The Backbone of AI: Optical Networking Deep Dive and Investment Opportunities In The Datacenter Gold Rush
We believe the optical networking industry is poised for significant growth due to a surge in AI technologies and applications, and believe there are several compelling investment and trading opportunities. Optical networking is a highly technical industry and can be a challenge for the average investor to comprehend, as well as follow all the different key technologies, terms and industry trends. I spent 3 years covering optical networking as a sell-side analyst. This deep dive is meant to provide a more simplistic overview of the optical landscape. We identify investment opportunities within the space, but will be initiating coverage on select names which we believe is best positioned to capitalize from these emerging trends.
Link to our research disclaimer.
Investment Thesis
We believe the optical networking component industry is poised for significant growth over the coming years due to a surge in global data consumption, which we anticipate will be led by robust adoption of AI technologies and applications. Optical networking systems power modern communication networks by enabling high-bandwidth, low-latency data transmission across data centers and telecom networks globally. Our bullish investment thesis is driven by the proliferation of hyperscale data centers, from companies like NVIDIA NVDA 0.00%↑, Meta META 0.00%↑, Google GOOG 0.00%↑, Microsoft MSFT 0.00%↑, Amazon AMZN 0.00%↑ and Oracle ORCL 0.00%↑ that require high-speed interconnects between data centers, and within data centers that only fiber optics can provide. We believe the rapid adoption of AI technologies, which require massive data processing, will grow demand for high-speed optical components and systems in hyperscale data centers from a current ~$4B market opportunity to ~$16B in 2028, or ~30% CAGR over that time frame. Over time, as enterprises adopt edge computing systems to process data closer to the source, the need for increased bandwidth will drive further demand for advanced optical components. This dual-stage growth trajectory positions optical companies to capitalize on both centralized and distributed computing expansions.
Consolidation over the last decade, including significant mergers like Lumentum’s acquisition of Oclaro, IPG Photonics, Neophontonics and Cloud Light, as well as Coherent’s integration of Finisar and II-VI, has streamlined the competitive landscape, making the optical space less cyclical than in the past. As a result, these companies benefit from stronger pricing power, diversified revenue streams, and exposure to secular growth trends such as 5G, AI and cloud computing. We believe the publicly traded optical component and system companies such as Lumentum LITE 0.00%↑, Coherent COHR 0.00%↑, Applied Optoelectronics AAOI 0.00%↑ and Fabrinet FN 0.00%↑ are worth exploring for investors seeking exposure to the growing demand for high-speed data transmission in this AI revolution. These companies are leaders in key optical components and systems such optical transceivers, amplifiers, coherent optics, and dynamic wavelength routing, serving both data center and telecom markets. As shown below, most of these stocks have seen incredible upside driven by the AI revolution that kicked off in November 2023. However, we believe we are in the 4th inning of this massive AI cycle, and believe select optical stocks have at least a ~1-2 year runway to see further upside.
(MUST READ) Timing Is Critical
Although consolidation over the last decade has made optical stocks less cyclical, it is still common for these names to go through high volatile boom bust cycles, so timing is essential when investing. Optical component stocks face significant bearish risks due to their heavy reliance on spending by Tier 1 telecom providers and hyperscalers, which are highly cyclical and vulnerable to macroeconomic pressures. Historically, during boom cycles optical companies experience aggressive capacity buildouts during growth phases, followed by often unanticipated sharp declines in demand, leading to margin compression and inventory overhang when the cycle ends. Furthermore, commoditization of key optical components further exacerbates pricing pressure over time, eroding profitability. This is why time to market is critical with the higher speed components. Additionally, geopolitical risks, including U.S.-China trade tensions and export restrictions threaten revenue streams, particularly for companies with significant exposure to Chinese customers or supply chains. Lastly, the optical component sector faces significant risks from evolving technology trends. The increasing adoption of Co-Packaged Optics (CPO) and integrated solutions, which we cover in our technology overview section, threatens the traditional pluggable transceiver market, potentially displacing established suppliers in favor of vertically integrated hyperscalers or specialized silicon photonics providers. This combination of cyclicality, commoditization, geopolitical and technological uncertainty makes the optical component sector inherently volatile and risky for investors. That said, higher risks often comes with high reward, where optical stocks see 2x - 10x upside during upcycles.
