What changed in this filing

/stocks — the workspaceLOADING

Item 1. Business

2024-02-13 compared with 2023-02-14 · 235 added, 212 removed, 135 unchanged (77% of the section changed)

Arista Networks is an industry leader in data-driven, client-to-cloud networking for next-generation data center and campus workspace environments.

Starting from the beginning, Arista set out to differentiate our architecture in two ways – uncompromising quality and operational efficiency built on an advanced networking operating system, Arista’s Extensible Operating System (“EOS”).

EOS, combined with a set of network applications and ethernet switching and routing platforms using merchant silicon, provides improved price/performance and time to market, delivering a cloud networking solution with high performance scale and availability, and enabling network automation, visibility, and security.

In recent years, we have brought the operational consistency and principles of cloud networking to the broader enterprise and campus markets with our Cognitive Cloud Networking approach, extending EOS across the enterprise data center and campus wired and wireless workspace.

CloudVision®, our workload orchestration and automation solution, leverages EOS based network wide data models to deliver turnkey automation and configuration capabilities across data center, wired and wireless workspaces, multi-cloud, and WAN routing use cases.

In 2020, we completed the acquisition of Big Switch Networks, a pioneer in network monitoring software and Awake Security, an Artificial Intelligence (AI)-driven network detection and response (NDR) company.

We believe the integration of these capabilities with our network-wide, data-based visibility and telemetry offerings delivers the next generation of operationally efficient network security and visibility solutions.

In 2022, we completed the acquisitions of Untangle Incorporated, a security asset for edge threat management for integration into our Cognitive Unified Edge (CUE) commercial and branch offering.

We also completed the acquisition of Pluribus Networks, a pioneer in unified cloud fabric networking, for integration into Arista’s Unified Cloud Fabric (UCF).

In addition, we continued to expand our 400G product portfolio for both enterprise and cloud customers with the introduction of the next generation 7050X/7060X and 7300X Series switches delivering performance and cost benefits for customers of all sizes as they transition to 400G.

We also continue to expand our cognitive campus edge portfolio with the new C-360 Wi-Fi 6E access point to take advantage of the greater capacity, higher bandwidth, and lower latency of Wi-Fi 6E.

Since we began shipping our products, we have grown rapidly, and, according to market research, we have achieved the second largest market share in data center Ethernet switch ports and revenue.

We have been profitable and cash flow positive for each year since 2010.

Arista Networks is an industry leader in data-driven, client to cloud networking for large data center, campus and routing environments.

Arista’s platforms deliver availability, agility, automation, analytics and security through an advanced network operating stack.

Since Arista’s inception, our founders have reimagined cloud networks for performance, scale and programmability with a focus on differentiating in three ways:

uncompromising quality, advanced open and standards-based technology and a robust quality assurance capability built on a suite of automated diagnostics.

At the core of Arista’s platform is Arista’s Extensible Operating System (EOS®), a modernized publish-subscribe state-sharing networking operating system.

Arista EOS, combined with a set of network applications and our Ethernet switching and routing platforms using best of breed merchant silicon, provides customers with a highly competitive and diversified portfolio of products with improved price/performance and time to market.

Our current portfolio of offerings are categorized in the following three product categories:

high-speed Data Center and Cloud Networking systems including newer artificial intelligence ("AI") Ethernet switching platforms.

Cognitive Adjacencies:

campus wired and wireless products and advanced routing systems addressing Core Routing, Edge Routing, Data Center Interconnect (DCI), Multi-cloud and Wide Area Networking (WAN) use cases.

Network Software and Services:

a suite of value-add software solutions that leverage Arista’s EOS to provide advanced end-to-end orchestration, automation, analytics, network monitoring and security.

Since we began shipping our products in 2008, we have experienced rapid growth, and, according to market research in 2023, we have achieved the leadership position in high-speed Ethernet port shipments of 100G and above and the second largest market share in overall data center Ethernet switch ports and revenue.

We have been profitable and cash flow positive since 2010.

Our Market Opportunity

We sell our products through both our direct sales force and channel partners, and compete primarily in the high-speed data center switching market for 10 Gigabit Ethernet and above, excluding blade switches.

We added advanced routing capabilities to our R-Series switches, which in addition to switching address the Data Center Interconnect (DCI) market, cloud-grade routing, next-generation network edge and wide area networking routing market.

We also compete in the enterprise campus market for 1 Gigabit Ethernet switching and above and in the cloud-managed wireless networking market.

I n addition, we participate in the Network Monitoring and Network Detection and Response security markets.

We believe that client-to-cloud networking represents a fundamental shift from traditional legacy network architectures.

As organizations of all sizes have moved workloads to the cloud, spending on cloud and next-generation data centers has increased rapidly, while traditional legacy IT spending has grown more slowly.

We sell our products through both a direct sales force and channel partners, competing primarily in the high-speed data center Ethernet switching markets for 10 Gigabit Ethernet ("GbE") and above, including the AI Ethernet switching market, the cloud-grade and enterprise routing markets, and the campus wired and wireless markets.

In recent years, we have also entered into the Network Monitoring and Network Detection and Response (NDR) security markets through both acquisition and organic development.

Our Customers

As of December 31, 2022, we had delivered our cloud networking solutions to over 9,000 end customers wo rldwide.

Our end customers span a range of industries and include large internet companies, service providers, financial services organizations, government agencies, media and entertainment companies, telecommunication service providers and other cloud service providers.

Purchases by both Microsoft and Meta Platforms accounted for more than 10% of our total revenue for the years ended December 31, 2022 and purchases by Microsoft accounted for more than 10% of our total revenue for the years ended December 31, 2021 and 2020 .

Cloud Networking and Market Drivers

Digital transformation is fundamentally changing the way IT infrastructure is built and how applications are delivered from a cloud environment.

In cloud computing, applications are distributed across thousands of servers.

These servers are interconnected by high-speed networking switches to form a pool of resources that allows applications to be rapidly deployed and cost-effectively updated.

Cloud computing enables ubiquitous and on-demand network access to these applications from internet-connected devices including personal computers, tablets, Internet of Things (IoT) devices, and smartphones.

Nearly all consumer applications today are delivered as cloud services, which was originated by internet leaders like Amazon, Google, Meta Platforms and Microsoft.

These internet leaders pioneered the development of large-scale cloud data centers in order to meet the growing demands of their users, including business customers.

Enterprise applications are rapidly moving to the cloud as cloud services are easier and more cost effective to deploy, scale and operate than traditional applications.

Enterprises and service providers around the world are adopting cloud computing technologies in order to achieve similar performance, operational efficiencies and cost reductions.

The aggregate network bandwidth in the cloud can be orders of magnitude higher than typical legacy data center networks.

Therefore, the networks in such cloud environments must be architected and built in a new way.

We refer to these next-generation data center networks as cloud networks.

Cloud networks must deliver high capacity, high availability and predictable performance, and must be programmable to allow integration with third-party applications for network management, automation, orchestration, and network services.

The expansion of AI computing and distributed applications is further pushing the envelope of predictable scale and performance in the network.

Our customers span a range of industries and geographies, including large cloud customers or hyperscalers, other internet providers, service providers, financial services organizations, government agencies and a cross section of enterprise customers.

Over the past five years, we have diversified the types of enterprise customers we sell to and have continued to expand our presence across a wide spectrum of industries including media and entertainment, healthcare, oil and gas, education, manufacturing, industrial, and more.

Meta Platforms and Microsoft, two of our cloud end customers, each accounted for more than 10% of our total revenue for the years ended December 31, 2023, and December 31, 2022.

Market Drivers

Digital Transformation

Digital transformation is fundamentally changing the way technology is integrated into business operations and as a result how IT infrastructure is built, and applications are delivered across cloud and end-customer environments.

The expanded dependency of business operations on the network has increased the complexity of the network and heightened the importance of network availability, predictable performance, open programmability and operational simplicity.

