AI Networking & Interconnect

Nvidia has committed at least $6.5 billion to photonics companies over the three months prior to May 2026, including $500 million to Corning for advanced optical connectivity solutions and $2 billion each to Lumentum, Coherent and Marvell, according to CNBC. The impact has been significant, with Marvell shares gaining 183.4% year to date, while Lumentum, Corning and Coherent have advanced 142.5%, 68.8% and 60.3%, respectively. Goldman Sachs projects the total addressable market for optical networking to grow from roughly $15 billion in 2026 to about $154 billion by 2028, a more than ninefold increase. Adoption of co-packaged optics is expected to accelerate starting in 2027, reaching roughly 20% to 30% penetration by 2028, while Mordor Intelligence estimates the global photonics market will reach $1.46 trillion by the end of 2026. Investors seeking diversified exposure can consider the Roundhill Photonics & Optics ETF LYTE, with $254.2 million in assets under management across 19 holdings; the KraneShares Photonic and Optical ETF LUMA, with $9.76 million in net assets; and the Tema Photonics & Optical ETF LAZR, with $94 million in net assets across 27 companies.

Amphenol's IT datacom business, which represented 43% of the company's total sales in the second quarter of 2026, saw revenues surge 89% year over year and 63% organically, with management saying the AI revenue run rate has reached roughly $10.5-$11 billion. Management noted that virtually all of the 22% sequential IT datacom growth in the quarter came from AI-related products, as customers require more copper, optics and power for larger, more complex AI clusters. The company's acquisition of CommScope contributed more than $1.2 billion in sales in the quarter, with CommScope's IT datacom business nearly doubling year over year; management expects that market to account for just under half of CommScope's 2026 sales, compared with roughly one-third in 2025. Amphenol faces tough competition from TE Connectivity, whose Digital Data Networks sales rose 34% year over year in the third quarter of fiscal 2026 with bookings up more than 70% year to date, and from Marvell Technology, whose data-center revenues increased 46% year over year to $2.17 billion in the second quarter of fiscal 2027. Amphenol shares have risen 16% year to date, trailing the broader Zacks Computer and Technology sector's 17% gain, and trade at a forward 12-month price/earnings of 25.32X versus the sector's 20.38X, while the Zacks Consensus Estimate for earnings stands at 72 cents per share, up 5 cents over the past 30 days and implying 53.19% growth.

Hewlett Packard Enterprise is sitting on a record pile of AI networking orders it cannot yet ship, and the gap between bookings and revenue is now the central question for the stock. Orders for the Networks for AI portfolio reached $700 million in fiscal Q3 2026, a new high, after the company blew past its cumulative fiscal 2026 target of at least $2 billion and raised that target to $2.5 billion to $3 billion. The hardware behind those orders includes routers and switches HPE is supplying to Oracle for one of the largest AI cloud infrastructure build-outs, work that extends more than a decade of engineering between Oracle and Juniper Networks, the business HPE acquired last year. Networking revenue rose 10% in fiscal Q3 2026 against a prior-year base that includes Juniper Networks, while networking orders rose 36%, and the segment brought in $2.9 billion of the $12.2 billion in total revenue HPE reported for the quarter. Management attributes the gap to supply, citing memory and wafer capacity that have gated supply since the start of fiscal 2026, and HPE has more than doubled its networking purchase commitments quarter over quarter while signing multiyear supply agreements to lock capacity. Networking revenue growth is guided to 11% to 13% in fiscal Q4 2026 and 14% to 17% in fiscal 2027, with the next scheduled look at the business coming at the Networking Investor Day on September 30, 2026.

Jim Cramer told Mad Money viewers to wait for Corning Incorporated to fall further before trimming or holding, citing the company's at-the-market selling program. Corning disclosed on September 11 that it had entered an agreement with Goldman Sachs to sell up to $2 billion of common stock through an at-the-market offering, with proceeds intended for general corporate purposes. The prospectus assumes raising the full $2 billion at $165.96 per share would require issuing 12,051,097 shares, increasing shares outstanding from 861,388,331 to 873,439,428. The offering came as Corning's optical business remained strong: at a September 9 Citi conference, CFO Ed Schlesinger said optical demand was very strong and orders continue to pick up, and second-quarter Optical Communications sales rose 32% year over year to $2.07 billion while core sales increased 17% to $4.74 billion. The bear case centers on execution, as Corning targets a $10 billion scale-up photonics business by the end of the decade tied largely to adoption of NPO and CPO technologies, and expects full-year 2026 capital expenditures of approximately $2 billion compared with $754 million spent in the first half of the year.

