NVDA NDR highlights diversified growth opportunities

Research evidence excerpt

NVDA NDR highlights diversified growth opportunities

IdeaMindustry will find ways to keep compute supply on track. This may in part be a

reference to something that we have written about in the past - that NVIDIA cut the

LPDDR5 content per rack in half so that they can meet rack demand - or could imply

future shifts. They said that compute, networking, and memory are all inter-related,

and it is possible to use the other two to optimize a shortage of the third. The

company also highlighted the importance of Groq technology, which uses SRAM

that "used to be much more expensive than DRAM". While some of this commentary

could dampen enthusiasm for memory stocks, we would highlight the starting place

that the offsets are happening due to a multi year shortage.

Importance of "nemotron" and open source model development. NVIDIA

developed the nemotron model to enable more application specific AI designs that

can't use more generic closed models; they used the example of circuit design. Open

model development allows enterprise customers to own the full stack, with the

ability to customize that usage around their own expertise.

Thoughts on the stock: NVIDIA remains our top pick in semis, which has been a

tough call since we made it in March the stock is higher but has underperformed a

sharp runup in the peer group - but our conviction remains high. We understand that

there is more leverage in other parts of the supply chain - in fact that was our

reason for rotating our top pick on NVDA last year to first Sandisk, then Micron -

but we now feel that this is the best value in the group. To be sure, the multiple is

likely constrained by the large market cap and index weighting, but in our view the

value disparity will revert over time. The points made here reinforced our