The public cloud leaders pioneered the development of large-scale cloud data centers to meet these growing demands from their users, including business customers.

Enterprises now have the option to move applications to the cloud as cloud services are generally easier and more cost effective to deploy, scale and operate than traditional applications.

These cloud metrics have become the baseline for performance, cost and efficiency of IT infrastructure investments.

Enterprises and service

providers around the world are also now adopting cloud computing technologies and principles to their own non-cloud or hybrid operations in order to achieve similar performance, operational efficiencies and cost reductions.

Arista addresses our customers' requirements through our approach to network architecture, our platforms and our software.

Our comprehensive R-series and X-series switching and routing portfolios running the highly programmable EOS, transform networks with simplified and scalable architectures across multiple use-cases.

Artificial Intelligence (AI)

The expansion of generative AI computing and distributed applications is further pushing the boundary of predictable scale and performance in the network.

A common characteristic of these AI workloads is that they are both data and compute intensive.

A typical AI workload involves large sparse matrix computations, distributed across hundreds or thousands of processors (CPU, GPU, TPU, etc.) with intense computations for a period of time and requires a high-bandwidth, scalable, lossless network in order to service these workloads.

Examples of key secular trends driving network transformation are illustrated below:

Arista's AI strategy is based on achieving two key objectives.

Arista first offers to customers the Arista Autonomous Virtual Assist ((AVA TM )) using natural processing language to provide AI assisted outcomes for security and observability.

Arista also provides network switching products intended to provide a robust interconnect that seamlessly links GPUs, compute and storage to deliver fast job completion time for training and generative AI workloads.

An overview of our AI enabled solutions is shown below:

As a proud founding member, Arista is committed to leading the Ultra Ethernet Consortium (UEC) to achieve scalable and efficient remote memory access, implemented with enhanced packet spraying, flexible ordering, and modern congestion control algorithms.

In the post-pandemic world, the traditional “campus” has been redefined and the boundaries between the office, home, teleworker and user have converged.

At the same time, the proliferation and sophistication of devices that connect the campus, such as smart devices, security cameras and Internet of Things ("IoT"), has grown dramatically.

The challenge lies in successfully transitioning the existing siloed campus into a data-driven, distributed campus model with a common experience, while addressing the growing security and availability needs.

Arista’s campus portfolio was driven by customers desiring the same quality and operational efficiency available from EOS and CloudVision® throughout their entire enterprise network.

We entered the campus market with a diverse portfolio of modular and fixed form factor Campus spine switches, Power-over-Ethernet (PoE) switches, and WiFi access points based on

EOS and managed through CloudVision.

We continue to expand our campus portfolio to offer the advantages of EOS across the entire enterprise network.

Most recently, we have added incremental Enterprise WAN products as well as embedded NDR security sensors into our campus switches to address more of the security challenges that face campus administrators.

Zero Trust Networking Security

Today, a zero trust networking approach to security is paramount for organizations looking to build a robust cybersecurity program.

Irrespective of which device, application, or user is accessing an enterprise resource, zero trust focuses on complete visibility and control over all activity on the network.

Arista’s zero trust networking principles, based on NIST 800-207, help customers address this challenge with three cornerstones:

visibility, continuous diagnostics, and enforcement.

The Arista NDR platform delivers continuous diagnostics for the entire enterprise threat landscape, processes countless points of data, senses abnormalities or threats, and reacts if/when warranted.

Limitations of Traditional Enterprise Data Center and Campus Networks

We believe that cloud networks and legacy networks are fundamentally different.

In a traditional data center, specific applications are installed on a small number of servers and most network traffic is server-to-client, or “north-south” traffic, which results in perhaps a few terabits/second of aggregate network bandwidth.

In the cloud, most network traffic is server-to-server, or “east-west” traffic.

The aggregate network bandwidth in the cloud can exceed 1 petabit/second, orders of magnitude higher than that of typical legacy data center networks.

Increasingly more complex and sophisticated use cases such as AI clusters are further pushing the boundaries of cloud networking bandwidth and capacity demands.

These large scale cloud networks require much higher network availability since network outages in the cloud are costly to customers.

Traditional network switches have evolved, and the features and capabilities of their operating system have expanded over many years without addressing the structural deficiencies of their underlying software architectures, making it difficult to achieve the required level of reliability.

Some networking vendors have built products that use proprietary protocols to address the scaling needs of next-generation data centers.

However, proprietary protocols are generally disfavored by cloud customers and cloud service providers because they create vendor lock-in and discourage vendor multi-sourcing.

Legacy enterprise networks are generally not programmable and, as a result, are extremely difficult to integrate with third-party applications for network management, automation, orchestration and network services.

This lack of integration forces customers to continue to rely on time consuming, manual box-by-box configuration and processes that can be error-prone and cost-prohibitive.

Traditional enterprise wired and wireless campus networks must cope with an ever-increasing number of endpoint IoT devices and remote work locations requiring users to be connected from anywhere.

The introduction of large scale, highly complex, public cloud environments and the digital transformation of end customer business models meant that the traditional ways of building networks were no longer adequate to meet the needs of customers for the deployment and provision of cloud applications and more recently generative AI applications, and new innovations were needed to push network performance forward.

Historically, most common network designs were rigidly hierarchical, based on a 3-tiered model developed in the early days of the internet for sparse north/south traffic patterns.

This model was limited in the number of devices that could be connected to a network and introduced many points of congestion as customers tried to scale the solution.

As more applications move to the cloud, network connections must scale, and the increased east/west traffic must be managed without congestion.

In addition, the switches and routers used to build these tiered networks were based on proprietary, application-specific integrated circuits ("ASICs") that historically underperformed when measured against Moore’s Law and operating systems that lacked the openness and programmability necessary to automate and effectively manage these networks.

As demand for scale and performance increased, the end-user experience was degraded, operational costs increased, and the host cost-per-connection increased.

Similarly, traditional enterprise networks have been mired in complexity, proprietary features and architectures, custom ASICs, siloed designs, and fragile software offerings built up over the previous three decades.

Because of this, operating a legacy network is riddled with challenges;

critical outages that cause risk-averse behaviors, labor-intensive rollouts that impede business initiatives, limited visibility that prevent problem detection and isolation, and overall lack of automation and uniformity that results in inefficiency.

In addition, there had historically been little or no attempt to address the needs of building and operating a network infrastructure at massive, cloud scale.

As a result, as enterprises moved applications to the cloud and implemented hybrid and multi-cloud strategies, there have been insufficient solutions that could configure, deploy, automate, and manage these scaled and dispersed resources.

In the post-pandemic world, enterprise campus wired and wireless networks must cope with an ever increasing number of endpoint IoT devices and remote work locations that require users to be connected from virtually anywhere.

Campus administrators have sought to address the resulting increased network complexities and bottlenecks through the adoption of a myriad of platforms, operating systems, proprietary features and network management tools.

Coupled with the explosive growth of IoT and the requirement for remote workloads, the operational costs of managing these complexities have become prohibitive.

Visibility is a critical component to a more efficient cloud-like network.

Being able to capture what a network is ‘thinking’ or ‘doing’ is the basis for true network automation and analytics.

Legacy networking has long suffered from limitations in network visibility largely due to inefficient polling mechanisms that only provide a limited subset of data.

As a result, the operators of legacy networks have been essentially blinded when it comes to true network insight.

Our Data-Driven Cloud Networking Solutions

The core of our platform is an advanced network operating system, Arista's EOS, which was designed from the ground up to be open, programmable, modular and reliable.

The programmability of EOS has allowed us to expand our software applications to address the ever-increasing demands of cloud networking, including data-driven workflow automation, network visibility, analytics and network detection and response, and has allowed rapid integration with a wide range of third-party applications for virtualization, management, automation, orchestration and network services.

