This discussion analyzes the evolving landscape of semiconductor networking and custom silicon, anchored by recent updates from Marvell and Broadcom. Ben and Jay review Marvell’s Industry Analyst Day, noting the company’s strategic pivot toward networking and optical interconnects over pure custom logic. They debate the technical and timeline challenges of shifting AI clusters from copper to optical (fiber) cabling, contrasting Google’s proprietary all-optical approach with the broader merchant market. The conversation then moves to Broadcom’s recent earnings, dissecting the market’s negative reaction to CEO Hawk Tan’s low-energy performance, the looming competitive threat of MediaTek at Google, and the complexities of modeling AI revenue. Finally, they close with a year-end review of AI stock performance, highlighting growing capital sensitivity and skepticism regarding the sustainability of current AI infrastructure spending.

In this episode, Ben and Jay discuss a range of topics including Ben's health update, Amazon's recent AWS event focusing on AI compute, the competitive landscape with Nvidia and Google, Marvell's earnings and challenges in custom silicon, networking innovations with DPUs, Marvell's acquisition of Celestial, Nvidia's investment in Synopsys, Intel's resurgence in advanced packaging, and the leadership changes at Apple. The conversation highlights the evolving dynamics in the tech industry, particularly in AI and cloud computing.

In this conversation, Ben Bajarin discusses the recent innovations surrounding NanoBanana and its new feature, Notebook LM, which has significantly impacted the way infographics are created. He highlights the ease with which users can transform data and text into visually appealing infographics, showcasing the power of AI in data visualization.

The conversation explores the dynamics of market competition, particularly focusing on AMD's position and the strategies of hyperscalers. The discussion delves into game theory as a framework for understanding potential future scenarios in the tech industry. They also discuss key neocloud earnings and engage in a conversation on if these companies are REITs or neoclouds. 

This episode argues that today’s AI exuberance fits a familiar pattern: bubbles misallocate capital on the way up but leave behind productive infrastructure that powers the next S-curve. We revisit the “boom-bust-build-out” cycle and apply it to compute and the grid, note why “good enough” AI latency could flip capex behavior, and push back on modeling everything in gigawatts—useful for planning, risky for strategy. On companies: ARM’s quarter was solid, with rising royalties/CSS stickiness and a strongly implied first-party chip effort complemented by the DreamBig memory-controller acquisition to improve AI-era CPU roles. Qualcomm benefits from a higher-ASP Android cycle, nurtures auto/smart-glasses adjacencies, and eyes DC inference. On supply: Elon hedges silicon with foundry allocations while fab-building talk reads as negotiating leverage; memory/storage stay tight with longer contracts and measured adds. Net: optimism about what survives the eventual correction, caution on GW-only thinking, and watchpoints around ARM’s vertical creep, Qualcomm’s DC push, and supply-chain discipline.

The conversation delves into the competitive landscape of cloud computing, focusing on Nvidia's ambitions and the fragmentation of the market with numerous Neo clouds. Jay Goldberg discusses the implications of this fragmentation for Nvidia and its customers, who are increasingly seeking to develop their own custom silicon.

In this episode, Ben Bajarin and Jay Goldberg discuss Intel's recent earnings report, highlighting a sense of stability in the market compared to previous downturns. They explore the demand for CPUs, particularly in the enterprise sector, and the implications of upcoming product launches. The conversation shifts to Intel's foundry developments, where they express optimism about new manufacturing processes and customer engagement. They also analyze the competitive landscape of AI compute infrastructure, particularly focusing on Amazon's challenges with its Tranium chips and the implications of Anthropic's partnership with Google. Finally, they delve into the future of AI agents, discussing the current limitations and potential advancements needed for these technologies to become viable.

In this episode, Ben Bajarin and Jay Goldberg discuss the recent industry discussion (confusion?) on GPU depreciation.  Jay Goldberg was also at OCP this week and they disucss takeaways from the networking show related to AI compute and AI datacenter. 

In this episode, Ben Bajarin and Jay Goldberg discuss the recent partnership between OpenAI and AMD, exploring its implications for the AI compute landscape. They delve into the competitive dynamics between AMD and NVIDIA, OpenAI's ambitions to become a hyperscaler, and the financial challenges associated with massive AI infrastructure investments. The conversation shifts to Intel's advancements showcased during a tour of their Fab 52, highlighting their competitive edge in semiconductor manufacturing. Finally, they touch on Qualcomm's acquisition of Arduino and its potential impact on the IoT sector.

Takeaways

OpenAI's partnership with AMD marks a significant shift in AI compute.
AMD is positioning itself as a serious player in AI GPUs.
OpenAI's ambition to become a hyperscaler raises questions about funding.
The demand for AI compute is concentrated among a few major players.
Intel's Fab 52 showcases advanced manufacturing capabilities.
Intel's 18A process is competitive but needs to prove itself in products.
Qualcomm's acquisition of Arduino aims to strengthen its IoT strategy.
The AI infrastructure build-out is unprecedented in scale and cost.
There is uncertainty about the long-term demand for AI services.
The competitive landscape in cloud computing is rapidly evolving. 

In this episode, Ben Bajarin and Jay Goldberg delve into the evolving landscape of hyperscalers and Neo clouds, discussing their definitions, the significance of power contracts, and the shift from Bitcoin mining to AI infrastructure. They explore the challenges posed by storage constraints in the age of AI, the implications of OpenAI's new app Sora, and Meta's acquisition of Rivos for custom silicon development. The conversation culminates in a reflection on the importance of semiconductor manufacturing and the future of technology.