In the midst of AI’s rapid transformation, BofA Securities analyst Vivek Arya sees a significant opportunity for NVIDIA Corp NVDA, Broadcom Inc AVGO and Marvell Technology Inc MRVL.

These companies are positioned at the heart of what Arya refers to as AI’s “Sputnik moment,” drawing an analogy to the Cold War era when the U.S. ramped up its efforts after the Soviet Union's surprise launch of Sputnik into space.

In this context, DeepSeek's rise in AI is seen as a catalyst for an even bigger wave of investment and BofA's analyst believes the hyperscalers and semiconductor companies will reap the rewards.

“DeepSeek: More Evolutionary Than Revolutionary” – But A Wake-Up Call For AI Investment

Arya is not overly impressed by DeepSeek's software contributions, as they are more about “evolutionary” progress rather than a “revolutionary” shift in AI. Still, the implications for the semiconductor space are clear. While “DeepSeek’s contributions are noteworthy,” they don’t represent a drastic change in course for the industry.

However, they do accelerate AI capex investments, especially in “AI compute TAM toward $500bn+” over the next several years.

For Nvidia, Broadcom and Marvell, the timing couldn't be better.

These companies are already in a prime position, supplying essential hardware for AI infrastructure. Arya maintains a Buy rating on all three, emphasizing that the “rising tide should benefit all” as the AI compute market expands significantly.

The Role of Hyperscalers, AI Hardware

One of the core insights Arya highlights is that the U.S. hyperscalers (tech giants like Amazon.com Inc AMZN, Alphabet Inc‘s GOOGL GOOG Google, and Microsoft Corp MSFT will continue to drive much of the AI spending in the coming years. These companies will push toward “artificial general intelligence (AGI)” and broader enterprise use cases. The key to their success will be investing not only in “software” but also in “advanced hardware.”

This is where Nvidia, Broadcom, and Marvell come in, offering both the hardware capabilities and adaptability to meet the ever-evolving needs of AI workloads.

Arya forecasts the total AI compute market to “double towards $500bn+” by 2028, a growth driven by the need for large-scale model training, AI inference, and cloud infrastructure upgrades. As the race for AI dominance continues, “U.S. hyperscalers” will remain at the forefront, driving the bulk of this spending.

‘Rising Tide Should Benefit All’

Nvidia, Broadcom and Marvell are all expected to benefit from this surge in demand. With DeepSeek’s AI software optimization now part of the broader landscape, it's clear that the hardware side of the equation — “ASIC versus general purpose GPU” — will continue to evolve.

Custom ASICs (like those from Broadcom and Marvell) offer cost efficiencies, but “merchant GPU (NVDA) could prove more adaptable to rapidly changing compute and model mix.”

For investors, the message is clear: These companies are well-positioned to capitalize on the growing AI infrastructure market. With AI compute TAM expected to rise to “$500bn+”, BofA sees the current downturn in the sector as an “enhanced Buy opportunity.”

Read Next: