AVGO

The company that trades today as Broadcom is the product of a roll-up rather than a single founding. The original Broadcom Corporation was started in 1991 by Henry Samueli and Henry Nicholas as a communications chip designer, and it grew into a major supplier of networking and connectivity silicon. The current corporate entity, however, traces its operating DNA to Avago Technologies, a chip business that was spun out of Hewlett-Packard and Agilent and taken private before listing publicly. Avago, led by Hock Tan, acquired the classic Broadcom Corporation in 2016 for roughly thirty seven billion dollars and adopted the better-known Broadcom name. That deal was the pivot point. Before and after it, Tan executed a sequence of acquisitions that reshaped the company every few years. LSI Corporation, a storage and networking chip maker, was bought in 2014 for around six and a half billion dollars. Brocade, a fiber channel storage networking firm, was acquired in 2017 for close to six billion dollars. Then the strategy widened beyond silicon. CA Technologies, a legacy enterprise software vendor, was bought in 2018 for nearly nineteen billion dollars, and the enterprise security business of Symantec followed in 2019 for about ten and a half billion dollars. The largest move came in late 2023, when Broadcom completed its acquisition of VMware for roughly sixty nine billion dollars, the deal that turned a chip company into a software-heavy conglomerate.

Today the business is split cleanly into two segments. Semiconductor Solutions is the larger and more historically central of the two. It covers networking silicon, including the Ethernet switch and routing chips that move data inside data centers, custom compute accelerators designed for specific cloud customers, wireless connectivity components found in smartphones, broadband access chips for set-top boxes and gateways, and storage connectivity products. The single most important customer relationship inside this segment is a large American smartphone maker that buys radio frequency and connectivity components in high volume, a relationship that contributes a meaningful slice of semiconductor revenue and recurs on the annual phone cycle. Infrastructure Software is the segment created and then enlarged through the CA, Symantec, and VMware acquisitions. It sells virtualization, networking, storage, and cloud infrastructure software to large enterprises, with VMware as the centerpiece. VMware's products, including vSphere, NSX, vSAN, and the bundled VMware Cloud Foundation stack, run private and hybrid cloud environments inside banks, governments, and large corporations that cannot easily move those workloads elsewhere.

The economic engine rests on a specific theory of value that Tan has applied consistently. The company targets what it calls franchise businesses, meaning products with high switching costs, deep technical moats, and customers who cannot readily replace them. In semiconductors that means leadership positions in narrow but essential categories rather than commodity volume. Broadcom's Ethernet switching silicon, sold under the Tomahawk and Jericho families, holds a dominant share of the high-end merchant switch market, and its newest generation of switch chips operates at bandwidths that competitors have struggled to match. The custom silicon business is structured differently from selling a catalog part. Broadcom co-designs application specific chips, often called XPUs in the AI context, for individual hyperscale customers, handling the design, intellectual property, packaging, and manufacturing coordination while the customer owns the end product. That model produces large, multi-year, high-value engagements with a small number of buyers. On the software side the durability comes from a different source. Enterprise virtualization and mainframe management software is embedded so deeply in customer operations that ripping it out carries real operational risk, which gives Broadcom strong pricing power and high renewal rates. After acquiring VMware, the company moved it from a sprawling catalog of more than eight thousand individual products toward a smaller set of subscription bundles centered on VMware Cloud Foundation, a shift designed to convert one-time license sales into predictable recurring revenue. That transition raised prices for some customers and drew complaints, but it also lifted software segment growth into double digits.

Broadcom's competitive position differs sharply by segment. In networking silicon it competes with Nvidia's networking arm, with Marvell, and with the in-house efforts of large cloud operators, but its switch and routing chips remain a default choice for builders of large data center fabrics. In custom AI accelerators it occupies an unusual position relative to Nvidia. Where Nvidia sells general-purpose graphics processors to the broad market, Broadcom helps the largest buyers design their own bespoke chips, which lets those customers lower their cost of running AI at scale and reduce dependence on a single merchant supplier. Public reporting has identified Google, Meta Platforms, and more recently OpenAI and Anthropic among the customers pursuing custom silicon with Broadcom, and the company has spoken about a large multi-year backlog tied to these AI programs along with an ambition to reach roughly one hundred billion dollars in annual AI-related chip revenue later in the decade. In enterprise software VMware competes with public cloud platforms that want to pull workloads out of private data centers, and with open-source virtualization alternatives, but the installed base and the operational stickiness of those deployments give Broadcom a defensible position even as it raises prices.

