Broadcom emerged as one of the clearest beneficiaries of artificial intelligence spending outside Nvidia, raising its forecast for fiscal 2024 AI-related chip revenue from $10 billion to $11 billion. Its shares had risen sharply as investors recognised the role of custom accelerators and networking components in hyperscale data centres. The company also brought a large software business into the portfolio through its $61 billion VMware acquisition. Together, those moves positioned Broadcom as a different kind of AI infrastructure supplier: less focused on selling general-purpose processors to every customer and more focused on designing specialised silicon for a small number of very large cloud platforms.
Custom AI Chips Open an Alternative to Nvidia GPUs
Nvidia’s graphics processors dominated AI training because they offered exceptional performance and a broad software ecosystem. Broadcom’s opportunity lay in application-specific integrated circuits, or ASICs, designed around the stable, high-volume workloads of hyperscale customers. A custom chip requires substantial upfront investment and lacks the flexibility of a general-purpose GPU, but it can lower capital and energy costs once deployed at enormous scale. For companies operating data centres across the world, even modest efficiency gains can justify a bespoke design.
Broadcom had already worked with large technology groups including Google and Meta and disclosed that a third unidentified customer was using its custom AI chips. Those relationships gave the strategy credibility because hyperscalers have the workloads, engineering teams and purchasing volume needed to support custom silicon. The market was not necessarily a direct replacement for every Nvidia product. Instead, it created a route for the largest buyers to reduce dependence on one supplier and optimise selected workloads that were predictable enough to reward specialisation.
AI Revenue Growth Changes Broadcom’s Market Story
The increase to an $11 billion forecast demonstrated how rapidly generative AI investment was flowing into infrastructure. Broadcom’s share price rose as analysts treated the company as one of the strongest AI opportunities beyond Nvidia. This enthusiasm reflected more than a single product cycle. Custom accelerators, high-speed networking and connectivity are all required to build large clusters, so Broadcom could benefit as data-centre operators expanded even when it did not supply the primary compute processor.
Investor optimism was reinforced by a ten-for-one stock split, a step that followed a similar move by Nvidia. A split does not change the underlying value of a company, but it can make individual shares more accessible and signal management’s confidence after a strong run. The market reaction illustrated how closely AI expectations had become linked with semiconductor valuations. That makes execution critical: forecasts based on a small number of large customers can produce substantial growth, but they also create concentration risk if deployment plans change.
VMware Adds Software and Diversifies Revenue
Broadcom’s acquisition of VMware added $2.7 billion to second-quarter software revenue and gave the company a major position in enterprise virtualisation and cloud infrastructure. Software diversification can moderate the cyclicality of semiconductor demand while creating additional relationships with large corporate customers. It also changes how Broadcom is evaluated. The company is no longer only a collection of chip businesses; it combines specialised hardware with infrastructure software that sits inside many enterprise data centres.
The strategic fit with AI is indirect but meaningful. Organisations adopting AI need to manage computing resources, private clouds and complex hybrid environments. VMware’s products can participate in that layer even when the model itself runs on hardware from another supplier. Integration still presents challenges, and customers will judge changes in licensing, support and product direction carefully. The acquisition adds scale and recurring revenue, but its long-term value depends on Broadcom maintaining the trust of enterprises that rely on the software.
Competition Broadens Without Ending Nvidia’s Lead
Broadcom’s growth did not mean Nvidia’s dominance had disappeared. Nvidia retained a formidable advantage in general-purpose accelerators, developer tooling and the ability to serve a wide range of customers. Broadcom was competing through a narrower model that suited a handful of hyperscalers. The two approaches can coexist: cloud companies may use Nvidia systems for flexible workloads while shifting mature, repeatable tasks to custom ASICs. The result could be a more diverse market rather than a simple winner replacing the incumbent.
That diversity matters for cost, supply and innovation. Large customers gain bargaining power and technical options, while Nvidia faces pressure to continue improving performance and efficiency. Broadcom’s forecast showed that the AI infrastructure boom was already distributing value beyond the best-known GPU supplier. Its lasting challenge is to convert a surge in spending into repeatable design wins and dependable software growth. If it succeeds, the AI chip market may remain led by Nvidia while becoming far less synonymous with a single company.
The custom-silicon model also changes the balance of power between cloud providers and component suppliers. A hyperscaler that controls more of its chip design can tune hardware and software together and negotiate from a stronger position. That advantage is available only to companies with extraordinary scale, so it may widen the infrastructure gap between the largest clouds and smaller operators. Broadcom benefits by supplying the specialised expertise behind those designs, but it must serve customers that increasingly want differentiation of their own. Managing that tension will be central to the durability and profitability of its AI business through future investment and replacement cycles.
Customer diversification would make that growth more resilient if one hyperscaler slowed its spending plans.
