· Event impact
Goldman Sachs sees custom AI chips (ASICs) matching GPU demand by 2027
Transmission path
A structural shift in AI hardware spending from general-purpose GPUs to specialized ASICs creates a new competitive dynamic, potentially diverting market share and capital from GPU leaders to ASIC specialists.
Market mechanism
A structural shift in AI hardware spending from general-purpose GPUs to specialized ASICs creates a new competitive dynamic, potentially diverting market share and capital from GPU leaders to ASIC specialists.
Extended read
A new forecast from Goldman Sachs suggests a significant evolution in the AI hardware landscape. The bank predicts that demand for custom-built Application-Specific Integrated Circuits (ASICs) will grow to match that of general-purpose GPUs by as early as 2027. This indicates that hyperscalers and other large AI developers are increasingly seeking tailored, power-efficient solutions for their specific workloads, moving beyond off-the-shelf GPUs. The growth potential is substantial, with separate forecasts from Bloomberg Intelligence projecting a 27% compound annual growth rate for the ASIC market through 2033. This trend is not just about a growing market, but a shift in its composition. The report identifies Broadcom and Marvell Technology as the key enablers and beneficiaries of this shift. These companies specialize in designing custom silicon for clients like Google, Amazon, and Meta. Broadcom, in particular, is expected to command a dominant 60% share of this emerging market, positioning it as a primary competitor to Nvidia in the next phase of AI infrastructure buildout.
Exposed assets
AVGO · MRVL · NVDA · SMH
Continuous event tracking, options routing, and portfolio overlap for this event and the assumptions it moves live inside Market Ontology. Start a trial →