Initiation Report: Arista Networks
A market leader exposed to a strong secular theme available at a compelling entry-point
I’ve started building a position in Arista after shares fell 30% from all-time highs. I expect to see more weakness in the shares over the next few months, and will be adding to holdings should that occur. I detail my thought process on Arista’s business and market position in the rest of this report.
Introduction
We recommend initiating a long position in Arista Networks (“ANET”, or “the Company” – NYSE:ANET), the market leading provider of high performance networking products and software, due its strong gearing to datacenter build-out and evolving network architecture trends. We believe ANET’s market position will enable the Company to be a secular winner over the medium and long term, driving above market returns.
Arista has demonstrated robust growth over the past several years driven by rising investments in datacenters, subsequently attracting the market’s attention and positioning the Company as a winner in the AI infrastructure theme. ANET substantially increased its guidance through 2024, as initial expectations of 12% revenue growth accelerated into 20% actual results, causing the market to revise up its long-term forecasts for growth, and consequently pushing the price of Arista’s stock up to a record of ~$130/share in late January 2025. The market is now reversing course as management has reiterated revenue guidance that was below expectations (17% growth in 2025), resulting in the stock trading down roughly 30% from recent highs to ~$90/share.
This volatility in the stock has opened up an opportunity to purchase ANET’s stock at a 21% discount to fair value in our base case assumption, with an expected compound annual return of close to 20% for the next five years, or 2.4x cash-on-cash, with strong confidence of a return in the range of 4 to 27% in our projection period. We believe that Arista has an exceptional management team that is heavily invested in the business, adding confidence that the Company will build on its market leadership, resulting in industry leading growth that supports our forecasts.
Competitive Moat
Arista Networks is the global leader in high-performance network switches, with an estimated 40-45% market share in the segment. The Company has developed a proprietary network operating system (EOS) that it deploys in tandem with internally designed networking hardware that is manufactured by third parties. At the end of FY 2024, revenues from network switches represented 65% of total revenue, while routing products and software amounted to 18% and 17% of total revenues, respectively.
In some ways, ANET’s competitive moat is a paradox; the success of the business is the result of an ecosystem that arises from the integration of the Company’s network software and hardware. However, rather than developing an ecosystem with high walls that forces customers to choose its products across an entire network, ANET’s software enables its customers to plug-and-play its products in their networks without the hardware and software lock-in that typically comes with deploying competing solutions. This self-reinforcing integration of software and hardware is strengthened by sticky customer relationships marked by joint engineering initiatives and deep customer support. Our breakdown of Arista’s moat is as follows.
A scalable, high-performance network operating system that enables TCO savings for customers
ANET’s EOS network operating system is provided for free to enable basic hardware functioning, with additional features like automation, telemetry and cloud integration available as add-ons on a subscription basis. Arista’s software has five key features that combine into a clear value proposition to customers operating enterprise, cloud, datacenter, or AI cluster networks: 1) an open architecture, 2) a modular operating system, 3) a non-containerized operating system, 4) automation, and 5) turn-key solution.
1. Open Architecture: The Company’s EOS platform results in low switching costs when adopting the Company’s products, as customers are able to deploy Arista products without upgrading existing hardware or operating systems. ANET’s EOS platform does this by seamlessly integrating Arista hardware into complex environments, such as architectures that combine enterprise, edge, cloud, and datacenter networks into an aggregated network. Arista’s software also enables compatibility with existing Software-Defined Networking (SDN) platforms, while enabling EVPN/VXLAN solutions without vendor lock-in, further reducing the incremental cost of adoption.
2. Modular OS: Arista’s software results in higher network reliability, increasing datacenter uptime that ultimately results in higher revenues, better unit operating costs, and improved ROIs on datacenter investments. A key principle of EOS is that individual network functions (routing, switching, security, and management) operate as separate processes, meaning that if one function fails, the entire system does not require a reboot. This feature is critically important in datacenters that target uptime in excess of 99%.
3. Non-containerized OS: By designing EOS as a non-containerized operating system, network efficiency is improved, resulting in higher speeds and lower relative power consumption. In a containerized OS, a network operating system operates at two layers; one layer where individual services operate (routing, switching, etc.) and one for the operating system that executes the services on the network. In contrast, a non-containerized network combines the individual services into the operating system, improving performance. The latter approach is particularly important in datacenter environments that require low latency or high throughput.
