DCT

5:25-cv-03738

Netflix Inc v. Broadcom Inc

Log In
Key Events
Amended Complaint

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 3:25-cv-03738, N.D. Cal., 11/17/2025
  • Venue Allegations: Venue is alleged to be proper in the Northern District of California because both Broadcom and VMware maintain principal places of business in the district and have allegedly committed acts of infringement there.
  • Core Dispute: Plaintiff alleges that Defendants’ virtualization, cloud infrastructure, and ethernet switching products infringe five U.S. patents related to virtual machine service availability, network subnet provisioning, and high-precision time synchronization.
  • Technical Context: The technologies at issue are foundational to modern cloud computing and data center operations, governing how virtual resources are managed for reliability and how time is synchronized across high-speed network components.
  • Key Procedural History: The complaint notes that Broadcom acquired VMware in a series of transactions concluding in November 2023. It also states that in a prior order dated October 29, 2025, the Court found claims of all five asserted patents to be directed to patent-ineligible subject matter; Plaintiff re-pleads these claims to avoid waiver while anticipating a stipulation to dismiss and appeal the ruling. The complaint further alleges a "culture of copying" at VMware, citing previous litigation where juries found VMware's infringement to be willful.

Case Timeline

Date Event
2003-03-17 ’102 Patent Priority Date
2005-04-27 ’912 Patent Priority Date
2006-04-11 ’931 Patent Priority Date
2007-12-25 ’102 Patent Issue Date
2008-09-25 ’751 Patent Priority Date
2008-11-04 ’931 Patent Issue Date
2010-01-19 ’912 Patent Issue Date
2010-02-02 ’751 Patent Issue Date
2014-08-29 ’472 Patent Priority Date
2019-06-25 ’472 Patent Issue Date
2022-05-26 Broadcom and VMware Inc. enter Merger Agreement
2023-11-22 Broadcom completes acquisition of VMware Inc.
2024-12-23 Netflix sends first notice letter to Defendants
2025-04-15 Netflix sends second notice letter to Defendants
2025-10-29 Court issues order on patent ineligibility for asserted claims
2025-11-17 Second Amended Complaint Filing Date

II. Technology and Patent(s)-in-Suit Analysis

U.S. Patent No. 10,331,472 - "Virtual Machine Service Availability"

The Invention Explained

  • Problem Addressed: In virtualized network environments, ensuring high availability for critical services often relied on redundancy, such as synchronous replication. This approach could negatively affect performance and cost efficiency, and a failure of redundant servers could still lead to service interruptions or periods of lower availability, which is particularly "undesirable for real-time or critical services" (’472 Patent, col. 1:44-61).
  • The Patented Solution: The invention describes a method for dynamically managing service availability in a server cluster. A service availability controller monitors services running on virtual machines. If the availability of a high-priority service is reduced, the system creates capacity by deactivating an instance of a lower-priority service on an available server. The higher-priority service is then activated on that newly available server, thereby improving "system reliability without increased hardware costs or performance impact" (’472 Patent, col. 1:65-2:8).
  • Technical Importance: This approach provided a more flexible and cost-effective way to maintain high availability for critical applications in virtualized data centers compared to relying solely on static, hardware-based redundancy (’472 Patent, col. 1:62-63).

Key Claims at a Glance

  • The complaint asserts independent claim 6 and dependent claims 7-10 (Compl. ¶100).
  • Essential elements of independent claim 6 include:
    • monitoring a first availability of a first service, the first service having a first availability requirement and a first availability tolerance;
    • detecting a reduction in the first availability of the first service;
    • creating capacity for the first service by deactivating a second service on a first active virtual machine on a server, the second service having specified availability characteristics relative to the first service; and
    • activating a second active virtual machine executing the first service on the server.
  • The complaint reserves the right to assert additional claims (Compl. ¶119, fn. 119).

U.S. Patent No. 7,313,102 - "System and Method for Subnet Configuration and Selection"

The Invention Explained

  • Problem Addressed: Before the invention, tools for managing IP address spaces ("subnets") were separate from automated provisioning systems. This required network administrators to manually determine available subnets that met a consumer's requirements for performance and security, a process described as tedious, error-prone, and hindering efficient resource allocation (’102 Patent, col. 1:29-48).
  • The Patented Solution: The patent discloses a method and system for provisioning subnets by first grouping them based on "logical properties" like security or performance characteristics (’102 Patent, col. 3:35-47). These groups are then assigned to specific network consumers. The invention provides for a graphical user interface (GUI) that enables "constrained selection," guiding the consumer through a step-by-step process of selecting a public or private address space, a gateway device, a subnet group, and a subnet mask to provision a suitable subnet without direct intervention from a network manager (’102 Patent, col. 1:55-61).
  • Technical Importance: The invention aimed to improve the accuracy, reliability, and security of network provisioning by automating and simplifying the process, thereby reducing administrative overhead and potential for human error (’102 Patent, col. 1:29-2:13).