Our cloud networking innovations started with pioneering a modern software platform, Arista EOS, which provides switching, routing, state-streaming and telemetry functions across all Arista platforms.

EOS established a new standard in networking for large-scale cloud operators, opened the door to the widespread adoption of merchant silicon hardware in

networks, and provided dramatic decreases in deployment and operating costs while delivering high reliability for cloud customers, service providers, enterprises, and more.

The Arista EOS network stack architecture provides a foundation for consolidation of streamed device state, telemetry, packet, flow, alert, sensor and third-party data into an aggregated Network Data Lake (Arista EOS NetDL™).

Arista EOS NetDL consolidates diverse datasets required for effectively applying AI/Machine Learning (ML) methods in Network Operations (NetOps) and Security Operations (SecOps) environments, and it presents a single application programming interface ("API") surface for access to network and network-related data for enhancing Arista and third-party applications.

An overview of our cloud networking solutions is shown below:

The key benefits of our cloud networking solutions are as follows:

Capacity, Performance and Scalability

Our data-driven cloud networking platform enables data center networks to scale to hundreds of thousands of physical servers and millions of virtual machines with the least number of switching tiers.

We achieve this by leveraging standard protocols, non-blocking switch architectures and EOS to meet the scale requirements of cloud computing, including AI workloads.

We have used active-active Layer 2 and Layer 3 network topologies to enable customers to build extremely large and resilient networks.

High Availability

Our highly modular EOS software architecture was designed to be fault-isolating and self-healing in order to deliver higher availability compared to legacy network operating systems.

In addition, customers can non-disruptively upgrade switches running in the network using Arista’s Smart System Upgrade ("SSU") application, without interrupting the network service.

Open and Programmable

EOS is built from the ground up using an open Linux microservices architecture, with publish and subscribe state sharing at its core, and open Application Programming Interfaces (APIs) for reliability and extensibility.

We offer multiple

types of extensibility for EOS that provide granular control and management, including our eAPI, OpenConfig, NetDL Streaming, CloudVision Studios and our EOS SDK.

These programmability options are imperative to enable automation and customization of our Cloud Networking platforms.

Workflow Automation

Our EOS software enables enterprises to provision networking resources in minutes with no manual intervention through our Zero Touch Provisioning (ZTP).

Capacity, Performance and Scale

Our data-driven cloud networking platforms enable data center networks to scale to hundreds of thousands of physical servers and millions of virtual machines with the least number of switching tiers.

We achieve this by leveraging standard protocols, non-blocking switch architectures and EOS to meet the scale requirements of cloud computing.

We architect active-active Layer 2 and Layer 3 network topologies to enable customers to build extremely large and resilient networks.

Emerging capabilities including recent developments related to AI will continue to place increased demands on networking infrastructure.

Arista’s strategic commitment to using merchant silicon is also a key competitive differentiator for Arista in addressing these capacity demands.

Merchant silicon not only provides the best price/performance available but allows Arista to bring next generation platforms to market early allowing customers to benefit from Moore’s Law.

Networks are only useful when they are available.

Arista’s modular EOS architecture and software testing innovations provide features and network designs that keep the network available even during maintenance and upgrades.

EOS publish-subscribe architecture provides self-healing resiliency through live patching, upgrades, fault isolation and containment and graceful process restart to reduce maintenance windows and allow for intelligent insertion and removal of network elements.

Openness and Programmability

Customers demand open, standards-based networking that avoids vendor lock-in and enables third-party integration to support best in-class technology ecosystems.

Arista EOS, built on Linux, features open standard protocols, such as Border Gateway Protocol ("BGP") and Ethernet VPN ("EVPN"), offering interoperable solutions.

Our well-structured set of APIs and EOS Software Development Kit ("SDK") as well as multiple DevOps integrations, enable enterprises to automate networking provisioning without manual intervention.

EOS also natively supports Ansible, CFEngine, Chef, Puppet, virtual network orchestration applications and third party management tools.

Our CloudVision is a network-wide approach for workload orchestration and automation that delivers a turnkey solution for enterprises looking to modernize their networks and move to cloud-class automation and operation without significant internal development.

In addition, we deliver continuous testing, version control and change management services through our Continuous Integration (CI) pipeline.

Network Visibility

Our EOS software provides a set of tools and applications that proactively monitor, detect and notify network managers when network issues arise, delivering real-time data to third-party network performance and security applications to provide detailed application visibility.

Our network visibility applications provide real-time insight into the status of the network.

Through the integration of DataANalyZer ("DANZ") features native to Arista switches with Big Switch’s monitoring software, we provide the DANZ Monitoring Fabric (DMF).

DMF delivers network traffic analysis, data analytics and contextual insights to enterprises looking for network-wide observability.

We focus on securing the network through features native to EOS, such as segmentation, as well as Network Detection and Response (NDR).

Our NDR is an advanced solution that delivers answers, not alerts.

By combining AI with human expertise, Arista NDR autonomously hunts for both insider and external attacker behaviors, while providing triage, digital forensics and incident response across the entire network — campus workspace, data center , IoT, operational technology (OT) and cloud networks.

Lower Total Cost of Ownership

Our cloud networking platform offers architectural and system advantages that provide our customers with cost-effective and highly available cloud networking solutions.

Monitoring network performance in real time is a core need for current and next-generation architectures to run dependably.

Arista EOS and CloudVision bring a modern approach to network telemetry and a replacement for legacy polling mechanisms.

CloudVision Analytics engines and CloudVision Telemetry Apps leverage the state streaming infrastructure of EOS and NetDL to give Arista customers an unprecedented level of visibility into their network operations.

Arista EOS also supports multiple telemetry tracers that bring deeper workload-level visibility by integrating with distributed applications like big data, cloud, container and virtualized environments.

Arista focuses on building security into the networking layers through features native to EOS, such as segmentation and encryption, as well as NDR powered by AI.

Arista’s zero trust networking principles, based on NIST 800-207, help customers address security challenges with three cornerstones:

visibility, continuous diagnostics, and enforcement.

The Arista NDR platform delivers continuous diagnostics for the entire enterprise threat landscape, processes countless points of data, senses abnormalities or threats, and reacts if necessary.

Operational Efficiency

Automation is critical to delivering operational efficiency and Arista has taken the lessons learned from building large public clouds and engineered network automation into our CloudVision management platform.

CloudVision eliminates burdensome manual tasks so organizations can become more agile in making changes to network infrastructure.

Fewer manual configurations enable faster time to service and improved availability for our customers.

Our open APIs enable standards-based integration with third-party tools as well, supporting additional automation capabilities.

Total Cost of Ownership

We believe our programmable, scalable leaf-spine architectures, combined with our applications, significantly reduce networking costs when compared to legacy network designs, enabling faster time to service and improved availability.

2 unchanged sentences

Cognitive Campus Workspace

Our solutions extend the principles of cognitive cloud computing to campus networks, with the intent of disrupting the wasteful, oversubscribed legacy three-tier architecture of access-aggregation-core.

The recent IoT-related expansion of devices to more disparate locations and functions has significantly increased the complexity and importance of campus networks.

We are addressing these campus trends by leveraging our single-image EOS extended across wired and wireless workspaces, data center, routing and multi-cloud to drive consistent cognitive controls and analytics for the network end-to-end.

Using this cloud-based approach reduces operational costs by incorporating a network that is a seamless end-to-end solution rather than silos of different places in the network.

Our cognitive campus workspace is a data-driven model coupled with a unified dashboard for wired/wireless edge for next generation zero touch campus deployments.

Together, with zero trust security, the cognitive campus drives multifaceted visibility for IoT and OT applications.

Our Cognitive Campus Networking solutions are based on three principles:

Universal Cloud Network - We offer our Universal Cloud Network as an alternative to brittle, proprietary solutions from legacy vendors.

Our Universal Cloud Network is an open, standards-based design focusing on data-driven control principles.