4. Automation: In short, Arista’s EOS leads to lower operating costs for network operators, as the platform’s automation features reduce the amount of labour required for network maintenance and operations. EOS has built-in network monitoring, management, and build-out tools, while also offering a web-based GUI, an integration layer (including for SDN), and network telemetry processing.
5. Turn-key solution: In addition to competing with peers that offer proprietary network components with high lock-in to vendor software, Arista also competes with white-box equipment providers that offer the lowest price point products but with no software support. Arista’s EOS software is fully developed and functional out-of-the box, in contrast to white-box switch providers like Broadcom, Mellanox, Intel, or Dell, where network operators need to develop their own operating systems with open-source solutions or integrate existing operating systems onto the white-box hardware. As a result, white-box hardware is often poorly optimized relative to Arista’s integrated offering and takes more time and effort to deploy.
Class leading hardware optimized for datacenters
ANET compliments its software with hardware that is developed to meet the highest performance requirements of its customers, particularly datacenter operators. The Company’s product portfolio advantages stem from: 1) hardware performance optimization via EOS, 2) system integration approach in product design to maximize innovation in customer pain points, and 3) industry leading performance.
1. Hardware performance optimization via software: Arista’s key differentiation is realizing that software has a greater impact on network performance than hardware innovation. Network components are essentially commodity products with marginal value-add available for hardware innovation. Arista’s hardware performance is thus enabled through its EOS system that is designed to work exclusively with Arista switches.
2. System integration approach to product development: Optimizing the network switch as a system of components that is best suited to customer applications is a source of value-add differentiation in an environment where individual components offer little discernible benefit to end users. When breaking down the costs of a network switch’s components, 35-50% is Switching ASIC, 10-15% is the network interface, 8-15% is optics and transceivers, 8-12% is PCBs and chassis, 7-10% is cooling systems, 7-10% is power supplies, and 5-20% is software. While ASICs represent the lion’s share of cost, ASIC design and manufacturing is capital intensive with little discernible value add. Arista has instead chosen to purchase ASICs, network interfaces, and optics from vendors and focus on designing chassis, cooling system and power supply components and outsourcing the manufacturing to third parties. While not as technically complex, these parts have dramatic impacts on switch reliability (overheated switches shut down, reducing network reliability) and power consumption (high efficiency switches reduce power costs). Arista’s R&D is thus targeted at addressing customer pain points, when peers have to spread dollars across developing a large portfolio of networking components and invest in CapEx to manufacture them.
3. Industry leading performance: The combination of Arista’s software and system integration approach to developing networking hardware results in best-in class performance in both low-latency and data-intensive applications. Low-latency networks typically have more consistent traffic rates and therefore can be optimized for speed, such as high-frequency trading firms, enterprise and cloud customers, or other low latency use cases. In contrast to low latency applications, certain networks have higher amounts of traffic concentrated in shorter periods, requiring switches that can handle this volatility. To accommodate these flows, switches need to have higher packet buffer sizes to ease congestion when traffic builds; Arista’s high-end products offer packet buffers up to 16GB, while Juniper switches are up to 12GB, and Cisco switches are up to 6GB. Applications that need deep packet buffers include storage, high-performance computing, and AI/ML.
Deep customer relationships and strong customer support
Given the importance of network performance in overall datacenter productivity, Arista has established sticky relationship with its customers, such as Microsoft, Meta, Oracle, and Apple. It has done so by establishing joint development programs with its customers over the past 14 years. As a result, Arista is able to gain a much better grasp of customer requirements and develop their products accordingly, adding to the flywheel effect of the Company’s R&D programs. In some cases, Arista has gone as far as developing customer-specific products, further strengthening these relationships.
Additionally, our channel checks suggest that Arista has vastly superior customer support relative to peers, even when dealing with smaller network operators. One source noted consistently receiving callbacks from Arista in 90 seconds after submitting a service ticket with technical service representatives all having a high level of expertise.
