Samsung Q2 25 – Nadir of Fortunes

More apologies are expected.

• Samsung reported another difficult quarter where a confluence of events caused profit to fall by 56% YoY in strong contrast to its competitors, highlighting once again the bonanza that Samsung is missing out on.
• Consequently, I expect more apologies to shareholders when full results are reported on 31^st July and a redoubling of its efforts to regain its crown as the king of memory.
• Q2 25 preliminary revenues / operating profit were KRW73tn – KRW75tn (KRW74tn) / KRW4.5tn – KRW4.7tn (KRW4.6tn), where revenues were broadly in line with forecasts, but profits were meaningfully behind.
• However, the notoriously leaky Korean market has been well aware of this for some time and has long since priced it in, which is why we are not seeing a worse reaction in the shares this morning.
• However, it is clear that Samsung is not out of the woods yet and that there is still a lot of work to do.
• This quarter includes an inventory write-down on chips that were designed for China but can no longer be sold there due to the increasing restrictions on the export of advanced AI chips to China.
• Profits also suffered as its inability to produce capable high bandwidth memory (HBM) for Nvidia led to further erosions of market share in contrast to its rivals, SK Hynix and Micron, who have powered ahead.
• HBM is the hottest segment of the market at the moment as it is an essential component of the AI data centre and is sold alongside Nvidia’s and everyone else’s GPUs.
• Given the scale of demand for AI GPUs and the amount of money being spent on capex for AI currently, both Micron and SK Hynix are already sold out for 2025, while Samsung is floundering.
• This was evident at GTC 2025 where both SK Hynix and Micron showed off their HBM products and roadmap, while HBM was a dirty word on the Samsung stand.
• This is why Samsung is trading at a substantial discount to its peers and the global technology sector in general, which, in my opinion, represents an opportunity.
• This is because Samsung’s history of getting out of messes like its current predicament is fairly good, and I think there is a better than even chance that it will fix the problems that it has in HBM and return to its former memory glory.
• My understanding is that Samsung has pretty much given up on the current generation of HBM and is instead putting all of its efforts into getting HBM4 right and qualifying with Nvidia as a supplier.
• HBM4 is the series that is likely to accompany Nvidia’s Rubin, which will be launched in 2026 and is expected to commence volume shipments in H2 2026.
• However, the signal that Samsung has fixed its problems will be when it qualifies its HBM4 chip with Nvidia, which I am looking for in the second half of this year.
• This is the catalyst for the recovery rally that should take the shares back up to the KRW85,000 level, erasing the discount and putting Samsung back in line with its peers.
• However, in the meantime, the Q2 25 results are going to be a pretty dour affair given the Chinese problems, where I expect to see management apologise again for the poor performance.  
• I hold a position in Samsung via the London-listed global depositary receipt (GDR) and remain very comfortable with it up to KRW85,000, at which point I would reconsider the position.