AI Data Center Networking Overview
Optical (or commonly referred to as photonics) fiber data transmission is the process of transmitting data through thin strands of glass or plastic (optical fibers) using light signals. These fibers carry information in the form of light pulses, which are generated by lasers or light-emitting diodes (LEDs). The data is encoded from electrical signals into these light pulses, which travel along the fiber to a receiver, where the light signals are converted back into electrical signals. In AI datacenter networking, flow of data is split into two main parts: the backend and the frontend.
Backend
The backend is like a supercomputer that processes massive amounts of data to power AI technologies. At its core are powerful servers, equipped with three main types of chips: CPUs (Central Processing Units), GPUs (Graphics Processing Units), and TPUs (Tensor Processing Units). CPUs act as the general managers, handling a wide range of tasks and managing system operations, while GPUs are the workhorses for crunching large amounts of data quickly, particularly for complex tasks like training AI models. TPUs, on the other hand, are specialized chips designed specifically for AI workloads, focusing on accelerating tasks like deep learning and neural network operations. This setup creates “AI clusters”, which we show below, that perform the intensive processing required for AI training, where models learn from data. As well as inference, which refers to trained models making predictions or decisions. Optical components play a critical role in the architecture of an AI cluster by enabling intra data center high-speed communication between servers, storage devices, and specialized processors (CPUs, GPUs, and TPUs).
Specifically, the large driver of optical components in the backend has been growing demand for GPUs, and with NVIDIA’s newly introduced Blackwell GPU chips we expect demand to accelerate. This is a result of NVIDIA's transition from Hopper to Blackwell GPUs, which doubles optical output bandwidth to 800G from Hopper's 400G. This in turn is driving optical transceivers' bandwidth from 800G to 1.6T. Taking it one step further, a single rack of NVIDIA’s GB200 platform now houses 72 Blackwell GPUs (NVL72), up from 8 Hopper GPUs, requiring massive interconnects within and between racks. For reference, NVIDIA’s NVSwitch help GPUs communicate in the same rack, while Quantum X800 Infiniband switch drives communication between racks. Each rectangular box below is is concsidered one rack. The transceiver-to-GPU ratio varies with cluster size, starting at 1.5:1 for a single NVL72 rack with 36 GPUs and increasing to 2.5:1 for mid-sized clusters (~3,000–10,000 GPUs). For larger clusters that exceed 10,000 GPUs, the ratio rises to 3.5:1 as interconnect complexity scales. Hyperscalers have indicated their need to buy 100s-of-thousands of GPUs, which we believe will drive demand for “millions” of optical transceivers (we dive into what a transceiver is below).
AI Datacenter Cluster
Frontend
On the other hand, the frontend connects these AI clusters to the rest of the network using Ethernet, enabling the transfer of data between the AI systems and traditional IT infrastructure for enterprises. The backend typically requires much more bandwidth than the frontend, with backend optics being the larger market because of higher demand and prices. For example, NVIDIA’s Hopper platforms, the backend uses 800G while the frontend uses 400G. However, connecting AI clusters to the frontend has significant implications for optical demand in enterprise systems. As AI clusters grow more powerful, the frontend connections must handle increasingly large data volumes between the AI infrastructure and traditional enterprise systems, such as storage, databases, and applications. Key enterprise networking equipment providers, such as Cisco CSCO 0.00%↑, Arista Networks ANET 0.00%↑, Juniper Networks JNPR 0.00%↑, and Huawei, are major purchasers of optical components to build the high-performance switches and routers needed for these networks. As the data center upgrade cycle slows, we believe optical vendors with high exposure to enterprise networks system OEMs will likely extend their growth runway.
Key Data Center Optical Component and Transport System Breakdown
Optical networking and component companies sell a range of products that serve different roles in transmitting, amplifying, routing, and processing light signals in data centers and telco networks. Within the data center, key optical components are divided into intra-data center and data center interconnect (DCI) optics.