Our collapsed Spline™ approach consolidates traditional campus core and aggregation layers into a simple single tier with high availability.

Cognitive Management Plane - There is a void in management plane consistency and a need for data-driven analytics in the campus, as in the data center.

We believe that a common model can be applied across both footprints, saving customers operational costs.

The Cognitive Management Plane (CMP) is a data-driven repository for the automated actions across network analytics.

Securing The Campus - Securing the campus requires a holistic approach to network segmentation, device compliance and auditing, as well as service integration with our security partners.

We deliver these capabilities through EOS, DMF, NDR and CloudVision.

Arista’s Cognitive campus is based on a data driven architecture and offers consistent, unified management across the campus edge for wired and wireless networks as well as integrated security & proactive network assurance.

Our Cognitive Campus Networking solutions are based on three capabilities:

Universal Cloud Network (UCN) - Offered as an alternative to brittle, proprietary solutions from legacy vendors, Arista UCN is an open, standards-based design focusing on data-driven control principles.

Arista’s Spline TM architecture, 7300 Series spine switches, 720/750 Series POE leaf switches, and Wi-Fi platforms consolidate campus layers into simpler topologies that reduce costs and improve reliability.

Cognitive Operations - The Cognitive management features built into the Arista CloudVision rely on NetDL to collect real time streaming telemetry from across the campus network and automates many critical IT functions.

These features provide real time visibility into the state of the network including traffic flows.

CloudVision’s AI-enabled AVA leverages data from NetDL for AI/ML-driven outcomes, helping to detect anomalies in the network, identify root causes and offer recommendations for mitigation.

The Wi-Fi access points in conjunction with CloudVision also provide proactive network assurance to monitor end user experience without the need for an overlay network.

Zero-trust Network Securit y - Securing the Campus requires a built-in approach to network segmentation, encryption, device compliance and auditing, as well as service integration with Arista’s security partners.

Arista delivers these capabilities through EOS and CloudVision AVA.

Arista campus leaf switches have an integrated AVA sensor that enables the access layer to provide Arista NDR capability without the complexity and cost of additional network monitoring devices.

Arista’s Macro Segmentation Service Group (MSS-G) provides a simpler, standards-based approach to segmenting traffic in the campus that is more flexible than other proprietary solutions.

We believe Arista’s Wireless Intrusion Prevention Service (WIPS) provides strong security while eliminating false positives.

Our Competitive Strengths

We believe the following strengths will allow us to maintain and extend our technology leadership position in data-driven cloud networking and next-generation data center and campus workspace Ethernet products:

• Purpose-Built Cloud Networking Platform.

We have developed a highly scalable cloud networking platform that uses software to address the needs of large-scale cloud companies, cloud service providers, and large enterprises including financial services organizations, government agencies and media and entertainment companies, including AI, virtualization, big data and low-latency applications.

Purpose-Built Cloud Networking Platform - We have developed a highly scalable cloud networking platform that uses software to address the needs of large-scale cloud companies, cloud service providers, and large enterprises, including AI, virtualization, big data and low-latency applications.

As a result, our cloud networking platform does not have the inherent limitations of legacy network architectures.

• Broad and Differentiated Portfolio.

Using multiple merchant silicon architectures, we deliver switches, capable of routing, with industry-leading capacity, low latency, port density and power efficiency, and have innovated in areas such as deep packet buffers, highly available modular hardware, and reversible cooling options.

Broad and Differentiated Portfolio - Using multiple merchant silicon architectures, we deliver switches, capable of routing, with industry-leading capacity, low latency, port density and power efficiency, and have innovated in areas such as deep packet buffers, highly available modular hardware, and reversible cooling options.

Our broad portfolio has allowed us to offer customers products that best match their specific requirements.

• Single Binary Image Software.

The single binary image of EOS software allows us to maintain feature consistency across our entire product portfolio and enables us to introduce new software innovations into the market that become available to our entire installed base without a “forklift upgrade” (i.e., a broad upgrade of the data center infrastructure).

• Rapid Development of New Features and Applications.

Our highly modular EOS software has allowed us to rapidly deliver new features and applications while preserving the structural integrity and quality of our network operating system.

Single Binary Image Software - The single binary image of EOS software allows us to maintain feature consistency across our entire product portfolio and enables us to introduce new software innovations into the market that become available to our entire installed base without a “forklift upgrade” (i.e., a broad upgrade of the data center infrastructure).

Rapid Development of New Features and Applications - Our highly modular EOS software has allowed us to rapidly deliver new features and applications while preserving the structural integrity and quality of our network operating system.

We believe our ability to deliver new features and capabilities more quickly than legacy switch/router operators provides us with a strategic advantage given that the requirements in cloud and next-generation data center and campus networking continue to evolve rapidly.

• Deep Understanding of Customer Requirements.

We have developed close working partnerships with many of our largest customers that provide us with insights about their needs and future requirements.

This has allowed us to develop and deliver products to market that meet customer demands and expectations as well as to rapidly grow sales to existing customers.

• Strong Management and Engineering Team with Significant Data Center Networking Expertise.

Our management and engineering team consists of networking veterans with extensive data center and campus networking expertise.

Our President and Chief Executive Officer, Jayshree Ullal, has over 30 years of networking expertise from silicon to systems companies.

Andy Bechtolsheim, our Founder and Chief Development Officer, was previously a founder and chief system architect at Sun Microsystems.

Kenneth Duda, our Founder and Chief Technology Officer, leads the software development effort of EOS.

• Significant Technology Lead.

We believe that our networking technology represents a fundamental advance in networking software.

Deep Understanding of Customer Requirements - We have developed close working partnerships with many of our largest customers that provide us with insights into their needs and future requirements.

This has allowed us to develop and deliver products to the market that meet customer demands and expectations as well as to rapidly grow sales to existing customers.

Strong Management and Engineering Team with Significant Data Center Networking Expertise - Our management and engineering team consists of networking veterans with extensive data center and campus networking expertise.

Our Chief Executive Officer and Chairperson, Jayshree Ullal, has over 40 years of networking expertise from silicon to systems companies, and Kenneth Duda, our Founder, Chief Technology Officer and Director, leads our software development team including EOS.

Our technical team also includes highly experienced leaders such as Hugh Holbrook, our Chief Development Officer, who leads our platform driven software engineering, and Andy Bechtolsheim, our Founder and Chief Architect, who was previously a founder and chief system architect at Sun Microsystems.

Significant Technology Lead - We believe that our networking technology represents a fundamental advance in networking software.

Our EOS software is a key cloud networking software stack that is state-driven and a result of tremendous research and development efforts.

Our Products and Technology

Our portfolio of products and technology consists of our Core Data Center/Cloud Switching Products, our Adjacent Campus and Routing Products and our Network Software and Services.

Extensible Operating System

Our portfolio of products and technology consists of Core Data Center/Cloud/AI Switching Products, Adjacent Campus and Routing Products and Network Software and Services.

Extensible Operating System (EOS)

The core of our cloud networking platform is our data-driven operating system, EOS, which runs on top of standard Linux and offers programmability at all layers of the stack.

System state and data are stored in EOS and maintained in a highly

efficient, centralized system database where data is accessed via an automated publish/subscribe model.

System state and data are stored in EOS and maintained in a highly efficient, centralized system database where data is accessed via an automated publish/subscribe model.

This distinct design principle provides module independence, self-healing resiliency, and multi-process software stability.

We have continued to evolve the EOS software stack transforming the centralized EOS network database into a multi-modal, multi-tenant, capable data lake.

The EOS Network Data Lake (NetDL™) unifies the multiple data types gathered in a network and allows for external data ingestion and enrichment.

NetDL™ aggregates data from systems, platforms, and services enabling smoother operations between NetOps, CloudOps, and DevOps operators.

We believe we are extending classic Software-Defined Networking (SDN) principles to data-driven networking control and an AI-enabled holistic view with NetDL™, while building on our core pillars of reliability, open standards, and programmability.