Key Claims at a Glance

  • The complaint asserts independent claim 1 and dependent claims 2-11 (Compl. ¶151).
  • Essential elements of independent claim 1 include:
    • grouping the subnets into subnet groups based on logical properties of the subnets;
    • assigning to each network consumer those subnet groups that are accessible to that network consumer; and
    • providing for constrained selection of a particular subnet by a network consumer accomplished by way of a graphical user interface with selectable fields, wherein the constrained selection includes selecting (i) an address space type, (ii) a gateway device, (iii) a subnet group, and (iv) a subnet mask.
  • The complaint reserves the right to assert additional claims (Compl. ¶150, fn. 144).

U.S. Patent No. 7,649,912 - "Time Synchronization, Deterministic Data Delivery and Redundancy for Cascaded Nodes on Full Duplex Ethernet Networks"

  • Technology Synopsis: The patent addresses the difficulty of maintaining precise time synchronization (per the IEEE 1588 standard) in networks that use "store and forward" switches, which can introduce random delays (Compl. ¶47; ’912 Patent, col. 2:19-25). The invention provides a method for use in daisy-chained networks where each node adjusts the timestamp in a time synchronization message "on the fly" to account for its own internal delay, and also adjusts the associated checksum and CRC code, before forwarding the message (Compl. ¶48-49; ’912 Patent, col. 9:44-10:12).
  • Asserted Claims: Claims 1-3 and 5-12, including independent claims 1 and 7 (Compl. ¶193, ¶51).
  • Accused Features: Broadcom’s ethernet switching products and associated software/firmware (e.g., StrataDNX devices, BroadPTP 1588 Software Suite, BroadSync firmware) that implement the PTPv2 specification for time synchronization (Compl. ¶193).

U.S. Patent No. 7,447,931 - "Step time change compensation in an industrial automation network"

  • Technology Synopsis: This patent addresses the "step change problem," where conventional time synchronization protocols (like IEEE 1588) fail to account for sudden, discrete jumps in a master clock's time, leading to system-wide inaccuracies (Compl. ¶64; ’931 Patent, col. 2:10-21). The invention proposes a method where a single node receives timestamps with associated offsets, determines if a step change has occurred (e.g., by comparing offsets), calculates a compensated timestamp, and selectively updates its local timestamp and offset if a step change is detected (Compl. ¶72).
  • Asserted Claims: Claims 27-32, including independent claim 27 (Compl. ¶241).
  • Accused Features: Broadcom’s Switching Solutions that implement the PTPv2 specification, particularly their hardware and software for processing timestamps and compensating for clock adjustments (Compl. ¶241).

U.S. Patent No. 7,656,751 - "Step time change compensation in an industrial automation network"

  • Technology Synopsis: Sharing a common specification with the ’931 Patent, this patent also addresses the "step change problem" in time synchronization (Compl. ¶78). It claims a system, rather than a method, comprising a "timestamp component" that captures timestamps and offsets from a network node, and a "time synch component" that identifies step changes to a master clock and synchronizes the local clock of the network node with the identified change (Compl. ¶86; ’751 Patent, Claim 1).
  • Asserted Claims: Claims 1-14, including independent claim 1 (Compl. ¶306).
  • Accused Features: The same family of Broadcom Switching Solutions as those accused of infringing the ’931 Patent (Compl. ¶306).

III. The Accused Instrumentality

Product Identification

  • For the ’472 Patent: The "Broadcom Load Balancing Accused Products," which include VMware Cloud Foundation, VMware Cloud on AWS, and other cloud solutions incorporating VMware NSX/NSX-T Data Center and/or VMware Avi Load Balancer (Compl. ¶100).
  • For the ’102 Patent: The "Broadcom Subnet Provisioning Accused Products," comprising largely the same family of VMware cloud and virtualization products, specifically those incorporating VMware NSX/NSX-T Data Center (Compl. ¶151).