Inside a single data center, intra-data center optics handles the connections between servers, switches, and other devices, typically over shorter distances of less than 2 kilometers. For these shorter links, transceivers—devices that send and receive data over fiber—are critical. They make sure the network is fast, efficient, and reliable, often using simpler, more cost-effective technologies. These systems prioritize density, making it possible to connect thousands of devices in a compact space. Data center interconnect (DCI) is the technology that manages optical transport systems between data centers, enabling them to share data over distances ranging from a few kilometers to thousands of kilometers. An emerging component in DCI infrastructures is coherent optical modules (ZR/ZR+), which are advanced transceivers that send large amounts of data quickly and reliably over long distances. These modules, combined with other technologies like wavelength division multiplexing (WDM), reconfigurable optical add-drop Multiplexers (ROADMs) and amplifiers help manage and maximize the capacity of data transmission over longer distances. While DCI systems can handle incredibly fast speeds they require specialized equipment to keep the signals strong and clear, which makes them more complex and expensive.
Below is an explanation of these key optical components, their primary function and leading providers.
Optical Transceivers (commonly referred to as optical modules)
Function: Optical transceivers are compact modules essential to optical communication, by converting electrical signals into optical signals for transmission and back to electrical signals at the receiving end across networks. These modules house a variety of optical and electronic components such as a transmitter (laser diode, driver), receiver (photodiode and TIAs) and digital signal processors (DSPs). They are designed to plug directly into switches, servers, and routers, enabling modules to be hot-swapped, which provides flexibility for network upgrades or scaling without requiring system downtime. Bill of material (BOM) is the total cost to build and is critical in optical module development. The driver accounts for ~10%, TIA accounts for ~10%, the DSP represents ~30%, and the remaining components account for ~50%. Optical transceivers come in various sizes and standards to support different network needs, and are defined by their bandwidth capacity, with the highest currently available being 800G modules. Optical modules are categorized into single-mode and multimode modules. Single-mode modules often use EML (Electro-absorption Modulated Lasers), which are better suited for longer-distance transmission but are more complex and expensive. Multimode modules utilize VCSEL (Vertical-Cavity Surface-Emitting Lasers), for shorter distances because of their affordability and are commonly used in AI data center networks.
Applications: Data centers and telecom networks, connecting high-speed equipment over fiber.
Leading vendors: China Innolight, Coherent, Eoptolink, Lumentum, Applied Optoelectronics
Key components in a transceiver:
Laser Diode: The core component for generating optical signals within a transmitter, which includes Electro-absorption Modulated Lasers (EML) used in long-range applications and Vertical Cavity Surface Emitting Lasers (VCSELs) used in short-range communication. Leading laser vendors include Lumentum, Broadcom AVGO 0.00%↑, Coherent, Applied Optoelectronics
Driver: Controls the laser operations within a transmitter. Leading vendors include Marvel MRVL 0.00%↑ and Macom MTSI 0.00%↑.
Photodiode: Receives optical signals and converts them back into electrical signals.
Transimpedance Amplifier (TIA): Amplifies the signal from the photodiode, and leading vendors include Marvel and Macom.
Digital Signal Processor (DSP): Circuitry that enhances data integrity with advanced modulation techniques, error correction and optimizes power consumption to improve the overall performance in high-speed transceivers. Leading vendors include Marvel, Broadcom, MaxLinear and Infineria INFN 0.00%↑.
Coherent Optics
Function: Coherent optics are advanced optical communication technologies that use sophisticated modulation techniques to transmit high-speed data over long distances with excellent signal quality. Among these, ZR and ZR+ transceivers are pluggable coherent modules specifically designed for DCI applications. Unlike traditional optical components discussed above, which are simpler and used for short-range intra-data center connections, ZR/ZR+ modules are more expensive due to the coherent optics and DSP integration but offer better performance for DCI.
Applications: Suitable for metro DCI and long-haul telco applications.
Leading Vendors: Lumentum, Ciena, Infinera, Cisco, and Coherent
Wavelength-Division Multiplexing (WDM) Systems
Function: WDM systems are on an optical network that enables multiple data streams, each carried on a different wavelength of light, to be transmitted over a single optical fiber. These systems rely on components like multiplexers (MUX) to combine wavelengths and demultiplexers (DEMUX) to separate them.
Applications: Primarily for telecom applications, enabling high-capacity data transmission across metro, long-haul, and subsea networks, as well DCI networks.
Leading Vendors: Lumentum Ciena, Huawei, Infinera, Nokia, and Cisco
Reconfigurable Optical Add-Drop Multiplexers (ROADMs)
Function: ROADMs are devices used in fiber-optic networks that provide dynamic wavelength management within WDM systems. They allow specific data streams, carried as different colors of light (wavelengths), to be added, removed, or passed through without needing to change them into electrical signals. This makes networks more flexible, as operators can adjust how data flows remotely without physical changes to the system.
