Broadcom (AVGO) stock is in the spotlight as the semiconductor giant begins shipping its breakthrough Tomahawk Ultra Ethernet switch, designed for AI and high-performance computing workloads. The Tomahawk Ultra is its latest innovation for AI and high-performance computing, and analysts have been bullish ahead of its launch.
The new switch delivers industry-leading performance, featuring 250-nanosecond switch latency at full 51.2 terabits per second throughput, which addresses the extreme demands of tightly coupled AI clusters. Key innovations include ultra-low latency switching, adaptable Ethernet headers that reduce overhead from 46 bytes to just 10 bytes, and lossless fabric technology that eliminates packet drops.
“AI and HPC workloads are converging into tightly coupled accelerator clusters that demand supercomputer-class latency,” said Kunjan Sobhani, lead semiconductor analyst at Bloomberg Intelligence, highlighting the product’s potential in what’s projected to be a “double-digit billions” market within years.
The Tomahawk Ultra’s 100% pin-compatibility with its predecessor ensures rapid deployment, while features like In-Network Collectives reduce AI workload completion times by executing operations directly within the switch chip.
With AI infrastructure spending continuing to surge and Broadcom positioning itself at the center of this transformation, the launch of the Tomahawk Ultra could be a significant catalyst for the stock’s performance in the coming quarters.
In the second quarter of its fiscal 2025 (ended in April), Broadcom reported record revenue of $15 billion, up 20% year-over-year, driven by explosive growth in AI semiconductors and continued momentum from its VMware acquisition. The semiconductor giant's AI revenue reached $4.4 billion, a 46% increase year-over-year and marking the ninth consecutive quarter of strong performance.
CEO Hock Tan highlighted Broadcom’s dominant position in AI networking, which now accounts for 40% of AI revenue and has grown by over 170% year-over-year. “AI networking, which is based on Ethernet, was robust and remains the preferred choice by our hyperscale customers,” Tan said, emphasizing Ethernet’s role as the standard for both scale-out and scale-up deployments.
