US Stock Market Move | Marvell Technology, Inc. (MRVL.US) opened with a more than 5% increase in trading, and it is reported that they are in deep discussions with Alphabet Inc. Class C regarding the development of a new AI chip.

date
21:58 20/04/2026
avatar
GMT Eight
On Monday, Marvell Technology (MRVL.US) opened up more than 5%, hitting a new all-time high, with a cumulative increase of 48% for the month.
On Monday, Marvell Technology, Inc. (MRVL.US) opened up more than 5%, hitting a new all-time high and accumulating a 48% increase within the month. According to reports, Alphabet Inc. Class C (GOOGL.US) is currently in deep discussions with Marvell Technology regarding the development of two new types of artificial intelligence chips. The core of these discussions revolves around two new chips aimed at significantly improving the efficiency of AI model operation. The first is the crucial "Memory Processing Unit" (MPU), designed to work in conjunction with Alphabet Inc. Class C's existing Tensor Processing Unit (TPU) to address memory bottleneck issues in AI computations; the second is the next generation TPU optimized for handling AI inference workloads. The goal is to complete the design and start testing production of the MPU before 2027, indicating Alphabet Inc. Class C's efforts to accelerate the construction of a more cost-effective vertically integrated hardware system to compete against NVIDIA Corporation's dominance in the GPU field. Analyst Aaron Rakers from Wells Fargo & Company highlighted in a recent research report that if the deal with Alphabet Inc. Class C is finalized, it would be a "meaningful incremental positive factor" for Marvell. Rakers emphasized that Marvell's advantage in this bidding process stems mainly from its recently launched Structera CXL solution. He wrote in a customer report that the potential MPU order might be a manifestation of Marvell's technological strength in Structera CXL, which significantly improves computing efficiency in large-scale data centers through rack-level memory pooling.