EOS Attributes

The modular and programmable architecture of EOS enables us to offer a set of attributes, capabilities and features that are essential for cloud networking and next-generation data centers.

High Availability

EOS is self-healing in the sense that individual processes can be restarted without impacting application traffic.

This architectural design principle supports self-healing resiliency in our software, easier software maintenance and module independence, higher software quality overall, and faster time-to-market for new features that customers require.

Programmable at All Layers

EOS is programmable at all layers from the Linux kernel to switch configuration, provisioning, automation and detailed monitoring of the network.

Customers can leverage tools such as the EOS Software Development Kit (“SDK”) and eAPI to implement fully customized infrastructure automation solutions.

Workflow Visibility

Through EOS, we have developed a wide range of applications available to our customers for purchase as additional licenses that enable enhanced network monitoring and visibility without requiring additional external monitoring devices.

This includes (i) DANZ, which provides access to raw network data for analysis by security, troubleshooting and performance management tools, (ii) Latency/loss ANalyZer (LANZ), which provides access to internal network performance loads and packet loss and latency occurring at the microsecond level, (iii) Network Telemetry, which provides network state information including correlations with the dynamic state of the systems operating on the network such as Hypervisors and, distributed job controls, and (iv) Network Tracers, which provide active integration and diagnostics for various workload conditions dependent upon network performance.

Network Automation

EOS supports Puppet, Chef and Ansible, which enable automatic network configuration in the same manner as servers and storage.

In addition, CloudVision provides visibility, orchestration and automation that greatly reduce network operational costs.

Core Datacenter/Cloud

We offer one of the broadest product lines of datacenter 1/2.5/5/10/25/40/50/100/400/800 Gigabit Ethernet switches in the industry, comprising of 7050/7060 Series, 7130 Series, 7150 Series, 7160 Series, 7170 Series, 7280 Series, 7500 Series and 7800 Series data-driven platforms.

We deliver switching platforms with industry-leading capacity, low latency, port density and power efficiency.

We have also innovated in areas such as deep packet buffers, embedded optics and reversible cooling.

Adjacent Campus and Routing

Cognitive Campus Switching and Routing

Our adjacent products include our Cognitive Campus switching products such as our 720XP fixed PoE switches, 750 modular PoE switches and 7300X3 spline switches, as well as our Universal Spine and Leaf Routing products such as our 7020R fixed routers, 7280R fixed routers, 7368X4 modular router, 7500R modular routers and 7800R modular routers.

Cognitive WiFi

Cognitive WiFi consists of our access point solutions (“APs”) that are tailored for a controller-less wireless network.

These APs are available in disaggregated options harnessing the power of cloud, machine learning and cognitive networking.

By integrating with CloudVision, Cognitive WiFi is based on a similar CMP model for cognitive analytics unifying the operational experience across wired and wireless.

It enhances real-time insight into the experience of WiFi clients to connect and utilize the network.

Cognitive WiFi also includes a suite of WiFi Tracer tools for wireless security, reachability and network health diagnostics.

The EOS NetDL TM unifies the multiple data types gathered in a network and allows for external data ingestion and enrichment.

NetDL TM aggregates data from systems, platforms, and services enabling smoother operations between NetOps, CloudOps, and DevOps operators.

EOS is packaged as a perpetual license on Arista hardware platforms, virtualized EOS (vEOS) for production or simulations use cases or containerized EOS (cEOS) for flexible platform support including third-party hardware.

Core Datacenter/Cloud/AI

Arista offers one of the broadest product portfolios of data-driven, high-speed, cloud and datacenter Ethernet switches.

Built on top of the superior quality and openness of Arista EOS, we deliver high performance, industry-leading capacity, ultra-low latency, feature rich, and powerful efficient solutions for customers building and operating cloud networks.

Our core switching portfolio contains both fixed and modular form factors, varying port configurations and densities, and options in power delivery all driven by customer requirements.

Arista also provides solutions for compute, GPU and storage interconnects in driving AI/ML workloads, leveraging its IP/Ethernet switches to deliver unparalleled performance and scalability.

With the exponential growth of AI applications, the need for standardized transport like Ethernet becomes paramount, enabling a power-efficient interconnect while overcoming the complexities of traditional approaches.

The 7800R AI spine and 7060X AI leaf switches, coupled with Arista's EOS innovations such as AI Analyzer along with optimal load balancing solutions offer compelling solutions for contemporary AI applications and deployment.

We continue to innovate with every generation of switching platforms.

Most recently, Arista has brought to market the Arista 7800R AI Spine and the 7060 AI Leaf to address the demanding scale and performance requirements driven by large-scale AI networks.

We also continue to be innovative in areas concerning deep packet buffer architectures, virtual output queuing, non-disruptive upgrades, embedded optics and next-generation optics, reversible cooling and overall system power efficiency.

Cognitive Adjacencies

Cognitive Campus Switching

Arista’s Cognitive Campus switching products, powered by EOS, offer consistent, unified management across the campus edge for wired and wireless networks as well as integrated security & proactive network assurance.

Our campus products include the Arista 7300 Series spine/SplineTM, 720/750 Series POE switches, and a broad range of indoor and outdoor Cognitive Wi-Fi Access Points.

Cloud-Grade Routing

Arista’s Cloud-Grade Routing platforms, powered by EOS, combine high performance routing, high port density, deep buffers, integrated DWDM and wirespeed encryption.

Our 7280R3 Universal Leaf and 7500R3 and 7800R3 Universal Spine platforms serve a variety of use cases including high speed multi-cloud connect, Data Center Interconnect (DCI), controller-based traffic engineering, peering, business VPNs, core routing and Secure Enterprise edge routing.

WAN Routing System

Arista’s new 5000 Series of WAN Routing System, powered by EOS, offers high performance and scale to meet enterprise modern WAN edge and aggregation requirements.

The 5000 Series supports 1/10/100GbE interfaces and flexible network modules while delivering from 5Gb to over 50Gb of bidirectional encrypted traffic with high VRF and tunnel scale.

Networking Software and Services

CloudVision is our network management plane solution for workload orchestration and workflow automation, which delivers a turnkey solution for cloud networking.

CloudVision’s abstraction of the physical network to a broader, network-wide perspective provides a simplified approach for consistent network operations across network domains, including data center, campus wired and wireless, routing interconnect, and multi-cloud networks.

CloudVision highlights include:

Centralized representation of distributed state based on NetDL™, allowing for a single point of integration and network-wide visibility and analytics;

Controller-agnostic support for physical and virtual workload orchestration through open APIs;

Turn-key automation for zero touch provisioning, configuration management and network-wide upgrades and rollback;

Compliance dashboard for security, audit and patch management;

Cognitive AI/ML, driven by our Autonomous Virtual Assist ("AVA") for dynamic insights and recommendations, built on a modern approach of real-time streaming for telemetry and as a replacement for legacy polling per device;

Granular visibility and historical troubleshooting with predictive insights across the unified edge wired and wireless networks, including IoTvision;

and finally Multi-domain segmentation for the zero trust enterprise, enabling macro-segmentation services (MSS®) for an open and scalable approach for network policy management and with dynamic integrations into security management systems from Arista’s security ecosystem partners.

DANZ Monitoring Fabric (DMF)

DANZ Monitoring Fabric (DMF) is a next-generation network packet broker (NPB) designed for pervasive, organization-wide network observability and security visibility, enabling IT to deliver multi-tenant monitoring-as-a-service.

Leveraging Arista's high-performance and versatile 1G/10G/25G, 40G/100G and 400G Ethernet switch platforms with DMF, IT operators can pervasively monitor all user, device/IoT and application traffic (north-south and east-west) by gaining complete visibility into physical, virtual and container environments.

DMF switch licenses are procured as subscription software.