Applications: Used in telecom networks for long-haul and metro optical transport systems, as well as a staple in modern data center interconnect environments.
Leading Vendors: Lumentum, Ciena, Cisco, Infinera, Fujitsu, and Huawei
Amplifiers
Function: An amplifier in optical networking is a device that sits between transceivers in an optical network that boosts the strength of optical signals, without converting into electrical signals, to overcome losses caused by long transmission distances. The most common type is the Erbium-Doped Fiber Amplifier (EDFA).
Applications: Amplifiers play a critical role in long-haul and metro telco optical networks, and extended DCI networks ensuring that signals maintain sufficient power across extended distances.
Leading Vendors: Lumentum, Coherent, Cisco, Fujitsu, Ciena, Huawei, Infinera
Optical Switches
Function: An optical switch is a device in fiber-optic networks that complement WDMs and ROADMs and directs light signals between different fiber paths without converting them into electrical signals. It’s used to reroute traffic or change network connections quickly, which is helpful during network upgrades or to handle failures. Optical switches work alongside ROADMs and amplifiers but serve a different purpose. While ROADMs manage individual data streams (wavelengths) on a fiber and amplifiers boost the signal strength over long distances, optical switches handle whole light paths, controlling how signals move between different fibers or network nodes.
Applications: Telecom networks rely more heavily on optical switching for large-scale, long-distance traffic, whereas data centers prioritize electronic and some optical solutions for low-latency and high-speed local connections.
Leading Vendors: Coherent, Cisco, Ciena, Infinera, Huawei, Fujitsu
400G, 800G, 1.6T Market Sizing
Data transmission is measured in bits per second and commonly referred to in gigabits (“G”), which denotes billions of bits per second. Optical transceivers are designed with multiple data transmission lanes, where increased bandwidth is achieved by either increasing the number of channels (lanes) or boosting the speed of data transfer per channel. This scalable architecture allows transceivers to meet growing data demands efficiently.
Optical transceivers operate over two primary types of physical fiber optic cables: single-mode fiber (SMF) and multimode fiber (MMF), each optimized for different applications. Single-mode fiber (SMF) has a narrow core (~8-9 µm), allowing light to travel in a straight, single path, minimizing dispersion and enabling long-distance transmissions (500m – 10km+). It leverages laser-based sources like Electro-Absorption Modulated Lasers (EMLs) or Distributed Feedback (DFB) lasers. In contrast, multimode fiber (MMF) has a wider core (~50 µm), which causes light to bounce around inside the fiber, taking multiple paths. This leads to higher signal dispersion, meaning the signal degrades more quickly, limiting MMF’s range to ≤100m. MMF uses lower-cost VCSEL (Vertical-Cavity Surface-Emitting Lasers), making it ideal for short-range data center interconnects. The type of fiber used also determines the connector type—single mode fiber transceivers typically use lucent connectors (LC) for single-fiber Wavelength Division Multiplexing (WDM) links or multi-fiber push-on (MPO) connectors for multi-lane parallel transmission, while MMF transceivers use MPO for high-density connections in data centers.
Transceivers are categorized using suffixes that define fiber type, number of lanes, modulation scheme, and transmission distance, helping standardize connectivity across different networking applications. Short-range multimode transceivers (e.g., SR, SR4, SR8) use parallel fiber lanes, optimized for high-speed connections within data centers, typically reaching distances of ≤100m. In contrast, single-mode transceivers (e.g., DR, FR, LR, ER) support longer distances by leveraging either parallel fiber lanes (MPO) or Wavelength Division Multiplexing (WDM) over a single fiber pair (LC).
A key innovation in data center photonics is the adoption of 200G lane speed optical components, which double the data transfer rates compared to traditional 100G lanes. These higher-speed lanes are driven by advancement in laser technology and are critical to enabling the transition to 1.6T and beyond, as they reduce the number of required lanes while increasing overall bandwidth efficiency. For instance, a 1.6T transceiver leveraging 200G lanes can achieve full-duplex operation with 16 lanes—8 for transmitting and 8 for receiving—rather than requiring 32 lanes of 100G. This advancement not only supports the growing demand for ultra-high-speed networking but also optimizes energy efficiency and minimizes the physical footprint of networking hardware, making next-generation data center infrastructure more scalable and cost-effective. Emerging technologies such as coherent optics for long-haul and metro networks and next-generation silicon photonics are expected to play a crucial role in enabling future 3.2T and 6.4T optical networking solutions.