Functionality and Market Context

  • The accused functionalities are part of a suite of software-defined networking and cloud infrastructure management tools. For the ’472 Patent, the relevant feature is the automatic scaling and migration of "virtual services" across "Service Engines" (data plane virtual machines) (Compl. ¶104). The product documentation describes this functionality as "autoscale," which responds to performance metrics like CPU utilization to maintain service availability (Compl. ¶107-108). This is a core capability for modern, elastic cloud services. A screenshot from the product's user interface shows a "health" indicator for monitored "virtual services," which is alleged to be part of the infringing method (Compl. ¶103, Fig. 4).
  • For the ’102 Patent, the relevant feature is the provisioning of virtual networks within VMware's NSX platform. The complaint alleges that users configure network "segments" or "subnets" within a Virtual Private Cloud (VPC) using a GUI that provides for selection of logical groupings ("tier-0" and "tier-1" subnets) and other properties like "access mode" (public, private, or isolated) (Compl. ¶154, ¶160). A table from the NSX Administrator Guide illustrates the concept of different subnet groups (Compl. ¶154, Fig. 17). This functionality is central to creating isolated, multi-tenant network environments in the cloud.
  • The complaint alleges these product families are commercially significant, noting that VMware Cloud Foundation represented $2.5 billion in revenue for Broadcom in one fiscal quarter (Compl. ¶119).

IV. Analysis of Infringement Allegations

’472 Patent Infringement Allegations

Claim Element (from Independent Claim 6) Alleged Infringing Functionality Complaint Citation Patent Citation
monitoring a first availability of a first service... The Accused Products monitor "virtual services" and display their status via a numeric, color-coded "health" indicator in the user interface. ¶103 col. 9:32-35
detecting a reduction in the first availability of the first service The Accused Products use metric-based thresholds, such as detecting when a Service Engine's (SE) CPU usage exceeds an 80% average, to trigger a scale-out or migration operation. A flowchart from product documentation shows this decision point (Compl. Fig. 7). ¶107-108 col. 9:36-37
creating capacity for the first service by deactivating a second service on a first active virtual machine on a server... In a "migrate" scenario, the system moves a virtual service to a new SE. The product documentation states that after 30 seconds, the "old SE will terminate the remaining connections and be removed from the virtual service configuration," which is alleged to be the deactivation step. ¶109-110 col. 9:38-44
activating a second active virtual machine executing the first service on the server In a "scale out" scenario, the system adds one or more new SEs (virtual machines) to a virtual service to handle increased traffic, thereby activating a new VM to execute the service. ¶112-113 col. 9:44-45
  • Identified Points of Contention:
    • Scope Questions: A potential point of contention may be whether migrating a service and subsequently removing the original virtual machine (the "old SE") from the service configuration meets the claim limitation of "deactivating a second service on a first active virtual machine." The complaint's theory appears to equate the migration of a single service between two VMs with a process involving two distinct services as recited in the claim, which raises a question of definitional scope.
    • Technical Questions: The complaint maps the "deactivating" limitation to a "migrate" workflow and the "activating" limitation to a separate "scale out" workflow (Compl. ¶109, ¶112). This raises the question of whether the accused products perform these steps in the specific, ordered combination required by Claim 6 in response to a single detected event, or if "migrate" and "scale out" are alternative, mutually exclusive responses to different underlying conditions.

’102 Patent Infringement Allegations

Claim Element (from Independent Claim 1) Alleged Infringing Functionality Complaint Citation Patent Citation
grouping the subnets into subnet groups based on logical properties of the subnets The accused NSX products provide for selecting from logical subnet groups, such as "tier-0" and "tier-1" subnets, which are based on the properties of the subnets. ¶153-154 col. 9:15-17
assigning to each network consumer those subnet groups that are accessible to that network consumer The NSX Administration Guide explains that users "can add subnets (networks) inside the NSX VPC that is assigned to them," which allegedly constitutes assigning accessible groups. ¶156-157 col. 9:18-20
providing for constrained selection of a particular subnet by a network consumer accomplished by way of a graphical user interface... The accused products provide a GUI for adding a subnet where a user can specify properties such as name, access mode (public/private), IP assignment, size, and IP CIDR. A screenshot from the NSX Administrator Guide shows this interface (Compl. Fig. 20). ¶158-160 col. 9:21-29
...wherein the constrained selection includes (i) selecting a public or private type address space... (ii) ...selecting a gateway device..., and (iii) selecting a subnet group..., and (iv) selecting a subnet mask... The GUI allegedly allows selection of address space type via "Access Mode," a subnet group via selection of tier-0/tier-1 subnets, and a subnet mask via a size or CIDR field. Gateway IP specification is also alleged to be supported (Compl. Fig. 21). ¶158-160 col. 9:48-61
  • Identified Points of Contention:
    • Scope Questions: The infringement analysis may turn on whether the accused product's workflow for creating a network segment constitutes the specific four-part "constrained selection" method recited in Claim 1. For example, does selecting from "tier-0" or "tier-1" sources function as "selecting a subnet group from those subnet groups that are accessible to the network consumer" as contemplated by the patent?
    • Technical Questions: What evidence does the complaint provide that the accused GUI requires the user to perform all four discrete selection steps—address type, gateway, subnet group, and mask—as part of the claimed method? The complaint presents different screenshots for different steps, raising the question of whether they are part of a single, integrated, and constrained workflow as required by the claim language.

