Nvidia rises after UBS keeps Buy rating amid Blackwell delay
Nvidia’s (NASDAQ:NVDA) stock rose about 4% on Monday after UBS retained its Buy rating on the shares, amid a delay in the availability of the company’s Blackwell graphics processing units, or GPUs.
The firm has a $150 price target on Nvidia’s shares.
Analysts led by Timothy Arcuri noted that after a series of customer discussions and more supply chain work, they were making slight adjustments to their 2025 EPS model to $4.87 from $4.95. They said that — initial Blackwell customer volume shipments are delayed maybe four to six weeks at most (putting them at the very end of January 2025); and this delay looks likely to be largely filled in by many customers taking more H200 chips due to very short lead times.
In addition, the analysts noted that lead customers should have first Blackwell instances stood up in April 2025 timeframe; and AI labs are still upsizing and lengthening their instance commitments and enterprises are rapidly growing as a proportion of the demand mix — both bullish indicators.
Arcuri and the team noted that Blackwell has started production at Taiwan Semiconductor Manufacturing (TSM) but there are some initial yield challenges due to the added complexity linked with CoWoS-L (local silicon bridge) for B100/B200 – essentially a larger interposer/package than CoWoS-S, which is being used for H100/H200, and this is producing less initial volume than planned.
CoWoS stands for Chip on Wafer on Substrate.
To maximize the available CoWoS-L capacity, Nvidia is replacing B100 with B200A which is using CoWoS-S, thereby freeing up scarce CoWoS-L capacity for B200/GB200 – this is higher value for Nvidia and also better matches customer demand mix as the vast majority of customers want rack-scale GB200 but most (other than two) are not fully ready with their liquid cooling infrastructure and the new B200A is allowing the company to offer a new 64 rack option (GB200A-NVL64) that is air cooled, according to the analysts.
The analysts added that some investors have wondered if there is a flaw in the Blackwell design at the chip level, the analysts said, they push back on this as chipset level simulation has been verified multiple times with feedback from the supply chain indicating there has been no “fine tuning” of design tools.
There is sufficient headroom and short enough lead time in H100/H200 supply for Nvidia to largely backfill lower initial Blackwell volumes with H200 units (and the supply chain is seeing upward revisions consistent with some backfill, as per the analysts.
Arcuri and his team said, based on customer discussions, they had long suggested initial Blackwell shipments were not going to start until mid-December, so this push-back of four to six weeks should be largely invisible to most, if not all, end customers.
The analysts added that they had highlighted in the past the evolving theme around air/liquid cooling and it remains a key issue that could shape the revenue mix for both Nvidia and AMD (AMD) (for MI350) in 2025. At a high level, these issues help to highlight the challenges that Nvidia (and AMD) face with an annual product cadence, especially with these rack scale systems.
However, the analysts noted that AMD has already gone to chiplet and was a pioneer of CoWoS with TSM, and Nvidia, according to them, has to go full chiplet for Rubin to reuse some of the IO die — this will be a challenge to see in 2026.
Nvidia (NVDA) has a Hold rating at Seeking Alpha’s Quant Rating system, which consistently beats the market. Meanwhile, the Seeking Alpha authors’ average rating is also Hold, but the average Wall Street analysts’ rating is more positive with a Strong Buy.