With Nvidia‘s flagship GPU Technology Conference (GTC) just days away, Bank of America is telling investors exactly what to watch and why the stock’s current valuation may not reflect what is coming.

Analyst Vivek Arya reaffirmed a buy rating on Nvidia (NVDA) with a $300 price target ahead of the March 16 GTC keynote from CEO Jensen Huang. The note flags three specific areas of focus that Arya believes could be the clearest signals of Nvidia’s trajectory through 2027 and 2028.

The timing matters. Nvidia shares are currently trading at what Arya describes as a historically depressed forward price-to-earnings multiple of 17x, a level he calls a trough following the Blackwell architecture’s massive $500 billion cumulative sales ramp.

The bank sees the GTC keynote as the catalyst that could begin closing that valuation gap.

Three things Bank of America is watching at Nvidia’s GTC

Arya’s note is precise about what investors should focus on when Jensen Huang takes the stage on March 16. The three areas are not just product announcements. They are signals about how far ahead of competitors Nvidia’s roadmap actually extends.

What BofA flagged as the key GTC catalysts

  • Product roadmap through 2028: BofA expects Nvidia to outline its full pipeline from the current Vera Rubin platform through to Feynman GPUs in 2028, a three-generation visibility that Arya says locks in developer and enterprise commitments well ahead of rivals.
  • Co-designed inference portfolio: The bank anticipates announcements across a new range of customized products including CPX chips for inference prefill workloads and a Language Processing Unit, or LPU, for low-latency decode, potentially integrated inside Nvidia’s next-generation rack systems.
  • Proprietary optics in scale-up networks: BofA is watching for details on Nvidia’s next-generation 102.4T Spectrum-6 switch and the 115T Quantum-X with co-packaged optics, technology the bank says could become essential infrastructure in large-scale AI cluster deployments.

Why Nvidia’s inference pivot is the bigger story

Much of the AI infrastructure conversation over the past two years has focused on training, the process of building large language models. Nvidia has dominated that market.

But the bank’s note signals that the next battleground is inference, the process of actually running those models at scale for end users.

  • Nvidia stock gets major reality check on ‘$100B’ number
  • Nvidia CEO delivers blunt 7-word rebuttal on software stocks
  • Bank of America resets Nvidia price target after earnings

More Nvidia: BofA describes the new CPX and LPU products as “a new wave of co-designed and disaggregated AI infrastructure,” adding that these architectures could become increasingly important as AI workloads shift from training toward inference at scale.

Arya specifically highlighted the LPU, developed in partnership with Groq, as a potentially strategic addition. The unit uses large amounts of fast on-chip SRAM memory to handle low-latency token generation, a capability that directly addresses one of the most demanding requirements in real-time AI applications.

BofA compared the potential long-term significance of the Groq deal to Nvidia’s 2020 Mellanox acquisition, which became the foundation of its networking and AI scaling business.

Related: History of Nvidia: Company timeline and facts

The NVDA valuation case BofA is making

Beyond the product roadmap, BoA’s note makes a straightforward valuation argument. At 17x forward earnings, Nvidia is trading at the low end of its historical range of 25x to 56x.

Arya’s $300 target is based on 28x calendar year 2027 earnings, a multiple he describes as justified, given Nvidia’s dominance across AI compute, networking, and software.

The data-center revenue projections underpinning that target are significant. Consensus estimates across Wall Street already call for Nvidia’s data-center business to reach approximately $750 billion in cumulative revenue across 2026 and 2027, rising to roughly $1 trillion for 2027 and 2028.

BofA believes any color Huang provides on the Rubin ramp at GTC could push those estimates higher still.

Nvidia has already committed approximately $95 billion in supply agreements for fiscal year 2027, including investments across the AI ecosystem in companies such as OpenAI and Anthropic. That level of supply chain commitment gives BofA confidence that the revenue pipeline is real, not speculative.

Bank of America believes that the details Nvidia CEO Jensen Huang provides on the Rubin ramp at GTC could increase the chipmaker’s data-center revenue projections.

Key questions the market still needs answered

Campbell/Getty Images Arya’s note is bullish but not without caveats. The bank is also looking for GTC to provide clarity on several unresolved questions that have kept some investors cautious.

What BofA still needs to hear from Jensen Huang:

  • The impact of the Middle East conflict on supply chains, sovereign demand, and energy cost inflation across U.S. and non-U.S. data-center buildouts
  • Nvidia’s ability to source enough wafers, memory, substrates, and optics against an annual product cadence
  • The timing of co-packaged optics introduction and whether it will be optional or required in Nvidia clusters
  • How Vera Rubin’s cost per token, estimated at roughly 10x cheaper than Grace Blackwell, translates into real-world customer economics

What BoA’s note means for investors

BofA’s note lands at a moment when Nvidia’s stock has been under pressure, despite fundamentals that the bank describes as intact. The 17x forward P/E that Arya calls a historical trough is the same multiple that has preceded some of Nvidia’s strongest multi-year rallies.

The GTC keynote on March 16 is not just a product event. For investors watching Nvidia (NVDA), it is the moment the market gets its clearest look yet at whether the $1 trillion data-center consensus for 2027 and 2028 is a ceiling or a floor.

Related: Bank of America names the U.S. auto stocks to own