V. Key Claim Terms for Construction

’472 Patent, Claim 6

  • The Term: "deactivating a second service"
  • Context and Importance: This term is critical because the plaintiff's infringement theory equates it with the "migrate" function in the accused products, where an existing virtual machine instance is removed from a service's configuration after the service moves to a new instance (Compl. ¶110). Practitioners may focus on whether "deactivating a service" requires stopping a distinct software service, as opposed to decommissioning the virtual machine on which it was running.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The specification states, "an instance of a lower priority service is deactivated to provide an available server" (’472 Patent, col. 2:1-3). This language focuses on the outcome—making a server available—which may support an interpretation where any action that frees up server resources, including migration, constitutes "deactivation."
    • Evidence for a Narrower Interpretation: The claim language recites "deactivating a second service on a first active virtual machine" to create capacity for a "first service" (’472 Patent, col. 9:38-44). This suggests two distinct services are involved, whereas the accused "migrate" function appears to involve only one service moving between two virtual machines. This may support a narrower construction requiring two different services to be part of the process.

’102 Patent, Claim 1

  • The Term: "constrained selection"
  • Context and Importance: The complaint highlights that during prosecution, the "providing for constrained selection" limitation was key to allowance (Compl. ¶36). The infringement case hinges on whether the accused NSX interface performs the specific, multi-part selection process defined by this term.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The patent's background discusses overcoming the "tedious tasks" of manual subnet determination (’102 Patent, col. 1:35-41). This purpose might support a broader reading where any GUI that guides a user through a simplified, logical provisioning process meets the "constrained" criteria.
    • Evidence for a Narrower Interpretation: The claim itself provides an explicit definition: "wherein the constrained selection includes (i) selecting a public or private type address space, (ii) if applicable, selecting a gateway device..., and (iii) selecting a subnet group..., and (iv) selecting a subnet mask..." (’102 Patent, col. 9:48-61). This explicit recitation of four required sub-steps provides strong evidence for a narrow construction that requires each element to be present in the accused method.

VI. Other Allegations

  • Indirect Infringement: For both the ’472 and ’102 patents, inducement is alleged based on Defendants' creation and distribution of marketing materials, installation guides, and instructional "VMware Hands-on Labs" that allegedly encourage and instruct customers on how to use the accused features in an infringing manner (Compl. ¶126, ¶175). Contributory infringement is also alleged, on the basis that the accused products contain components specially made to practice the claimed methods and have no substantial non-infringing use, particularly when features like "auto-rebalance" or subnet provisioning are used (Compl. ¶138, ¶183).
  • Willful Infringement: Willfulness is alleged for all asserted patents. The complaint alleges Defendants had pre-suit knowledge based on notice letters sent by Netflix in December 2024 and April 2025 (Compl. ¶121, ¶168). It further alleges a "culture of copying" and "willful blindness toward patents" at VMware, citing jury findings of willful infringement against VMware in prior, separate litigation (Compl. ¶146-148).

VII. Analyst’s Conclusion: Key Questions for the Case

  • A primary issue will be one of claim scope versus technical implementation: For the ’472 patent, can the act of migrating a virtual service and decommissioning the original host VM be construed to meet the claim language "deactivating a second service on a first active virtual machine," or is there a fundamental mismatch between the patent's two-service concept and the accused product's single-service migration function?
  • A second core issue will be one of evidentiary proof: For the ’102 patent, does the accused VMware NSX interface, in practice, implement the specific, four-part "constrained selection" method that was critical for the patent's allowance? The case may depend on whether Plaintiff can demonstrate that the accused product's GUI performs each of the recited selection steps as part of an integrated, claimed method.
  • A significant threshold question will be the impact of prior patent eligibility rulings: The complaint's explicit acknowledgment that a court has already found claims of all five asserted patents to be directed to ineligible subject matter under 35 U.S.C. § 101 frames the entire dispute. How the parties and the court navigate this procedural history will be a critical factor in the case's trajectory.