Meta Platforms raised its full-year 2026 capital expenditure guidance to a range of $130 billion to $145 billion, including principal payments on finance leases, narrowed from a prior $125 billion to $145 billion range in its Q2 2026 report on July 29, 2026. The forward guidance nearly doubles Meta's full-year 2025 capex of $72.215 billion, and Q2 capital expenditures alone reached $31.1 billion, driven by servers, data centers, and network infrastructure. To fund the build, Meta ended Q2 with $90.3 billion in cash and marketable securities and $83.7 billion in debt, and announced a strategic venture with BlackRock to develop a one gigawatt data center in El Paso, Texas. CFO Susan Li said Meta is demand constrained today, and CEO Mark Zuckerberg said the company is receiving quite a number of offers at a meaningful premium over what we paid for the compute, framing direct compute sales as one leg of a portfolio that also includes APIs, business agents, productivity tools, and subscriptions. The strain is visible in the quarterly numbers: Q2 free cash flow was $784 million, down 91.31% year over year, and operating margin compressed to 31% from 43%, even as Q2 revenue reached $60.801 billion, up 27.96% year over year and above the $60.286 billion consensus, with advertising revenue of $59.4 billion, up 27%.

GlobalFoundries has signed a new multi-year agreement with Marvell Technology to increase production capacity for its silicon germanium technology. Under the deal, GlobalFoundries will expand SiGe production at its Burlington, Vermont facility to support Marvell's growing requirements, with the added capacity aimed at next-generation pluggable optical transceivers, Near-Packaged Optics and Co-packaged Optics. The collaboration is expected to serve rising demand for high-speed optical connectivity driven by artificial intelligence and cloud data center infrastructure. GlobalFoundries' existing solutions support 200G-per-lane optical connectivity, while its development roadmap targets higher-speed generations as bandwidth requirements rise. The company faces competition from United Microelectronics, which has begun mass production of photonic IC wafers, and Taiwan Semiconductor, which is strengthening its CoWoS packaging and silicon photonics capabilities.

Bank of America forecasts the global semiconductor market will nearly double to $3.2 trillion by 2030 from $1.7 trillion in 2026, citing continued strength in AI infrastructure, memory and data-center demand. Within that total, BofA expects memory sales to climb from $937 billion in 2026 to $1.8 trillion by 2030, core semiconductor sales to rise from $739 billion to $1.35 trillion, and server-related sales to increase from $359 billion to $848 billion. The bank also projects wafer-fabrication equipment spending will more than double to $359.8 billion by 2030 from $155.9 billion in 2026, with 2027 largely fully booked and contracted across compute, networking and memory vendors and supply expected to stay tight in 2028. The Philadelphia Semiconductor Index gained around 3.14% on Thursday after five negative sessions, though it remains up about 91.19% over the past year and 59.14% year to date. BofA flagged a potential slowdown in hyperscalers' AI spending as the biggest risk to its outlook, even as recent calls for a more measured pace of AI development from figures including Dario Amodei, Sam Altman and Elon Musk have weighed on the broader AI trade.

United Microelectronics Corporation shares have soared 250.1% over the past year, far outpacing the industry's 25.8% gain, as improving foundry demand and AI exposure drive optimism. The company's 22/28-nanometer nodes contributed 37% of second-quarter revenues, with 22nm sales hitting another record, and management expects third-quarter wafer shipments to rise by a high single-digit percentage sequentially with capacity utilization exceeding 90%. UMC expects AI-related revenues to approach $300 million in 2026 and exceed $1 billion within three years, and it recently began mass-production delivery of silicon photonics wafers from its 12-inch Singapore facility, with plans to open its own 12-inch silicon photonics platform to broader customers in 2027. Its collaboration with Intel on a 12nm process platform is expected to begin production in Arizona in 2027. The 2026 earnings estimate now stands at $1.23 per share, up 75.7% over the past 60 days, and the stock carries a Zacks Rank #1 (Strong Buy).