While 100G (100 billion bits per second) is already widely deployed in both telecom and data center networks as a standard speed, 200G/400G has emerged as the new standard in hyperscale data centers. However, 800G shipments are accelerating in data centers with the quick rise of high data density AI applications and cloud computing, with 1.6T (terabits per second or 1 trillion) planned for production in 2025. Module pricing varies from single mode to multi mode, with 800G multi-mode ASP ~$450, 800G single mode ASP ~$700 and initial 1.6T modules estimated to start around ~$2k per module. However, keep in mind as competition and production yields increase ASPs will come down over time.
As shown below, the market for cloud data center photonics (optics) is projected to grow significantly, driven by the increasing adoption of high-speed optical transceivers and advanced switching solutions for AI and data center applications. The total addressable market is expected to expand from $4.5B in 2023 to $16B by 2028, with a compound annual growth rate (CAGR) of ~30%. Key growth areas include 200G/400G/800G/1.6T intra-data center transceivers, 400G/800G/1.6T ZR/ZR+ modules for data center interconnect (DCI), and advanced transport solutions like mux/demux and amplification systems. The push toward 800G and 1.6T speeds is central to this growth, with ~80% of the market expected to operate at 800G+ by 2028. However, 100G/200G/400G will still have a market as hyperscalers update legacy equipment, as well as deploy solutions to the frontend.
Largest Optical Component Customers
Amazon, Meta, Microsoft, Google and Oracle, are all aggressively building AI data centers, and each has unique strategies reflecting their business priorities. This aggressive expansion is the main driver to the surge in demand for high bandwidth optical components and systems. We highlight these companies' unique strategies, as well as their key optical vendors.
Amazon, through AWS, focuses on scalable cloud-based infrastructure, integrating custom AI chips like Trainium and Inferentia to optimize performance and cost-efficiency for diverse customers. Key vendors include: China Innolight, Eptolink.
Meta prioritizes in-house solutions tailored to support its AI-driven social platforms and metaverse ambitions, including custom hardware and data centers designed to handle massive workloads for generative AI and AR/VR technologies. Key vendors include: Applied Optoelectronic (speculation).
Microsoft, leveraging its Azure platform, emphasizes partnerships, such as its deep collaboration with OpenAI, and invests in specialized AI supercomputers to deliver advanced generative AI capabilities to enterprise customers. Key vendors include: Applied Optoeletronics.
Google, with its Tensor Processing Units (TPUs), blends hardware and software innovation to power its AI research, cloud services, and consumer applications like search and YouTube, while building one of the most energy-efficient AI infrastructures globally. Key vendors include: Lumentum, Coherent, China Innolight
NVIDIA is also a key customer of optical components by being the leading vendor of cutting edge GPU systems, which include key optical networking components. All hyperscales are indirectly purchasing optical components through NVIDIA. However, as we highlight in our emerging optical trends section below, some hyperscalers are rumored to be taking an unbundling approach. This refers to hyperscalers purchasing GPUs from NVIDIA, but sourcing optical components directly from suppliers to lower cost and create customized data center solutions. Key vendors include: Coherent, China Innolight, Mellanox (internal manufacturing capabilities with Fabrinet), Ecoptolink.
Oracle is building a robust AI infrastructure by introducing one of the largest cloud-based AI supercomputers. The company is expanding its global data center network with plans to build 100 new centers, ensuring resources are available for advanced AI workloads.
OpenAI recently announced plans to build its own data centers in the U.S, which represents a significant new customer opportunity for optical component companies. These data centers will require high-speed optical interconnects to handle the massive data throughput and low-latency demands of AI workloads, driving substantial demand for advanced optical technologies.
As shown below, we believe tracking Capex spend for these hyperscalers can be an indicator of demand trends for optical components. Although they do not break down spending by technology, the elevated growth levels we continue to see likely bodes well for optical component companies.
Hyperscalers Quarterly Capex Spend ($Billion)
Source: Company Filings
Investor Gifts
Looking to become a better investor or find that perfect gift for your finance friend or family member? Check out these products from our partner Investor Gifts.