Additionally, DMF’s advanced services include deep hop-by-hop visibility, predictive analytics, contextual insights and scale-out packet capture — integrated through a single dashboard — to provide simplified network performance monitoring (NPM) and SecMon workflows for real-time and historical context.

For enterprises, service providers and cloud providers, DMF provides a one-stop network observability solution for production data centers, enterprise campus/branch and 4G/5G mobile networks.

Arista Network Detection and Response (NDR)

The AI-driven Security Platform, driven by our AVA, deeply analyzes billions of network communications to autonomously discover, profile and classify every device, user and application across any network.

Using a multi-dimensional ensemble machine learning approach, Arista NDR then models complex adversarial behaviors and detects threats by connecting the dots across entities, time, protocols and attack stages.

By combining artificial intelligence with human expertise, Arista NDR hunts for both insider and external attacker behaviors, while providing triage, digital forensics and incident response across the entire network.

CloudEOS™ is our multi-cloud and cloud-native networking solution enabling a highly secure and reliable networking experience with consistent segmentation, telemetry, provisioning and troubleshooting for the entire enterprise.

It can be deployed across the enterprise edge, WAN, campus workspace, data center, on-premises Kubernetes clusters, and multiple public and private clouds.

CloudEOS provides multi-cloud connectivity across the entire enterprise cloud environment with high-performance virtual and container-based instances of EOS software that simplify network operations and integrate with declarative cloud provisioning toolchains like Terraform, Ansible, and other popular CloudOps and DevOps tools.

CloudEOS is designed for consumption on Amazon AWS, Microsoft Azure, and Google public clouds via their marketplace and service catalogs, and it is also available as a cloud-native instance for deployment in Kubernetes clusters.

With CloudEOS and CloudVision, customers can integrate their cloud network deployments with the elasticity and automation of the public cloud, private cloud and cloud native platforms.

CloudVision is Arista’s modern, multi-domain management platform that leverages cloud networking principles to deliver a simplified end-to-end network operations experience for our Enterprise market.

Unlike traditional domain-specific management solutions, CloudVision enables consistent zero touch network operations across data center, campus wired and wireless, routing interconnect and multi-cloud networks helping to break down the complexity of siloed management approaches.

CloudVision is built on Arista’s NetDL architecture and leverages real-time network state to provide an abstraction of the physical network to a broader, network-wide perspective allowing for a more efficient approach for several operational and network telemetry capabilities.

CloudVision’s cloud-native architecture gives customers a choice to consume CloudVision as a subscription service or an on-premise licensed appliance.

Arista A-Care Services

We have designed our customer support offerings, Arista A-Care Services, to provide our customers with high levels of support.

Our global team of support engineers engages directly with client IT teams and is available at all times over e-mail, by phone or through our website.

Our global team of support engineers engages directly with client IT teams and is always available over e-mail, by phone or through our website.

We offer multiple service options that allow our customers to select the product replacement service level that best meets their needs.

We stock spare parts in over 200 locations around the world through our third-party logistics suppliers.

All of our service options include unlimited access to bug-fixes, new -feature-releases, online case management and our community forums.

All our service options include unlimited access to bug-fixes, new-feature-releases, online case management and our community forums.

DANZ Monitoring Fabric (DMF)

DANZ Monitoring Fabric (DMF) is a next-generation network packet broker (NPB) architected for pervasive, organization-wide visibility and security, delivering multi-tenant monitoring-as-a-service.

Leveraging Arista's high-performance and versatile Ethernet switch platforms with DMF, IT operators can pervasively monitor all user, device/IoT and application traffic (north-south and east-west) by gaining complete visibility into physical, virtual and container environments.

Deep hop-by-hop visibility, predictive analytics, contextual insights and scale-out packet capture, integrated through a single dashboard, enables simplified network performance monitoring (NPM) and SecMon workflows for real-time and historical context across production data centers, enterprise campus/branch and 4G/5G mobile networks.

DMF switch licenses are sold as subscription licenses.

Arista Guardian Network Identity (AGNI)

To overcome the new security challenges and the explosion of clients in today’s perimeter-less enterprise networks, Arista delivers a novel AI-driven network Identity service, Arista Guardian for Network Identity (AGNI) to connect the network, users, and devices across remote and geographically dispersed locations.

Based on Arista’s flagship CloudVision, the new AGNI platform brings scale, simplicity, and security across users, their associated endpoints, and IoT devices.

AGNI integrates with Arista NDR and other third-party XDR and EDR solutions for post-admission control functionality.

AGNI is sold as subscription licenses.

Arista's AI-driven Network Detection and Response (NDR)

An AI-driven Security Platform, powered by AVA TM , Arista NDR analyzes billions of network communications to autonomously discover, profile and classify every device, user, and application across perimeter, core, IoT, and cloud networks.

Based on this deep understanding of the attack surface, the platform then detects threats to and from these entities, while providing the context necessary to respond rapidly.

The analysis begins with AVA Sensors that span the network and perform deep packet inspection.

These sensors are available in a variety of form factors:

physical hardware, virtual, cloud-based, and now also incorporated into Arista campus switches.

Arista NDR licenses are sold as a subscription license.

CloudEOS is Arista’s multi-cloud and cloud-native networking solution supporting autonomic operation to deliver an enterprise-class, highly secure, and reliable networking experience for any cloud.

As part of the Arista EOS and CloudVision product family, it delivers consistent segmentation, automation, telemetry, provisioning and troubleshooting for the enterprise edge, WAN, campus, data center and multiple public and private clouds.

To provide a scalable and automated network experience, CloudEOS integrates with Arista CloudVision to simplify the operator's experience of interconnecting and managing multi-cloud, cloud-native and on-premises enterprise networks.

Our portfolio of products and technology consists of our Core Data Center/Cloud Switching Products, our Adjacent Campus and Routing Products and our Network Software and Services.

Sales and Marketing

1 unchanged sentence

We also sell in conjunction with various technology partners.

To facilitate channel coordination and increase productivity, we have created a partner program, the Arista Partner Program, to engage partners who provide value-added services and extend our reach into the marketplace.

Authorized training partners perform technical training of our channel partners and end customers.

1 unchanged sentence

Our partners generally do not stock inventory received from us.

Our sales organization is supported by systems engineers with deep technical expertise and responsibility for pre-sales technical support and solutions engineering for our end customers, systems integrators, OEMs, and channel partners.

Our sales organization is supported by systems engineers with deep technical expertise and responsibility for pre-sales technical support and solutions engineering for our customers, systems integrators, OEMs, and channel partners.

In general, the personnel in our sales organization are formed into teams, and each team is responsible for a geographical territory, has responsibility for a number of major direct end-customer accounts or has assigned accounts in a specific vertical market.

1 unchanged sentence

Our marketing activities consist primarily of technology conferences, webinars, web marketing, trade shows, product demonstrations, seminars and events, public relations, analyst relations, demand generation and direct marketing to build our brand, increase end-customer awareness, communicate our product advantages and generate qualified leads for our field sales force and channel partners.

Our rapid historical growth may have reduced the impact of seasonal or cyclical factors that might have influenced our business to date.

As our increasing size may cause our growth rate to slow, seasonal or cyclical variations in our operations may become more pronounced over time and may materially affect our business, financial condition, results of operations and prospects.

We operate on a December 31st year-end and have typically experienced higher sequential product revenue growth in the fourth quarter, followed by revenue that is sequentially flat-to-declining in the first quarter of the following year.

We believe that this seasonality results from a number of factors, including the procurement, budgeting and deployment cycles of many of our end customers.

In addition, industry-wide supply chain shortages have resulted in extended lead times and have impacted, and may continue to impact our ability to manufacture and ship products to our customers in a timely manner, which may disrupt typical seasonal trends.

We operate on a December 31st year end and typically have lower sequential quarter over quarter revenue growth in the first quarter of each fiscal year, often followed by stronger sequential revenue growth in the ensuring quarters.