Nvidia CEO Jensen Huang said he expects the company to sell roughly twice as many chips next year as it does this year, pointing to continued demand for artificial intelligence technology across industries. Huang made the projection while attending an AI-focused meeting hosted by King Charles III in Scotland, where he said businesses and governments in many countries keep investing in AI because of its potential economic benefits. The unit-growth outlook comes after Nvidia projected about $673 billion in revenue for fiscal 2028, representing growth of roughly 70% from the prior year, though the company does not disclose its total chip sales volume, making the two figures difficult to compare directly. Nvidia's portfolio includes data-center graphics processors such as Blackwell and Rubin, along with CPUs, networking chips and processors used in robotics and vehicles, and Huang has previously said the company shipped six million Blackwell GPUs over four quarters. Huang also joined executives from Google DeepMind, OpenAI and Anthropic at the Scotland meeting, where discussions included AI safety and product testing.

Arista Networks raised its full-year revenue outlook, citing continued robust demand for AI data center networking infrastructure. The company, which trades on the NYSE under ANET, said the higher guidance reflects management's expectation of a bigger top-line number for the current financial year. Management highlighted Arista's role in AI data center buildouts, and the company's AI data center focus sits within a much broader communications infrastructure footprint spanning data centers, AI clusters, campuses, and routing setups across the Americas, EMEA, and Asia-Pacific. Investors are watching whether upcoming quarters keep revenue above the US$3b mark and show continued strength in AI related switching and routing orders, along with how much of that business comes from a small group of hyperscale customers, since concentration there could still swing results either way. The update leaves intact existing risks around customer concentration and tougher competition from Cisco and NVIDIA.

MACOM Technology Solutions Holdings introduced a new 3.2T optical front-end chipset for AI and cloud data center interconnects and showcased it, alongside a broad portfolio of high-speed optical and connectivity components, at ECOC 2026 in Málaga, Spain. The launch underlines MACOM's push into 448 Gbps-per-lane and PCIe 6.0/7.0 ecosystems, aiming to become a core supplier for next-generation AI infrastructure and advanced telecom networks. The ECOC announcement closely echoes MACOM's OFC 2026 demonstrations, where it highlighted 3.2 Tbps optical transmit solutions and PCIe 6.0/7.0 technologies. MACOM's narrative projects $2.1 billion revenue and $605.5 million earnings by 2029, requiring 24.4% yearly revenue growth and about a $428.7 million earnings increase from $176.8 million today, while the most optimistic analysts already expected about US$2.7 billion in revenue and US$871.2 million in earnings by 2029. The company's forecasts yield a $398.36 fair value, a 51% upside to its current price.

Marvell Technology announced new collaborations with GlobalFoundries, Microsoft and Utimaco focused on advanced connectivity and payment security. The GlobalFoundries partnership targets expanded production of advanced SiGe chips used in next generation optical links for AI data centers, giving Marvell more assured capacity for SiGe chips used in pluggable optics, NPO and CPO and tightening a manufacturing relationship that already spans more than a decade. Marvell is also working with Microsoft and Utimaco to provide a cloud native payment security platform delivered through Microsoft Azure, combining LiquidSecurity hardware with Utimaco software to shift part of its security hardware footprint into a cloud delivered, usage based service. The first checkpoint for that effort is the Azure Payment HSM v2 public preview, which begins on 17 September 2026 in Western U.S. and Western Europe. Investors can then watch for disclosed customer adoption metrics, expansion of availability regions, and Marvell's capacity plans for higher speed SiGe optical products beyond the current 200G per lane benchmark.

Fujikura has sharply raised its full-year forecast for the fiscal year ending March 2027, more than doubling its operating profit projection from 211 billion yen to 432 billion yen. Net sales are expected to rise 48.4% to 1.755 trillion yen, ordinary profit to increase 127.1% to 453 billion yen, and net profit to grow 107.4% to 326 billion yen. The upgrade follows first-quarter operating profit for the fiscal year ending March 2027 of 104.8 billion yen, which exceeded half of the previous full year's 188.7 billion yen, with progress at the first-quarter mark standing at 24.3% for operating profit, 22.9% for net sales, and 24.7% for net profit, all on track with the plan. For the fiscal year ended March 2026, consolidated net sales rose 20.7% year on year to 1.1823 trillion yen, operating profit climbed 39.2% to 188.7 billion yen, and net profit jumped 72.5% to 157.1 billion yen, with the information and communications business generating 653 billion yen in net sales and 152.7 billion yen in operating profit on growing data center demand, accounting for more than 80% of consolidated operating profit. Forecast net profit per share is 196.88 yen, and the planned dividend is 38 yen per share.