Competitive moat outcomes and outlook
As a result of the Company’s advantage in software and hardware, Arista has become the market leader in high performance switches for datacenters. This leadership has developed due to a product portfolio that meets customer needs for best-in-class performance, high uptime, and lower up-front and operating costs. Essentially, the Company has carved out a niche for itself by focusing on traffic management within datacenters, while competitors have opted to develop solutions for the general networking market (Cisco), or for communication service providers (Juniper). In addition, the Company has been able to combine the benefits of an integrated networking operating system that pairs with proprietary hardware for class-leading performance without forcing customers into hardware and software lock-in for the rest of their networks in the way that peers typically do. Ultimately, this focus on the high-performance segment of the networking market has allowed the company to gain greater leverage on its R&D spend, enabling it to investment more aggressively in its software platform to enable superior performance relative to peers, while maintaining hardware leadership by focusing on the design of value-add components, rather than spreading R&D dollars across commodity components, or investing in physical manufacturing facilities.
In conclusion, we see Arista’s moat as stable or growing. We believe their focus on open architectures is resonating with datacenter customers that prefer flexibility over walled garden approaches by Cisco, Juniper, and Nvidia, and investments in software will maintain this lead. Additionally, our view is that Arista remains a leader in high-performance hardware, as they will be the first to market with 800G ethernet, are actively developing 1,600G for release in the medium term, and are working with customers to displace Nvidia’s InfiniBand networking solution as the de facto option for intra-datacenter networking.
Strategy
Within the networking industry, we see three trends that are currently impacting Arista and its peers: 1) the transition from traditional network architectures to Hyper-Converged Infrastructure and AI datacenters, 2) the entry of Nvidia into the networking segment with its NVlink and InfiniBand offerings, and 3) in-sourcing of datacenter costs by hyperscalers.
1. Transition from traditional network architectures to HCI and AI datacenters
To understand current network architecture trends, its important to begin with how networks originally were designed as this view provides a glimpse in how other competitors think about their businesses and are competitively positioned. Originally, network architectures were hierarchical with three distinct layers (Core, Distribution, and Access), siloed server resources into separate compute and storage assets due to technology limitations, had high latency as a result of network hardware constraints, and required a large degree of manual configuration.
As a result of advancements in storage speeds, networking equipment, and software, previously siloed server resources were increasingly integrated together, removing barriers between server compute components while also flattening architectures to improved network efficiency. A visual summary of these developments is provided below[1].
Concurrent with the transformation of network hardware stacks has been the evolution of networks to incorporate multiple heterogeneous compute resources and access points. Whereas previously, most networks had a primary access point, increasingly networks have multiple shared compute resources such as cloud, internal servers, edge networks, and datacenters. As a result, network architectures that made it straightforward to acquire and deploy network equipment with locked in software (Cisco) have evolved to networks that increasingly must be able to integrate individual end points running a diverse menu of hardware and software vendors that are optimized for each use case. In other words, the increasing complexity of networks has decreased the effectiveness of a one-size-fits-all approach.
At the same time, the emergence of datacenters, and more recently AI clusters, has also altered the landscape of network architectures. From an end user perspective, in traditional and HCI architectures, a large portion of hardware functioning is devoted to routing: directing traffic on an inter-network basis, or across various end points. However, at the datacenter operator level, there is a higher level of demand for switching activity: managing traffic within a network. In simplified terms, each datacenter operates as a single network to the outside world, so overall data center performance is maximized by efficient switching across the pooled compute, memory, and storage resources.
Putting this all together, we believe that current trends in the networking market suggest sustained and growing demand for high-performance switching products as well as the need for flexible software that provides compatibility with a wide variety of other hardware and software vendors. This more surgical approach to designing, building, and operating networks is challenging the business models of players that have relied on tightly integrated software and hardware platforms with low flexibility.
2. (Re)Emergence of InfiniBand and NVlink in the networking industry
InfiniBand is a networking architecture that is an alternative to ethernet. InfiniBand began as a response to ethernet limitations in the late 1990s, and has re-emerged for similar reasons, especially after Nvidia’s purchase of Mellanox in 2019. InfiniBand’s primary benefit is lower latency and higher throughput relative to Ethernet, enabled by a specialized transport layer that eliminates the need for switches and routers, offers device-to-device connections, and the ability to transfer data without CPU involvement through RDMA (Remote Direct Memory Access).