Leadership is concentrated and consistent, which is central to understanding the company. Hock Tan has run the business since the Avago era and is the architect of the acquisition strategy, the cost discipline, and the focus on franchise products. His approach to integrating acquisitions is well known in the industry. He retains the high-margin core of an acquired business, divests or winds down the parts that do not fit, and drives the remainder hard for cash flow and margin rather than top-line growth at any cost. The board granted Tan an equity award in 2025 structured to keep him leading the company through fiscal 2030, signaling continuity at the top. Henry Samueli, a co-founder of the original Broadcom Corporation, serves as chairman of the board and remains the company's technical conscience as a longtime chief technical officer figure. The finance function is in transition. Kirsten Spears, the chief financial officer through this period, is retiring, and Amie Thuener is set to take over as CFO in mid-2026. The broader board includes a mix of semiconductor industry veterans and investors, among them Kenneth Hao of the private equity firm Silver Lake, which has been closely associated with several of Broadcom's largest deals.

The forward strategy is a continuation of what has worked, sharpened by the AI cycle. The near-term growth story is custom AI silicon and the networking that connects it. Broadcom is positioning itself as the company that builds the bespoke accelerators and the high-bandwidth switching fabric for the largest AI data centers, a role that benefits from heavy capital spending by cloud operators. The software strategy is to extract durable, recurring, high-margin revenue from VMware by completing the subscription transition and cross-selling the broader infrastructure software portfolio into VMware's large enterprise base. Underneath both is the same financial logic that has defined the company for a decade. Broadcom uses the strong cash flows of its franchise businesses to service the debt taken on for acquisitions, to fund a large and growing dividend, and to repurchase shares, while remaining open to the next large deal when one fits the franchise criteria. Investors evaluating the company are effectively underwriting both an AI infrastructure bet and a private equity style cash-compounding machine inside one corporate structure.

The risks are specific and follow directly from the strategy. The first is acquisition and integration risk. A company built by serial large deals carries substantial debt and depends on integrating very different businesses without destroying the value it paid for, and the VMware price increases have created customer friction that competitors are trying to exploit. The second is customer concentration. The semiconductor segment leans heavily on a single large smartphone customer, and the fast-growing custom AI business depends on a small number of hyperscale buyers whose own plans could change, be brought in-house, or slow with the AI capital spending cycle. The third is cyclicality and the AI cycle itself. Much of the recent growth narrative is tied to spending on AI data centers, and any pause or correction in that spending would weigh heavily on the part of the business the market currently prizes most. The fourth is regulatory and geopolitical exposure. Large technology acquisitions face antitrust scrutiny in multiple jurisdictions, and the semiconductor business is exposed to export controls and to the manufacturing concentration of the broader chip supply chain. A fifth, quieter risk is execution dependence on a single leader. Tan's record is the core of the investment thesis, and the company's identity is closely tied to his judgment on capital allocation, which makes eventual succession a genuine variable even with his term extended.

Taken together, Broadcom is best understood as two strong businesses welded onto a disciplined capital-allocation engine. One business sells some of the most essential and defensible silicon in modern networking and AI infrastructure, and the other sells software that large enterprises find difficult and risky to replace. The company has grown less by inventing new markets than by buying franchise positions, optimizing them, and using the resulting cash flows to do it again at larger scale. The durable facts are the moats in switching silicon and enterprise virtualization, the custom-chip relationships with the largest cloud operators, the consistent and acquisition-driven leadership under Hock Tan, and a balance sheet built to carry debt against reliable cash generation. The open questions are how long the AI capital cycle sustains the current growth, how durable the VMware pricing model proves as customers adjust, and how the company transitions when its defining executive eventually steps back. For an investor reading this page, AVGO represents a concentrated bet on infrastructure that sits underneath both the cloud and the AI buildout, run by a management team whose method is well documented and whose risks are equally clear.