MACOM Technology Solutions announced a front-end chipset solution for 3.2T applications designed to accelerate next-generation optical interconnects for artificial intelligence infrastructure. According to a company press release, the chipset delivers up to 3.2 terabits per second of throughput across eight channels operating at 448 gigabits per second PAM-4, a signal modulation technology used to increase data transmission speeds. The solution is engineered to support high-performance data processing across cloud data centers, high-speed networking, and next-generation AI infrastructure. Shares of the network chips maker jumped 4.4% in the afternoon session on the news, trading at $264.04, up 4.9% from the previous close. MACOM is up 50.9% since the beginning of the year, but at $264.04 per share it remains 35.5% below its 52-week high of $409.68 from May 2026.

Qualcomm has given Amazon an equity-linked incentive to deepen their AI-infrastructure partnership, with Amazon able to eventually purchase as much as $60 billion of Qualcomm data-center products while purchase-linked warrants could hand Amazon roughly $4 billion of Qualcomm shares at $161.26 each. Qualcomm shares gained approximately 2.1% to $188.64 Thursday, a level 7.37% above the stock's $175.69 GF Value. The warrant value represents roughly 6.7% of the maximum purchasing framework, directly tying Amazon's buying activity to Qualcomm's equity story. The two companies are also working together on custom AI inference silicon and optical connectivity capable of reaching 1.6 terabits per second. No minimum purchase commitment, delivery timetable or margin profile has been disclosed.

Generac struck a deal with Amazon to supply backup power generators for its data centers, with initial deliveries expected to total $2.4 billion between 2027 and 2028, and also granted Amazon the right to buy up to $340 million worth of its stock. Generac shares soared 19% on the news, while Amazon's stock moved 1.3% higher. Fluence Energy shares tumbled 14% after the battery storage maker cut its full-year guidance, now expecting $2.4 billion in revenue for 2026 versus its prior guidance of $2.9 billion to $3.1 billion, and anticipating a loss of $200 million before interest, taxes, depreciation and amortization, more than its previous guidance range of $30 million loss to $10 million EBITDA. Workday rose 5% after CNBC's David Faber reported that efforts to gain financing for a bid to take the cloud-based human resources software platform private are continuing, citing unidentified sources. Tower Semiconductor rose almost 9% after agreeing with privately-held New Photonics to a high-volume shipment of laser-integrated optical engines designed to meet growing demand for high-bandwidth, energy-efficient optical interconnects in AI infrastructure, and Vital Farms climbed 9% after Axios said the Austin, Texas-based pasture-raised egg and butter producer is exploring strategic alternatives, including privatization.

CoreWeave deployed multi-rack NVIDIA Vera Rubin NVL72 clusters on CoreWeave Cloud on Wednesday, moving the architecture from validation toward broader availability. CoreWeave had already become the first cloud provider to bring up and validate NVIDIA's Vera Rubin NVL72, and in August it won a multi-year agreement with Hudson River Trading to build its next-generation AI research and model-development platform using Vera Rubin NVL72 and Spectrum-X Ethernet. NVIDIA says Rubin is in full production and is designed for training and inference. CoreWeave said AI products and services beyond GPUs, including storage, CPU, networking and software, exceeded $400 million of annual recurring revenue in the second quarter, while booked annual recurring revenue for managed inference rose from $1 million to more than $100 million within months of launch. Active power exceeded 1.5 gigawatts in the second quarter and contracted power reached about 4.2 gigawatts by Aug. 11, with CoreWeave targeting more than 1.85 gigawatts of active power by year-end 2026 and at least 8 gigawatts by 2030, against expected 2026 capital expenditures of $35-$39 billion and second-quarter net interest expense of $640 million.

Astera Labs unveiled new Leo Smart Memory Controller solutions, including the Leo X-Series and Leo 2 E & P-Series, targeting AI and cloud workloads. The expanded Leo lineup is built for large-scale AI uses such as KV-cache and agentic AI, aiming to improve memory efficiency for hyperscalers. Astera Labs is collaborating with major CPU, GPU, and memory vendors and is showcasing the Leo family at the AI Infra Summit. The company designs connectivity chips that link CPUs, GPUs, and memory inside cloud and AI data centers, and its Leo controllers plug directly into the core plumbing that lets large AI workloads access high volumes of data efficiently.