Nvidia’s acquisition of the InfiniBand capability was motived by a desire to optimize the performance of their GPUs at the datacenter level. Nvidia also leveraged its InfiniBand capabilities to maximize performance of the GPUs at the server rack level, resulting in the NVlink capability deployed in their GB200 product with 72 GPUs per rack. While InfiniBand offers clear performance benefits, it has been used by Nvidia to reinforce their position in GPUs by capturing a greater position of the datacenter value chain (compute and networking, rather than compute only), but at the expense of hardware flexibility. In other words, Nvidia’s deployment of InfiniBand technology is motivated by a desire to force customers to choose their GPUs because of the performance benefit, driving pricing power in both segments.
However, InfiniBand does have distinct drawbacks relative to ethernet. On hardware cost alone, InfiniBand is anywhere from 2-2.5x more expensive than ethernet switches. InfiniBand also costs more to install and operate given its specialized nature and consumes more power except in specific use cases.
We believe the early lead in performance of InfiniBand over ethernet has narrowed over the past year. Recent advancements in ethernet have closed the latency and throughput gaps, such that ethernet is now good enough in most datacenter use cases. With technology that is now able to compete with InfiniBand head-on, the far-reaching benefits of ethernet, including lower hardware costs, ease of installation and operations, and broader ecosystem will begin to factor more into buying decisions than the highly niche benefits of InfiniBand. As an analogy, InfiniBand today is like a Formula 1 car, whereas ethernet is a high-performance road car. Yes, the Formula 1 car is faster in very controlled environments, but it takes an entire team of engineers to operate, is very expense, and loses its benefits once it’s outside a racetrack. In contrast, the road car is faster than just about any other car on the road, yet anyone can get in and drive it, while having the flexibility to service it at a dealer network. We speculate that only a small subsection of datacenters operators will find use cases that justify InfiniBand, and most will be happy with the performance / cost ratio of ethernet.
3. In-sourcing of datacenter costs by hyperscalers
The vast majority of datacenter spending is concentrated in a handful of large companies: Microsoft, Amazon, Meta, Google, Apple, and Oracle, with datacenter CapEx estimated at over $300 billion in 2025. Of this spending, compute hardware represents 40-60% of total CapEx. Given the magnitude of this spending, as well as the resources that these companies have, we believe that the hyperscalers are highly incentivized to in-source as much of this CapEx spend as is reasonable, resulting in far reaching implications for the datacenter chip market; Arista’s CEO noted that Nvidia is capturing 80%+ of the GPU market currently, but expected this would fall to 50% as hyperscalers begin deploying their internally developed semiconductors in the next several years.
With hyperscalers increasingly focused on developing and deploying their own chips, we believe this opens the door for Arista to increase its value-add positioning as a networking equipment supplier. Whereas datacenter developers are currently deploying a highly integrated Nvidia GPU/InfiniBand solution, these operators will increasingly look to their networking partners for solutions that optimize datacenter performance across separate compute and network vendors. Additionally, networking hardware constitutes a much smaller proportion of total datacenter up-front investment (10-25%). We believe this data point substantially reduces the urgency for hyperscalers to insource networking hardware, particularly in a sector that they lack a core competency in.
However, we do acknowledge that Arista is exposed to the threat of increasing adoption of white-box networking hardware by hyperscalers that are willing to invest in their own network software development efforts. At this time, we believe the increased performance that Arista offers through its software platform is enough to hedge this risk for the time being but will keep an eye out for indications that hyperscalers are beginning to focus their efforts in this area to reduce costs.
Financial Performance and Benchmarking
In the context of the networking industry, Arista has produced category-leading growth over the short, medium, and long-term. We summarize LTM revenues and growth rates for the sector in the following chart.
As we can see, Arista has substantially outperformed pureplay networking peers such as Cisco, Juniper, and Extreme Networks. Broadcom, although not a direct competitor to Arista, has been the only player in the networking segment able to match Arista’s growth rate. However, Broadcom’s growth has benefited from a number of acquisitions (8 over the past decade), including the VMWare acquisition that closed in late 2023. This acquisition alone added 300 bps to Broadcom’s 10-year revenue growth rate. To supplement our financial analysis, we have provided a comparison of Arista’s key financial metrics to public comparables below.
In gross margin terms, Arista is roughly equivalent to Cisco, below Broadcom, and then above Juniper, Extreme, HP Enterprise and Dell. While Broadcom earns better gross margins due to its focus on manufacturing high value add components, this advantage requires more operating and capital investment to sustain it. Juniper and Extreme, while most closely comparable to Arista, have lower margins as a result of their focus on service provider and campus networks, which appear to have less pricing power and/or lower value-add. HP Enterprise and Dell gross margins are weighed down by their server and white box networking hardware businesses that are driven by volume and scale rather than pricing power.