We believe that this seasonality results from a number of factors, including the procurement, budgeting and deployment cycles of many of our customers.

The effects of recent supply chain disruptions and our rapid growth may have reduced the impact of seasonal or cyclical factors that might otherwise have influenced our business and broader industry performance.

If our growth rates slow, seasonal or cyclical variations in our operations may become more pronounced over time and may materially affect our business, financial condition, results of operations and prospects.

In addition, any supply chain shortages and manufacturing disruptions that result in extended lead times may impact our ability to manufacture and ship products to our customers in a timely manner, which may disrupt typical seasonal trends.

Research and Development

We believe our future success depends on our ability to develop new products and features that address the needs of our end customers.

Our in-house engineering personnel are responsible for the development, quality, testing, documentation, support and release of our products.

We plan to continue to invest in resources to conduct our research and development efforts.

Our success relies on the timely enhancement of existing products and the development of new solutions and features to address changing customer needs and technological advancements.

Our in-house engineering personnel are responsible for the development, testing, documentation, support and release of our products.

We have a highly skilled team of software and hardware engineers with extensive experience in wired and wireless networking technologies, network protocols, network security, operating systems, programming languages, compilers, databases, hardware system design, Field-Programmable Gate Array programming, high-speed signal integrity, and other related technologies.

Our research and development strategy focuses on advancing our core products and expanding into new markets while maintaining high product quality.

We are focusing research and development efforts in (1) adapting EOS for new silicon architectures, especially to support the unique requirements of AI workloads;

(2) adding or enhancing EOS control plane and management plane functionality;

(3) expanding our CloudVision management stack with enhanced automation, provisioning, monitoring, and security capabilities;

(4) building related services, such as NDR and Network Access Control (NAC);

(5) improving the security and scalability of our software development infrastructure and software supply chain;

and (6) maintaining high product quality throughout.

We have dedicated significant time and resources in test automation, ensuring high test pass rates, and working with our support group whenever customers experience technical defects in our products.

Collaboration with customers and other industry leaders is integral to our approach, though uncertainties persist regarding the successful development and market acceptance of emerging technologies.

Looking ahead, we plan to continue to invest in resources to conduct our research and development efforts to evolving and extending our portfolio's capabilities, ensuring our products continue to address dynamic market needs and solidify our industry leadership.

Manufacturing

We subcontract the manufacturing of all of our products to various contract manufacturers.

Our primary manufacturing partners are Jabil Circuit, Sanmina Corporation, Flex Ltd.

We subcontract the manufacturing of all our products to various contract manufacturers.

Our primary manufacturing partners are Jabil Inc., Sanmina Corporation, Flex Ltd.

and Foxconn Hon Hai.

This approach allows us to reduce our costs, manufacturing overhead and inventory position and allows us to adjust more quickly to changing end-customer demand.

require all of our manufacturing locations to be ISO-9001 certified.

We require all our manufacturing locations to be ISO-9001 certified.

We have four direct fulfillment facilities worldwide to hold finished goods inventory, perform product transformations, and install our EOS software to ship to customers and partners.

Our contract manufacturing partners procure the majority of the components needed to build our products and assemble our products according to our design specifications.

This allows us to leverage the purchasing power of our contract manufacturing partners.

This allows us to leverage the purchasing power of our contract

manufacturing partners.

We retain complete control over the bill of materials, test procedures and quality assurance programs.

Our personnel work closely with our partners and review on an ongoing basis forecasts, inventory levels, processes, capacity, yields and overall quality.

Our personnel work closely with our partners and review on an ongoing basis the forecasts, inventory levels, processes, capacity, yields and overall quality.

Our contract manufacturing partners procure components and assemble our products based on our demand forecasts.

These forecasts represent our estimates of future demand for our products based upon historical trends and analyses from our sales and product management functions as adjusted for overall market conditions.

For example, industry-wide supply chain shortages have resulted in extended lead times for components and required us to extend the time horizon of our demand forecasts, and increase our purchase commitments for long lead time components.

For example, when industry-wide supply chain shortages resulted in extended lead times for components, we were required to extend the time horizon of our demand forecasts and increase our purchase commitments for long lead time components.

Our products rely on key components, including merchant silicon, integrated circuit components and power supplies purchased from a limited number of suppliers, including certain sole source providers.

2 unchanged sentences

The supply of components may also be adversely affected by geopolitical conditions such as international trade wars like the U.S.

trade war with China and the impact of public health epidemics like the coronavirus.

trade war with China, the Houthi attacks on marine vessels in the Red Sea, and the impact of public health epidemics like COVID-19.

Our product development efforts also depend upon continued collaboration with our key suppliers, including our merchant silicon vendors such as Broadcom and Intel.

1 unchanged sentence

This enables us to focus our development resources on software core competencies and to leverage the investments made by merchant silicon vendors to achieve cost-effective solutions.

Once the completed products are manufactured and tested, our contract manufacturing partners ship them to various direct fulfillment facilities in the United States, the Netherlands and Singapore for final configuration, quality-control inspection and shipment to our distribution partners and end customers.

After the products are shipped to our end customers, our products are installed by the end customers or by third-party service providers such as system integrators or value-added resellers on their behalf.

After the products are shipped to our customers, our products are installed by the customers or by third-party service providers such as system integrators or value-added resellers on their behalf.

The markets in which we compete are highly competitive and characterized by rapidly changing technology, changing end-customer needs, evolving industry standards, frequent introductions of new products and services and industry consolidation.

We expect competition to intensify in the future as the market for cloud networking expands and existing competitors and new market entrants introduce new products or enhance existing products.

The data center and campus networking markets have been historically dominated by Cisco, with competition also coming from other large network equipment and system vendors, including Extreme Networks, Dell/EMC, Hewlett Packard Enterprise and Juniper Networks.

We expect competition to intensify in the future as the market for cloud and AI networking expands and existing competitors and new market entrants introduce new products or enhance existing products.

The data center and campus networking markets have been historically dominated by Cisco, with competition also coming from other large network equipment and system vendors, including Nvidia, Extreme Networks, Dell/EMC, Hewlett Packard Enterprise and Juniper Networks.

Most of our competitors and some strategic alliance partners have made acquisitions and/or have entered into or extended partnerships or other strategic relationships to offer more comprehensive product lines, including cloud networking solutions.

For example, Broadcom acquired Brocade Communications Systems and entered into an agreement to acquire VMWare, Extreme Networks purchased certain data center networking assets from Broadcom/Brocade and Avaya as well as Aerohive Networks, Dell acquired EMC, Hewlett Packard Enterprise acquired Aruba Networks and Juniper Networks acquired Mist Systems.

We also face competition from other companies and new market entrants, current technology partners and end customers who may acquire or develop network switches and cloud service solutions for internal use and/or to broaden their portfolio of products to market and sell to customers.

Some of these competitors are developing networking products based on off-the-shelf or commoditized hardware technology, or “white box” hardware, particularly where an end customer’s network strategy seeks to emphasize deployment of such product offerings or adopt a disaggregated approach to the procurement of hardware and software.

End customers may also increase their adoption of networking solutions based upon open-source network operating systems that may be provided for free and used either on “white box” or proprietary hardware.

of new competitors into our markets or the increased adoption of these new technology solutions or consumption models may cause downward pricing pressures, result in lost sales or otherwise have a material adverse effect on our business, prospects, financial condition and operating results.

For example, Broadcom has acquired VMWare and Hewlett Packard Enterprises has entered into an agreement to acquire Juniper Networks.

With the emergence of AI networking, new competitive technologies may enter the market to address the requirements of AI clusters.

Ethernet, today, faces competition from both InfiniBand (IB) and NV Link interconnects for back-end AI networking clusters.

IB has traditionally been used in supercomputer clusters due to its high reliability, low latency and high bandwidth.