AFL, Corning Incorporated, Sumitomo Electric Industries, Ltd., and TeraHop PTE. LTD. announced the completion of the SDM4 Multicore Fiber Multi-Source Agreement and the public release of the SDM4 Multicore Fiber MSA Technical Specifications, Version 1.0. The specification defines common geometrical, optical, mechanical and measurement requirements for a four-core multicore fiber intended for AI data centers, campus networks and other short-reach optical links, completing the objective the MSA set out to achieve: an industry-defined technical foundation for multi-vendor interoperability and future international standardization of four-core multicore fiber. Key elements include a four-core, square-lattice 2x2 core arrangement with a core pitch of 40 plus or minus 1 micrometers, a nominal cladding diameter of 125 micrometers, optical characteristics suitable for O-band short-reach applications, and crosstalk limits of minus 40 dB or less at 1 km and 1310 nm for both adjacent and diagonal core pairs. Antonio Castano, Chair of the SDM4 MCF MSA, said the specification establishes a common technical foundation that can help accelerate adoption of multicore fiber across the industry by aligning around a shared four-core fiber design. The MSA said frequent revisions are not anticipated and that future evolution of requirements for this class of fiber is expected to be advanced through relevant international standards-development organizations, with the published specification serving as a technical reference for discussions within ITU-T, IEC and organizations such as IEEE.

Corning Incorporated is expanding its partner base across the technology, cloud, networking and telecom sectors, signing a series of multibillion-dollar agreements tied to AI infrastructure buildouts. In the first half of 2026, Corning entered a multiyear agreement worth up to $6 billion with Meta to supply optical fiber, cable and connectivity products for Meta's U.S. data-center expansion, and also signed a multibillion-dollar, multiyear agreement with Amazon to support its AI data center expansion. Corning has additionally formed a multiyear commercial and technology partnership with NVIDIA under which it plans to increase its U.S. optical-connectivity manufacturing capacity tenfold and expand U.S. fiber production by more than 50 percent. Verizon and Corning announced a multibillion-dollar, multiyear agreement covering more than 80 million miles of high-density optical fiber and connectivity solutions from 2027 through 2032, supporting both broadband expansion and the long-haul network needed to connect AI infrastructure. Beyond those large contracts, Corning is collaborating with US Conec as a licensee of the PRIZM TMT optical ferrule technology, designed to accommodate higher fiber counts in tighter spaces for denser AI data centers. Corning faces competition from Amphenol Corporation and Ciena Corporation, and its growth outlook remains dependent on the pace of AI infrastructure investment and the timing of large customer deployments.

Marvell Technology's data center business reached a record $2.17 billion in the second quarter of fiscal 2027, up 46% year over year and 18% sequentially, and accounted for 79% of total company revenues. The data center end market made up 74% of Marvell's fiscal 2026 revenues, and the company expects data center revenues to grow approximately 60% in fiscal 2027. Marvell reported strong demand for 800G optical DSPs while its 1.6T products ramp rapidly, and scale-out switching is expected to more than double in fiscal 2027 on broader adoption of its 51.2T switch products. The company expects its custom business to accelerate in the second half of fiscal 2027 and more than double year over year in fiscal 2028, driven by XPU and XPU-attach solutions, with an expanded hyperscaler agreement covering AI inference accelerators, storage controllers, network interface controllers, memory interface controllers and near-memory compute. Marvell also expanded its position in next-generation AI infrastructure through its relationship with NVIDIA, including NVLink Fusion, scale-up switching and silicon photonics, and introduced products such as the 102.4 Tbps Teralynx T100 switch. Marvell shares have gained 170.3% year to date, and the Zacks Consensus Estimate for fiscal 2027 earnings suggests year-over-year growth of 48%.

Ciena Corporation announced three-year financial targets at its Investor Forum at its R&D facility in Ottawa, Ontario, projecting a revenue compound annual growth rate of approximately 30% from 2026 through 2029, an adjusted gross margin of about 50%, an adjusted operating margin of 32-35% and a free cash flow margin of roughly 20%. Management said the targets reflect the opportunities Ciena sees ahead and its confidence in execution, supported by a differentiated portfolio, growing customer demand, expanded supply capacity and new addressable market opportunities. The long-term targets follow strong third-quarter fiscal 2026 results, when Ciena reported revenue of $1.67 billion, up 37% year over year, adjusted gross margin of 46.4%, a record adjusted operating margin of 22.5% and $116 million in free cash flow, while backlog increased by $800 million to $8.5 billion and management expects fiscal 2026 backlog to exceed $10 billion. Ciena has also secured long-term agreements for certain key components through 2029, and for fiscal 2027 management provided an early outlook for revenue growth of at least 30%, or $8.3-$8.4 billion, with adjusted gross margin of 45-46% and adjusted operating margin of 25-27%. Beginning with first-quarter fiscal 2027 results, Ciena will change its financial reporting structure to four segments: Optical Systems, Interconnects, Global Services and Routing and Other.