When we look at EBIT margins in the space, we see Arista as a clear leader with an EBIT margin of 42%; Broadcom is the closest at 30%, then Cisco at 22%. We believe this reflects Arista’s focused hardware R&D investments as well as running a higher leverage sales force as a result of a smaller product portfolio, and less overhead associated with running various manufacturing sites. In terms of returns on capital, Arista also leads the space with 29%, while Dell is the closest at 26%, as a result of strong capital returns from their mature business. We also note that Broadcom’s pursuit of growth through M&A negatively impacts its ROIC, which is only 11%. For context, since 2011, Broadcom has spent a cumulative $75 billion on M&A, or 28.3% of revenue over that period. During that time, Arista invested less than $500 million on inorganic growth, representing 1.5% of cumulative revenue. We are impressed with Arista’s ability to drive market leading growth with a disciplined approach to capital allocation and avoidance of inorganic levers relative to other players
In terms of cash flow, Arista’s asset-light business results in strong conversion of operating income into free cash flow. Typically, working capital is a source of cash due to a deferred revenue balance that tracks revenue growth as customers prepay for equipment or sign-up for ANET’s software services. At times, the Company’s reliance on third-party suppliers can negatively impact cash flows as Arista must invest in component inventory to maintain lead times, a situation the Company encountered in 2022. CapEx is generally less than 1% of revenues.
Arista has no debt, and held cash and short-term investments of $8.3 billion as of December 31, 2024.
Valuation & Expected Returns
To arrive at out our valuation and expected returns, we started by forecasting the business in line with the medium-term targets for 2025 and 2026 that management provided in their Q3 2024 investor presentation. We note that management typically provides conservative guidance; actual revenue growth in 2024 was 20% compared to guidance of 10-12%, and updated their guidance for 2025 to be 17%, at the top end of the range for prior guidance of 15-17%. From there, we forecasted out to 2029 to arrive at various five-year growth rates based on each case. Key highlights of each case are as follows.
1. Base Case
We assume mid-teens revenue growth rates until 2026, with gross margins and EBIT margins contracting in line with management’s outlook. From 2026, we forecast an acceleration of growth resulting in a compound annual revenue growth rate of 18%. We see a minor gross margin expansion due to positive mix shift as the Company launches 1,600GB ethernet products in 2027 and beyond, while EBIT margins realize scale benefits after 2026. We note that cash R&D expense is 12% of revenues in 2029.
2. Weak Case
We assume slower growth of 14% across the forecast period as competitors gain share relative to Arista. Additionally, we forecast 180 bps of gross margin deterioration in 2029 relative to 2024 due to pricing pressure, and a terminal EBIT margin of 41.5% (vs. 47.1% in 2024) as the company invests aggressively in R&D and sales and marketing to maintain its advantage in technology and customer support.
3. Best Case
Our long-term growth rate is projected at 20.5% on a compound annual basis until 2029. We forecast that gross margins improve by 130 bps due to pricing power in hardware and service offerings. We see additional leverage in EBIT margins from elevated 2024 levels, ultimately reaching 48% in 2029.
We summarize the key outputs of our forecast in the following table and compare them with actual data for FY2024 period below.
At Arista’s current share price of ~$90/share, we calculate the Company to be trading at a discount to fair value of 21% in the base case scenario when assuming an exit multiple of 35x EBIT. We note that Arista has traded between a multiple of 15-55x but is consistently valued between 30-40x. Our estimate of long-term return is 19.6%, or 2.4x cash-on-cash in 2029. A sensitivity of expected return and forecast scenario is provided in the following table.
The range of expected outcomes suggests a 30% loss of principal from current levels in the weak case scenario and a 10x EBIT multiple, although the company only traded below 15x EBIT for a very brief period during the middle of the COVID pandemic bear market. Realistically, we could expect a 4% compound return to 2029 as a worst case scenario and have confidence in an outcome between the base and best case scenarios given management’s penchant for conservatism, leading to operating results that would justify a premium multiple between 30 and 40x.
