Marvell Shipments to Triple by 2027 as Hyperscalers Cut Nvidia Dependency
Marvell Technology’s AI accelerator shipments are forecast to triple by 2027 as hyperscalers develop custom chips, helping it nearly double in 2026 while Nvidia rises 7%. Google Cloud grew 63% in Q1 by deploying custom TPUs that cut Nvidia dependency, and D-Wave stock jumped 40% on Nvidia’s Ising models.
1. Hyperscaler Custom Accelerators Growth
Major cloud providers are shifting from off-the-shelf GPUs to custom ASICs, driving Marvell Technology’s AI accelerator shipments to forecasted growth of 3x by 2027. This shift fueled Marvell’s stock to nearly double in 2026 while Nvidia’s share gain was 7%, signaling rising competition in data-center AI hardware.
2. Google Cloud’s TPU Deployment
Google Cloud reported 63% revenue growth in Q1, led by adoption of in-house Tensor Processing Units that handle generative AI workloads and inference tasks. These custom TPUs reduce reliance on Nvidia GPUs by optimizing training efficiency and latency for AI models across Google’s services.
3. D-Wave’s Ising Model Impact
D-Wave Quantum’s stock surged 40% after Nvidia launched Ising AI models designed to calibrate quantum systems, boosting error-correction speed by 2.5x and cutting error rates threefold. This development underscores Nvidia’s influence on adjacent quantum computing ecosystems and potential new market applications.