CoreWeave is expanding its capabilities for next-generation AI workloads with the deployment of multi-rack NVIDIA Vera Rubin NVL72 clusters on CoreWeave Cloud, alongside two new capabilities for its AI Object Storage platform: cross-region write acceleration and a lower-cost Archive tier. The multi-rack Vera Rubin NVL72 deployment allows hundreds of NVIDIA Rubin GPUs to operate as a single scale-out cluster, with each NVL72 rack combining 72 Rubin GPUs with 36 Vera CPUs, NVIDIA NVLink 6, ConnectX-9 SuperNICs and BlueField-4 DPUs, connected across racks using NVIDIA Spectrum-X Ethernet networking and automated through CoreWeave's Mission Control platform. CoreWeave says its Local Object Transport Accelerator caches data locally and can reduce latency by up to 8x compared with traditional storage-cluster reads while delivering up to 7 GB/s of throughput per GPU, while the new cross-region write capability lets customers write data locally as CoreWeave replicates it to another region in the background. The company faces intensifying competition in the AI cloud market from Microsoft and Nebius Group N.V., which plans to invest £1.7 billion in U.K. AI infrastructure and expects connected power capacity to reach roughly 800 MW to 1 GW by 2026 end. CoreWeave shares have gained 24.4% year to date against the Internet Software industry's fall of 0.6%, and the Zacks Consensus Estimate for CRWV's earnings for the current year has been revised downward over the past 60 days.

Yuanta Securities (Thailand) said in an analysis note after visiting the factory of Stars Microelectronics (Thailand), or SMT, that the recovery trend in its business has become clearer, with second-quarter 2026 results reflecting a rebound in revenue, capacity utilisation and margins, and it expects the third and fourth quarters of 2026 to improve significantly on the prior quarter. SMT currently has advance orders covering roughly 8 to 12 months of production and targets 2026 revenue of no less than 2.6 billion baht, while for 2027 it expects growth of more than 10%, or above 3 billion baht. Yuanta views current estimates as based on fairly conservative assumptions and sees upside if orders convert to revenue faster than expected, particularly in wafer dicing, where market supply is constrained. Yuanta sees the Optical business becoming SMT's new growth engine on the back of AI growth, since the Optical and OSAT segments benefit directly from AI infrastructure investment and carry higher margins than the traditional business. Revenue from SMT's AI-related customers is still at an early stage and accounts for less than 5% of total revenue, but has the chance to rise to at least 10% next year. Meanwhile, its collaboration with Lumentum, a major player in the global Optical supply chain, marks an important starting point covering joint investment and development work. Yuanta expects SMT revenue of 2.67 billion baht in 2026, rising to 3.11 billion baht in 2027, and forecasts normalised profit of 150 million baht in 2026, rising to 256 million baht in 2027, or growth of about 70% from the previous year. On 2027 valuation, the stock trades at a PER of about 19 times, still below peers trading at roughly 28 to 58 times. Yuanta maintains its Buy rating on SMT with a target price of 7.60 baht, based on a PER of 25 times, compared with the closing price on 15 September 2026 of 5.85 baht, implying upside of about 29.9%.

AXT Inc. shares climbed 11.49 percent on Wednesday to close at $64.30 after Needham & Company named the company one of five semiconductor beneficiaries in the accelerating co-packaged optics sector. Following a five-day visit to Semicon Taiwan 2026, Needham said co-packaged optics will play a huge role in debottlenecking artificial intelligence networks through 2028 and beyond, and that AXT's indium phosphide substrate sits at the center of that shift. Needham also named Coherent Corp. and Lumentum Holdings among the similar beneficiaries, signaling a more bullish view across the industry. The firm upgraded AXT to buy from hold last July with a $90 price target, which still implies 56 percent upside from Wednesday's close. Investors will watch for business updates next week when AXT participates in the Morgan Stanley ASIA Best Corporate Day on Monday and Tuesday, September 21 to 22, including any guidance on third-quarter financial performance and updated operational and financial guidance for the rest of the year. In the second quarter, AXT swung to a net income of $13.03 million from a $7.67 million net loss a year earlier, as total revenues more than doubled to $47.59 million from $17.97 million on record indium phosphide revenues. Hedge fund holders fell to 33 in the second quarter from 37 in the first, but their committed capital rose 34 percent to $487.9 million from $363.7 million quarter-on-quarter.