We provide the key components that drive our expected return in the following graph. Of note is the minimal amount of return that we expect from multiple expansion.
Management and Execution Capabilities
At a high level, we characterize Arista’s management team as successful operators with highly relevant industry experience that have material shareholder interests in the Company, and are compensated in ways that are strongly levered to financial performance and positive shareholder outcomes.
In our view, the key members of the management team are Jayshree Uppal (CEO), Kenneth Duda (co-founder and CTO), and Andy Bechtolsheim (co-founder and chief architect). Combined, these three senior leaders hold a 17.8% interest in Arista, with a market value of $21.6 billion, the majority of which is held by Andy (14.5%). In recent history, we saw no share dispositions for any of these leaders, except for a 2.5% reduction in Kenneth’s stake. Other senior executives have personal stakes valued between $2-28 million. The vast majority of senior leaders at the Company were senior executives at Cisco with tenures in excess of ten years prior to joining Arista, demonstrating strong industry knowledge.
Management’s pay is structured to favor shareholder interests. The Company avoids base salary increases, and reports setting base salaries for executives at levels that are in the lowest 25th percentile of the market for peers. At risk pay is 95.6% of total pay, of which 92.6% is performance- or tenure-based equity compensation, while the remainder is short-term cash bonus incentives.
Over the past three years, management has achieved an average performance of 97% of target for its short-term incentives. For equity-based incentives, the management team has achieved five of eight objectives over the same period. Interestingly, we note that the company missed its revenue targets in 2022 and 2023 (by 1% or less) but still substantially exceeded the 2-year revenue CAGR established for PRSUs issued in 2022 (26% vs. 15% target). This suggests to us that the Board is setting fairly aggressive stretch targets for the management team on a yearly basis. We note that management has exceeded its operating income target over the past three years (ending in 2023; the Company’s proxy statement for 2024 results is not available at the time of writing), despite modestly missing revenue goals.
We do however acknowledge that Arista is exposed to succession risk at the senior level. The CEO is currently 64, and has worked for Arista since 2008, while the Chief Platform Officer and SVP of Software Engineering, John McCool is 65 and has been at Arista since 2017. We see both positions as key leadership roles with a material impact on the Company’s success, so will be looking for updates on transition timelines and prospective candidates in regards to both roles.
Governance and Shareholder Alignment
Arista’s board of directors has eight seats, of which two are non-independent seats held by Jayshree (CEO and Chair), and Kenneth Duda (co-founder and CTO). What stands out to us in Arista’s board is the material size of some board member’s stakes despite not being founders or representing institutional shareholders. For example, Charles Giancarlo, a board member since 2013 holds a 0.03% stake in Arista valued at $36 million. This sizeable interest is held despite it being far in excess of the minimum holding requirement of three times a board members annual cash retainer. To us, this suggests a board that is highly incentivized to drive towards objectives that benefit shareholders and is bullish on the Company’s prospects. In total, insiders (board members and management) hold close to 18% of the Company’s shares which equates to roughly $21.7 billion in value. We note that only one board member (Charles) has sold down a portion of their shares (4%) over the past year.
In terms of compensation, board members are paid roughly 75% in stock, and the remainder in cash. Board members bring experience from firms such as Dolby Labs, National Semiconductor, Cadence Design, Pure Storage, Silver Lake Partners, Cisco, Zoom, Digital Ocean, Citrix, and Forrester Research, contributing expertise across semiconductors, networking, data storage, cloud computing, virtualization, private equity and technology market research. In short, we believe the board is highly qualified to provide value-add guidance to Arista’s development.
Further Research
To complete our research, we would like to conduct expert calls with customers across various datacenter opportunities to better understand Arista’s relative position in software and hardware relative to peers. We would also like to probe further on the expected use cases of InfiniBand networking and ethernet and what it means for the relative adoption of Arista’s products going forward and confirm or refute our current conclusions, as well as understand the TCO math of Arista products and software, and relative ease of deploying Arista’s EOS platform. Beyond datacenters, we would like to conduct calls with campus network operators to understand how Arista is positioned, given this is a market that Arista is expanding into, and has not historically been a core competency. Finally, we would like to speak to sales representatives at Cisco, Juniper, and Extreme to understand how successful Arista is at competing against their offerings.
[1] https://www.wallarm.com/what/what-is-converged-infrastructure-guide-by-wallarm