Broadcom Shares Gain on Release of New Ethernet Switch for AI Workloads

Broadcom Inc. (NASDAQ: AVGO) announced the shipment of its breakthrough Tomahawk Ultra Ethernet switch on Tuesday, engineered specifically for high-performance computing and artificial intelligence workloads. The semiconductor giant’s stock gained 2.22% in premarket trading, reaching $281.72 as of 11:47 AM EDT, building on its impressive year-to-date performance of over 22%.

The new switch represents a significant technological advancement in AI infrastructure, delivering industry-leading ultra-low latency of 250 nanoseconds and massive throughput of 51.2 Tbps. This product launch comes as Broadcom continues to capitalize on the growing demand for AI-optimized networking solutions in data centers worldwide.

Tomahawk Ultra: Broadcom’s Effort at Redefining Ethernet for AI Scale-Up

The Tomahawk Ultra represents a multi-year engineering effort involving hundreds of engineers who reimagined every aspect of the Ethernet switch architecture.

Built from the ground up to meet the extreme demands of tightly coupled AI clusters, the switch achieves 250ns switch latency at full 51.2 Tbps throughput while delivering line-rate switching performance even at minimum 64-byte packet sizes. This translates to supporting up to 77 billion packets per second, a critical capability for AI training and inference workloads that require massive data throughput with minimal delay.

The switch incorporates several breakthrough features that address traditional Ethernet limitations in AI environments. It implements adaptable, optimized Ethernet headers that reduce overhead from 46 bytes down to as low as 10 bytes while maintaining full Ethernet compliance, boosting network efficiency significantly.

Additionally, the lossless fabric technology eliminates packet loss through Link Layer Retry (LLR) and Credit-Based Flow Control (CBFC), ensuring the reliability demanded by data-intensive AI workloads.

Perhaps most significantly, the Tomahawk Ultra includes In-Network Collectives capability, which accelerates AI performance by executing operations like AllReduce and Broadcast directly within the switch chip rather than burdening expensive AI accelerators.

This endpoint-agnostic feature enables immediate adoption across diverse system architectures and can substantially reduce job completion times for machine learning workloads while improving utilization of compute resources.

Join our Telegram group and never miss a breaking digital asset story.

Broadcom Shares Gain as Firm Releases New Product for AI Workloads

Broadcom shares have demonstrated exceptional momentum in 2025, with the stock trading at $281.72 in premarket hours, up $6.12 or 2.22% from the previous close of $275.60. The company’s year-to-date return of 22.19% significantly outpaces the S&P 500’s 6.64% gain, reflecting strong investor confidence in Broadcom’s AI-focused strategy. Over the past year, AVGO has delivered an impressive 66.25% return compared to the S&P 500’s 11.39% gain, with the stock reaching new record highs amid the AI infrastructure boom.

The company’s financial metrics underscore its strong market position, with a massive market capitalization of $1.325 trillion and trailing twelve-month revenue of $57.05 billion. Broadcom maintains healthy profitability with a 22.64% profit margin and generates substantial cash flow of $26.15 billion in levered free cash flow. The forward P/E ratio of 33.56 suggests investors are pricing in continued growth from AI-related products like the Tomahawk Ultra, while analyst price targets range from $210 to $400 with an average of $286.36.

Industry analysts are particularly bullish on Broadcom’s positioning in the AI networking space. Bloomberg Intelligence’s Kunjan Sobhani noted that the Tomahawk Ultra’s ability to deliver sub-microsecond switching and lossless transport marks “a pivotal step toward meeting the demands of an AI scale-up stack projected to be double-digit billions in a few years.” The 100% pin compatibility with the previous Tomahawk 5 ensures rapid time-to-market for customers, potentially accelerating revenue recognition from this new product line.

Disclaimer: The author does not hold or have a position in any securities discussed in the article. All stock prices were quoted at the time of writing.