Hewlett Packard Enterprise expanded its global AI infrastructure partnership with Oracle, bringing HPE Juniper Networking into Oracle AI data centers, and separately agreed to a large-scale digital modernization project with Tottenham Hotspur Football Club focused on stadium hybrid cloud and AI infrastructure. Both collaborations highlight new client deployments for HPE across AI data centers and digitally enabled sports venues. The Oracle expansion supports the view that Juniper networking widens HPE's role in AI-centric data centers, where Cisco and Arista are key rivals, and shows Juniper integration moving into multi-year deployments. The Tottenham project reinforces HPE's GreenLake and as-a-service infrastructure business by turning a complex, multi-site venue into a single hybrid cloud environment. The article does not answer how far HPE can shift its overall revenue mix away from legacy servers toward these recurring, software-rich contracts.

Apple is considering building enterprise servers powered by its M-series Ultra processors, according to The Information, potentially pairing those chips with networking equipment from Nvidia. The servers would reportedly be sold to businesses rather than used only inside Apple's own data centers, putting two very different chip strategies inside the same machine. Apple has spent years designing processors to reduce its dependence on outside suppliers, while Nvidia has become the backbone of much of the AI data-center market. Such a move would push Apple silicon beyond Macs, iPhones and its own computing ecosystem into the larger enterprise infrastructure market, where it would compete more directly with Dell, Hewlett Packard Enterprise and Supermicro. The project is still under consideration, so there is no guarantee Apple moves forward.

Alphabet reported Q2 FY2026 Google Search & other revenue up 17% to $63.27B and Google Cloud revenue up 82% to $24.77B, with YouTube ads climbing 13% to $11.06B. The company posted its 12th straight quarter of double-digit revenue growth and its 11th consecutive EPS beat, with Q2 EPS of $9.11 against a $3.0427 consensus, while FY2025 revenue crossed $400B for the first time on $73.27B of free cash flow. Alphabet guided 2026 CapEx to $175B-$185B, which pushed Q2 free cash flow to negative $5.86B, and long-term debt roughly doubled from $46.5B to $98.2B, with the buyback suspended in Q2 2026. The cloud backlog nearly doubled to over $460 billion in Q1, nearly 90% of the Fortune 100 now use Gemini Enterprise, and the Gemini App has 950M monthly active users. Alphabet raised its dividend 5% to $0.22 per share, and Waymo just passed 500,000 fully autonomous rides per week.

Fabrinet shares have dropped about 22.3% since its last earnings report, underperforming the S&P 500. The company reported fourth-quarter fiscal 2026 non-GAAP earnings of $4.10 per share, up 54.7% year over year and beating the Zacks Consensus Estimate by 6.49%, while revenues rose 44.6% to $1.316 billion, beating consensus by 2.64%. Data center revenues totaled $669 million, up 68% year over year and 13% sequentially, becoming the largest category at 51% of sales, with data center interconnect exiting the quarter at an annualized revenue run rate above $1 billion. For the first quarter of fiscal 2027, Fabrinet expects revenues of $1.375-$1.425 billion, implying 43% year-over-year growth at the midpoint, and non-GAAP earnings of $4.10-$4.25 per share. Management estimates Building 10 at the Chonburi campus, on track for completion by early 2027, can add $3-$3.5 billion of revenue capacity, taking total capacity to roughly $8.5-$9.3 billion, with a path to $12.5-$14 billion of annual revenue capacity over the coming years.