Both IB and NV Link are often sold as part of a vertical solution along with the GPUs from Nvidia.

We also face competition from other companies and new market entrants, current technology partners and customers who may acquire or develop network switches and cloud service solutions for internal use and/or to broaden their portfolio of products to market and sell to customers.

Some of these competitors are developing networking products based on off-the-shelf or commoditized hardware technology, or “white box” hardware, particularly where a customer’s network strategy seeks to emphasize deployment of such product offerings or adopt a disaggregated approach to the procurement of hardware and software.

Customers may also increase their adoption of networking solutions based upon open-source network operating systems that may be provided for free and used either on “white box” or proprietary hardware.

The entrance of new competitors into our markets or the increased adoption of these new technology solutions or consumption models may cause downward pricing pressures, result in lost sales or otherwise have a material adverse effect on our business, prospects, financial condition and operating results.

In the NDR market, our Arista NDR offerings compete with other network security vendors including Cisco, DarkTrace, and ExtraHop.

Lastly, in the network packet broker (NPB) market, Arista DANZ Monitoring Fabric competes with Gigamon, Keysight, and other network monitoring software providers.

In the network packet broker (NPB) market, Arista DANZ Monitoring Fabric (DMF) competes with Gigamon, Keysight, Cisco, Netscout, AVIZ networks and other network monitoring software providers.

Our relationships with our strategic alliance partners or suppliers may also shift as industry dynamics change.

14 unchanged sentences

We believe the combination of EOS, a set of network applications and our 1/2.5/5/10/25/40/50/100/200/400/800 Gigabit Ethernet platforms make our offering highly competitive for both cloud and enterprise data centers.

However, many of our competitors have greater name recognition, longer operating histories, larger sales and marketing budgets and resources, broader distribution and established relationships with channel partners and end customers, greater access to larger end-customer bases, greater end-customer support resources, greater manufacturing resources, the ability to leverage their sales efforts across a broader portfolio of products, the ability to leverage purchasing power when purchasing subcomponents, the ability to bundle competitive offerings with other products and services, the ability to develop their own silicon chips, the ability to set more aggressive pricing policies, lower labor and development costs, greater resources to make acquisitions, larger intellectual property portfolios and substantially greater financial, technical, research and development or other resources.

While we have extended our demand planning horizons and increased our component purchase commitments, there is no guarantee that suppliers will meet their commitments or that ultimate customer demand will directly match our demand forecasts.

However, many of our competitors have greater name recognition, longer operating histories, larger sales and marketing budgets and resources, broader distribution and established relationships with channel partners and end customers, greater access to larger customer bases, greater customer support resources, greater manufacturing resources, the ability to leverage their sales efforts across a broader portfolio of products, the ability to leverage purchasing power when purchasing subcomponents, the ability to bundle competitive offerings with other products and services, the ability to develop their own silicon chips, the ability to set more aggressive pricing policies, lower labor and development costs, greater resources to make acquisitions, larger intellectual property portfolios and substantially greater financial, technical, research and development or other resources.

Intellectual Property

Our success and ability to compete depend substantially upon our core technology and intellectual property.

We rely on patent, trademark and copyright laws, trade secret protection and confidentiality agreements with our employees, end customers, resellers, systems integrators, manufacturers, and others to protect our intellectual property rights.

We rely on patent, trademark and copyright laws, trade secret protection and confidentiality agreements with our employees, customers, resellers, systems integrators, manufacturers, and others to protect our intellectual property rights.

We file U.S and foreign patent applications to protect our intellectual property and believe that the duration of our issued patents is adequate relative to the expected lives of our products.

6 unchanged sentences

and other jurisdictions, and Arista, EOS, and CloudVision are among our core trademarks.

We control access to and use of our software, technology and other proprietary information through internal and external controls, including contractual protections with employees, contractors, end customers and partners.

Our software is protected by U.S.

and international copyright, patent and trade secret laws.

Despite our efforts to protect our software, technology and other proprietary information, unauthorized parties may still copy or otherwise obtain and use our software, technology and other proprietary information.

Despite our efforts to protect our software,

technology and other proprietary information, unauthorized parties may still copy or otherwise obtain and use our software, technology and other proprietary information.

In addition, we intend to expand our international operations, and effective patent, copyright, trademark and trade secret protection may not be available or may be limited in foreign countries.

19 unchanged sentences

We are proud to be an S&P 500 company with both a female CEO and CFO.

We were also recognized by Comparably in 2022 amongst the Top 50 best large companies for diversity and Top 100 best large companies for culture.

We were also recognized by Comparably in 2023 amongst the Top 50 best large companies for diversity and Top 100 best large companies for happiness, work-life balances as well as compensation.

50% of our board of directors are women or underrepresented minorities.

2 unchanged sentences

We offer mentorship opportunities to our employees facilitated by our Women@Arista employee resource group.

In 2022, we initiated a returnship program primarily for female caregivers with Path Forward returning to the workforce.

In addition, we support under-represented employee affinity organizations and actively recruit from historically black colleges and universities, women’s colleges and Hispanic/Latinx and African-American professional societies.

We affirm the principle of equal employment opportunity without regard to any protected characteristic, including but not limited to race, religion, national origin, color, gender, age, disability, pregnancy, marital status, ancestry, military status or sexual orientation.

7 unchanged sentences

Compensation and Benefits

We provide competitive and comprehensive benefit packages that are designed to help employees make the best decisions for themselves, their family and their lifestyle.

Arista offers a broad variety of physical and mental wellness offerings to our global employees in a virtual as well as on-demand format, including fitness classes, webinars on practical wellness takeaways, strategies for stress reduction and a variety of other topics.

We provide competitive and comprehensive benefit packages that are designed to help and empower employees to make the best decisions for themselves, their family and their lifestyle.

In addition, in the United States, we offer our employees an employee stock purchase plan, healthcare and retirement benefits, paid time off and family leave, flexible time away, family planning benefits, backup resources for childcare and elder care, and other employee assistance programs including behavioral health and emotional support assistance.

3 unchanged sentences

Along with our traditional healthcare benefits, we have created a detailed injury and illness prevention program to better protect employees from occupational risks of injury or illness.

We periodically host wellness days, whose purpose is to raise awareness on health issues, increase education on preventive medicine and available services and shift employee behavior through interactive activities and live presentations.

We also maintain a community engagement program, which provides opportunities for our employees to engage in community service.

We periodically host wellness weeks, whose purpose is to raise awareness on health issues, increase education on preventive medicine and available services and shift employee behavior through interactive activities and live presentations.

We also maintain a community employee engagement program, which provides opportunities for our employees to engage in volunteering and community service.

Training and Development

Our employees receive periodic training on various topics, including our Code of Ethics and Business Conduct, information security, data privacy, intellectual property, anticorruption, and other topics.

We also offer a variety of webinars on health and wellness topics, fitness classes, and social engagement activities.

We also partner with non-profit organizations to provide volunteer opportunities to our employees.

In addition, Arista provides extensive training and accreditation opportunities to employees in Sales, Customer Engineering and Software Research and Development roles including our Arista Certified Engineering (ACE) certification program as well as mentorship programs facilitated by our Women@Arista employee resource group.

Additional career development content including management training is available through our E-Learning portal to facilitate a culture of lifelong learning.

Our engineering teams have the opportunity to further develop their technical skills through our internal Arista PREP Training Program, technical summits, and participation in industry conferences or associations.

We also offer a variety of webinars on physical and mental health, career development and financial wellness topics, fitness classes, and social engagement activities.

We also partner with non-profit organizations across the globe to provide volunteer opportunities to our employees.

Available Information

7 unchanged sentences

Compared sentence by sentence after normalising whitespace, quotation marks, case and digits, so re-formatting and restated figures do not read as changed language. Wording changes appear as one removal and one addition. The current filing and the prior one are authoritative.

What changed 10-K

Item 1. Business