All articles tagged with #custom chips
Broadcom CEO Hock Tan projects AI semiconductor revenue will double to $115 billion in 2027 and again to $230 billion in 2028. Despite this aggressive growth trajectory, the stock trades at a forward P/E of 19x, significantly lower than peers like NVIDIA and AMD. Analysts view the current price as undervalued, though supply constraints remain a key risk.
Broadcom reported $16.7 billion in AI semiconductor revenue for Q3 FY2026, a 221% year-over-year increase. CEO Hock Tan projects AI revenue will reach $115 billion in fiscal 2027 and $230 billion in fiscal 2028. The company launched VMware Private AI Cloud and secured a design win with OpenAI's 'Jalapeno' accelerator, which outperforms NVIDIA GPUs at half the cost for inference. Free cash flow hit $13.67 billion, and the company raised its dividend for the 15th consecutive year.
Microsoft and Intel have struck a custom chip deal worth over $15 billion, with Intel producing chips designed by Microsoft using its 18A process. The partnership marks a significant move for Intel's chip foundry services and a shift in the industry as more companies seek to produce their own self-designed chips. This deal positions Intel to compete with TSMC and signifies a strategic shift for both companies in the chipmaking world.
Nvidia is reportedly investing $30 billion to establish a new unit for custom chip solutions, targeting various sectors including AI, telecom, cloud, automotive, and gaming hardware. The move comes as the company faces increasing competition in the generative AI arena, despite its dominance with products like the H100, A100, and H200 AI chips. The investment reflects the growing demand for more efficient and specialized hardware in the tech industry, driven by the escalating energy footprint of generative AI and the need to improve efficiency.
NVIDIA is forming a new business unit to develop custom chips for AI, game consoles, auto, and telecom markets, aiming to compete with industry leaders like Broadcom and Marvell Technology. The move comes as the company seeks to diversify its supply chain and capitalize on the growing demand for custom chips, with analysts predicting the market to reach $30 billion by 2023. NVIDIA has reportedly engaged with major tech companies like Amazon, Meta, Microsoft, Google, and OpenAI to offer custom AI chip services, signaling a strategic shift to consolidate its position in emerging domains of the artificial intelligence market.
Nvidia is in discussions with major tech companies such as OpenAI and Google to develop customized AI chips, potentially posing a threat to competitors like Broadcom. As the demand for AI chips grows, Nvidia aims to solidify its position in the semiconductor market by creating custom chips tailored to specific needs. The company's success has led to a shortage of its chips, prompting plans to address the supply-demand mismatch and expand its reach across various industries.
Nvidia is venturing into the custom chip market by forming a group to help clients build their own solutions using Nvidia IP or chiplets, potentially leveraging their relationship with TSMC or Samsung for fabrication. This move could position Nvidia as an AI licensing giant and allow clients to tap into Nvidia's in-house and AWS supercomputers for design optimization. The speculation suggests that Nvidia is preparing to transform the industry and monetize through licensing deals, potentially indicating a shift in the semiconductor market.
Nvidia's stock reached a new high after reports of plans to create a new business unit focused on custom chips, particularly for AI. This move could strengthen Nvidia's position as the world's most valuable chip company. The company's shares rose over 3% to $717.71 per share, and it is set to report its Q4 fiscal 2024 results on February 21.
Nvidia is establishing a new business unit to develop custom chips for cloud computing and AI, aiming to tap into a $30 billion market and counter the increasing competition in the semiconductor industry. The company's dominant position in AI chips has led major customers like Microsoft, Alphabet, and Meta Platforms to seek alternatives, prompting Nvidia to offer custom AI chips to meet specific needs. The move is expected to impact competitors like Broadcom and Marvell, and Nvidia also plans to target the telecom, automotive, and video game markets with its custom chip offerings.
Nvidia is reportedly aiming to enter the custom chip space, seeking to build a new business unit for custom-designed AI chips in response to the growing trend of companies developing their own specified chips for specific needs. This move is seen as a proactive effort to address potential market share erosion and expand opportunities in the energy-intensive applications and chips market, particularly in competition with major cloud providers like Microsoft, Amazon, and Meta.
Nvidia is establishing a new business unit to create custom chips for cloud computing and other industries, aiming to tap into the growing market for bespoke AI chips and defend against increasing competition. The company, which currently dominates the high-end AI chip market, is seeking to assist customers in developing custom AI chips to address specific needs, potentially reducing energy consumption and design costs. With plans to target telecom, automotive, and video game markets, Nvidia's move into custom chip design poses a threat to established players like Broadcom and Marvell, and the company is also in talks with Ericsson for a wireless chip incorporating its GPU technology.
Meta Platforms plans to deploy a new version of its custom chip, "Artemis," in its data centers this year to support its AI push, aiming to reduce dependence on Nvidia chips and control costs associated with running AI workloads. The chip, designed for inference, could potentially save the company hundreds of millions of dollars in annual energy costs and billions in chip purchasing costs. Meta also plans to have roughly 350,000 flagship "H100" processors from Nvidia by the end of the year, accumulating the equivalent compute capacity of 600,000 H100s in total.
Microsoft has developed its own custom AI chip, called Azure Maia, and an Arm-based CPU, named Azure Cobalt, to power its Azure data centers and meet the growing demand for AI. The Azure Maia chip is designed for large language model training and inference, while the Azure Cobalt CPU is optimized for general cloud services. Both chips are part of Microsoft's deep overhaul of its cloud server stack to optimize performance, power, and cost. The Maia chip has 105 billion transistors and supports faster model training and inference times. Microsoft is also sharing its rack designs with partners but keeping the chip designs in-house. The chips are expected to be available in 2024.
Amazon is racing to catch up with Microsoft and Google in the field of generative AI by developing custom chips, Inferentia and Trainium, for training and accelerating AI models. These chips offer an alternative to Nvidia GPUs, which have become difficult and expensive to procure. While Amazon has made progress with its custom silicon, others like Microsoft and Google have acted faster and invested more in capturing the generative AI market. However, Amazon's cloud dominance and existing customer base could give it an edge in enabling customers to expand into generative AI applications. AWS is also offering a range of developer tools and services focused on generative AI, such as AWS HealthScribe and SageMaker.
Meta has unveiled its in-house infrastructure for AI workloads, including generative AI, with the aim of pushing the boundaries of AI research at scale. The company has developed a custom chip, the Meta Training and Inference Accelerator (MTIA), capable of both training AI models and running them, and a supercomputer, the Research SuperCluster (RSC), containing 2,000 Nvidia DGX A100 systems sporting 16,000 Nvidia A100 GPUs. Meta is also developing another chip, the Meta Scalable Video Processor (MSVP), designed for the processing needs of video on demand and live streaming. The company is feeling increasing pressure from investors concerned that it's not moving fast enough to capture the potentially large market for generative AI.