GlobalFoundries is broadening its growth prospects by expanding collaborations across emerging semiconductor applications, including AI, optical networking, power management, automotive, edge computing and quantum technologies. The company recently signed a long-term manufacturing agreement with Monolithic Power Systems, under which Monolithic will deploy its proprietary process technology at GlobalFoundries' advanced 300mm facility in Singapore to address automotive architectures, industrial robotics and automation, and smart power stages for AI and cloud infrastructure. In the second quarter, GlobalFoundries' communications infrastructure and data-center end market generated more than 60% year-over-year growth, driven by demand for optical networking across its silicon photonics and silicon germanium platforms. In September, the company finalized a definitive agreement with the U.S. Department of Commerce's CHIPS Research and Development Office for up to $375 million in R&D funding over five years. GlobalFoundries also announced a strategic collaboration with RAAAM Memory Technologies, whose Gain-Cell RAM memory can take up 40% less space on a chip and use up to 60% less power than traditional SRAM, with GlobalFoundries providing the FDX platform manufacturing technology. The financial impact of these collaborations depends on successful commercialization, customer adoption and production ramps.

Worldwide data center capital expenditures accelerated sharply in 2Q 2026, growing 92 percent, according to a newly published report from Dell'Oro Group. Continued AI infrastructure investment supported growth across compute, storage, networking, and physical infrastructure, while rising memory and storage prices significantly increased server average selling prices. Baron Fung, Vice President of Research at Dell'Oro Group, said spending remained concentrated in NVIDIA Blackwell Ultra and hyperscaler custom accelerators, while agentic AI created incremental demand for general-purpose compute, storage, and complementary networking. Neocloud providers and AI model builders posted the fastest capex growth among customer segments, reflecting the early stages of their infrastructure buildouts, and Dell led server OEM revenue, followed by Supermicro and Lenovo, while white-box server revenue reached a record high. Fung added that ongoing accelerator deployments and emerging agentic AI and AI-related storage workloads should sustain strong capex growth through the remainder of 2026 and beyond, although supply constraints could limit the pace at which planned infrastructure is deployed.

Broadcom CEO Hock Tan pushed back on fears of a broader AI slowdown and defended the company's long-term revenue targets. The semiconductor and infrastructure software company has growing exposure to specialized AI chips and networking, selling networking technology that connects large clusters of AI accelerators and emerging as a key participant in custom chips built for hyperscale customers. That position gives Broadcom exposure to AI spending without full reliance on the market for general purpose GPUs, a distinction that may matter more as technology firms build more custom silicon for cheaper, more effective handling of massive AI workloads. Tan's optimism offers investors a direct rebuttal to recent worries that the enormous infrastructure cycle behind AI equities could be cooling. The next test will be Broadcom's results and outlook, which will show whether customer orders continue to support management's long-term optimism.

Broadcom, Micron Technology and Marvell Technology all raised forward guidance in their latest earnings reports, citing accelerating AI infrastructure demand. Broadcom reported fiscal Q3 2026 revenue of $29.59 billion, up 85.5% year over year, with AI semiconductor revenue of $16.7 billion, up 221% year on year, and guided Q4 AI semiconductor revenue to $21.7 billion, up 236% year over year, outlining a trajectory reaching approximately $115 billion in fiscal 2027 and $230 billion in fiscal 2028. Micron posted fiscal Q3 2026 revenue of $41.46 billion, up 345.7% year over year, and guided fiscal Q4 revenue to $50.0 billion plus or minus $1.0 billion at gross margin near 86%, with full-year fiscal 2026 capital spending of approximately $27 billion. Marvell reported Q2 FY2027 revenue of $2.739 billion, up 36.5% year over year, and raised its fiscal 2027 data-center growth outlook to approximately 60% from approximately 50% previously, with CEO Matt Murphy flagging the October 6, 2026 Investor Day as the next catalyst. Micron's fiscal Q4 report, expected late this month, is the next test of whether hyperscaler capital spending holds through year-end.

Arista Networks has been named a Zacks Rank #1 Strong Buy stock after posting a strong beat-and-raise quarter in early August, with its Q3 earnings estimate surging 19%, its FY26 EPS estimate up 11%, and its FY27 outlook 14% higher. The AI infrastructure company, which counts Microsoft and Meta among its largest clients, beat the Q2 earnings estimate by 15% on August 4 and reported its first $3 billion quarter, boosted by its Arista 2.0 platform strategy. Arista more than doubled its revenue between 2022 and 2025, from $4.38 billion to $9.00 billion, and more than doubled its GAAP earnings over the same period, from $1.07 a share to $2.75 per share. The company is projected to grow revenue another 39% in FY26 and 26% in 2027 to reach $15.80 billion, while adjusted EPS is expected to rise 36% in 2026 and 24% next year to $4.99 per share. Arista holds $13.3 billion in cash and equivalents against near-zero debt, and its average Zacks price target implies 30% upside from current levels.