TL;DR:
- Google reaffirms its commitment to Broadcom as its AI chip supplier, dismissing reports of a potential breakup by 2027.
- Broadcom’s stock initially dipped by 4.3% in response to rumors that Google might design its tensor processing units (TPUs) in-house.
- Google has intensified its investments in chip technology to compete with Microsoft in the generative AI applications market.
- The report emerged from a pricing dispute between Google and Broadcom over TPU chips and Google’s exploration of alternative suppliers.
- Marvell Technology emerges as a potential replacement for Broadcom in server-chip integration.
- Google praises Broadcom as an excellent partner, with no anticipated change in their collaboration.
- Marvell Technology declined to comment on the situation.
- Broadcom is positioned as the second-largest beneficiary of the generative AI boom, following Nvidia.
- J.P. Morgan analysts estimate Broadcom could earn $3 billion in revenue from Google this year due to increased TPU processor orders.
- Google and Broadcom collaborate on AI chip design, with Broadcom set to contribute to Google’s sixth-generation processor.
- The broader tech industry sees a trend of major companies developing custom chips tailored to their specific needs, driven by the rising costs of Nvidia’s H100 chip.
Main AI News:
In the dynamic realm of tech alliances, Alphabet’s Google remains resolute in its commitment to its AI chip supplier, Broadcom. Dismissing recent speculations suggesting otherwise, Google asserts its unwavering partnership with Broadcom, even in the face of rumors that the tech titan was contemplating severing ties with the chipmaker as a key supplier of artificial intelligence chips by 2027.
Following the initial release of The Information’s report, Broadcom experienced a brief dip in its stock value, plummeting by as much as 4.3%. The report alluded to Google’s contemplation of in-house design for the coveted tensor processing units (TPUs), a strategic move that could potentially translate into billions of dollars in annual savings should Google decide to proceed.
Throughout the year, Google has strategically bolstered its investments in chip technology, strategically positioning itself to compete head-to-head with Microsoft for dominance in the burgeoning generative AI applications market, a realm exemplified by innovations like ChatGPT.
The report’s genesis lay in the backdrop of a pricing dispute between Google and Broadcom concerning TPU chips. It was further noted that Google had been actively exploring alternatives, including a potential collaboration with Marvell Technology as a successor to Broadcom in the domain of server-chip integration.
In response to these developments, a Google spokesperson asserted, “Our work to meet our internal and external Cloud needs benefits from our collaboration with Broadcom; they have been an excellent partner, and we see no change in our engagement.”
Meanwhile, Marvell Technology opted not to comment on the matter, observing a 1.3% decline in its own stock value.
Broadcom’s standing as the second-largest beneficiary of the generative AI boom, trailing closely behind Nvidia, has cemented its importance in the semiconductor landscape. In June, Broadcom’s CEO, Hock Tan, anticipated that generative AI technology could constitute over 25% of the company’s semiconductor revenue in the coming year.
Notably, J.P. Morgan analysts projected substantial revenue for Broadcom, estimating potential earnings of $3 billion from Google in the current fiscal year, prompted by a notable upswing in orders for its TPU processors.
In terms of AI chip development, Google has been co-designing its chips in collaboration with Broadcom, with the semiconductor firm already poised to contribute to Google’s sixth-generation processor, as highlighted by analysts. Furthermore, Broadcom has been actively engaged with Meta Platforms, partnering on the development of custom chips for the social media giant.
Across the technology landscape, major players, from Microsoft to Amazon.com, have embarked on the race to develop bespoke chips tailored to their specific workloads. This heightened pursuit of custom chip solutions has intensified in recent times, driven partly by surging costs associated with Nvidia’s H100, the chip at the heart of many generative AI applications, which has nearly doubled in price from its initial valuation of $20,000.
Conclusion:
The reaffirmation of Google’s partnership with Broadcom in the AI chip domain signals stability in their collaboration. The potential for in-house chip design underscores Google’s commitment to cost-efficiency. This intensified focus on AI chips positions Google to compete effectively in the generative AI market, while Broadcom’s strong presence in this sector remains unshaken. As custom chip development gains momentum across the tech industry, it is apparent that market players are actively seeking tailored solutions to optimize performance and mitigate rising costs, epitomized by the price surge of Nvidia’s H100 chip.
