the greatest pricing power.
For investors focused on the long-term evolution of AI infrastructure, expanding your view beyond GPUs to the bottlenecks in equipment and interconnects may offer deeper insight into industry cycles than simply tracking GPU designers’ quarterly shipments.