Emerging Optical Component Trends & Technologies
The optical component sector faces significant risks from evolving technology trends. The increasing adoption of unbundling, Co-Packaged Optics (CPO) and integrated Silicon Photonic solutions, which we cover below, threatens the traditional pluggable transceiver market, potentially displacing established suppliers.
Unbundling: Key CPU providers such as Intel and AMD, and GPU vendors led by NVIDIA, with AMD as a key competitor purchase optical components to build into their larger CPU/GPU platforms. However, an “unbundle” trend is emerging with hyperscalers. Rather than buying complete CPU/GPU systems with optics included, hyperscalers are leaning towards building their own systems where they can buy components separately for cheaper prices. While this should not impact the total shipments of optical components, it does create a greenfield opportunity for optical suppliers to possibly gain significant new customers. Google has already adopted this approach, and industry reports suggest Amazon is about to follow Googles suit, and we would not be surprised to see Microsoft and Meta to be next. We estimate this shift could open up ~$1B in new market opportunity for the optical component players.
Hyperscalers Sourcing Domestically: Potential new tariffs on Chinese imports, which could be reintroduced under a Trump administration, stand to benefit domestic optical component suppliers like Lumentum, Coherent, and Applied Optoelectronics. Combined with hyperscalers unbundling their optical component procurement, this shift allows U.S.-based companies to capture a larger market share by offering cost-effective and customizable solutions. As hyperscalers prioritize efficiency and localized sourcing, domestic manufacturers are well-positioned to capitalize on these trends.
Co-packaged Optics (CPO) is a new technology that integrates optical transceivers directly with electronic chips of the switches (CPUs or application specific integrated circuits (ASICs)) within the same package, eliminating the need for separate pluggable optical modules. By shortening connection paths and improving integration, CPO reduces power consumption, lowers latency, and increases bandwidth, making it ideal for large-scale AI data centers. However, its adoption faces challenges, with a shift in manufacturing responsibilities from optical module makers to switch chip producers. CPO is still in the early development stages, but key beneficiaries include switch chip producers like Broadcom, Marvell, and NVIDIA, which are actively developing CPO solutions. The shift to CPO may disrupt the current optical module providers, but key discrete optical components are likely to remain essential in CPO systems.
Linear-drive Pluggable Optics (LPO) is a new approach in optical module technology that simplifies data transmission by removing the traditional digital signal processors (DSPs). Instead of converting analog signals to digital and back again, as DSPs do, LPO uses linear analog processing to directly transmit analog signals over optical links. This reduces power consumption, lowers costs of the module (DSP ~30% of BOM), and significantly decreases latency, making it especially appealing for large-scale AI applications. However, by relying on the DSPs within network switches to handle signal correction, LPO sacrifices some of the noise resistance and compatibility functions of traditional systems. It also faces challenges like shorter transmission distances, increased system complexity, and protocol compatibility issues. Overall, LPO represents a cost-efficient, low-power alternative for specialized use cases, but broader adoption depends on overcoming its technical limitations and achieving full compatibility across networks. Linear receive optics (LRO) has been discussed as a middle-ground solution between traditional optical modules and LPO. LRO continues to use the DSP on the transmitting side while eliminating the receiver DSP.
Silicon Photonics (SiPh) uses silicon to transmit data with light, integrating optical components on silicon wafers. This technology reduces power consumption, costs, and heat, while offering higher reliability and compatibility with emerging technologies like Linear-drive Pluggable Optics and Co-packaged Optics. SiPh is more scalable and cost-effective than traditional methods. For instance, its use of cost-efficient CW lasers can cut module costs significantly compared to EML-based solutions, and its tight integration improves performance and reliability. SiPh is gaining traction as demand for high-speed 800G and 1.6T modules grows. SiPh faces challenges such as improving production yields and delayed readiness for 1.6T modules, but its advantages in power efficiency and cost make it a strong contender for future adoption. Key players like China Innolight, Lumentum, and Coherent are driving SiPh’s progress, and it is expected to become a significant alternative to EML in mid- and short-distance applications, while EML retains its edge in long-distance use.
Investable Optical Component Companies
Below we identify investment opportunities within the optical space, and who we believe is best positioned to capitalize from these emerging trends.