Samsung Profits Plunge 95%
cnn.com
cnn.com
On first blush, the headline just looks like a consequence of quarterly revenue miss when there's an implied floor to both cost of sales and operating expense in the face of elevated FX headwinds and substantial macro uncertainty.
TSMC posted poor results last week, not dissimilar to Texas Instruments on Tuesday. Now we see Samsung join the club, and the trend is likely to be no different for Intel after market close today.
All the while Nvidia has been trading in the $270-280 range on pure AI hype euphoria, completely oblivious to $2+ billion of deprecated Ampere architecture inventory glut, the export restrictions hammer that dropped last year when China sales alone accounted for over 21% of FY22 revenue, and more recently, AIB manufacturers informed that RTX 4070 production is being paused a mere 1 week after launch in an attempt to prevent inventory buildup, just to call out a few.
[1] https://images.samsung.com/is/content/samsung/assets/global/...
As a reference, between 2008 and 2011 when ATI/AMD had the biggest technological lead over Nvidia's disaster architectures, even then when AMD was at near 50% market share it merely posted 19M of profits across 3 financial years! In the same time frame, despite terrible performances, Fermi architecture already going into the 400w consumption range, even then Nvidia posted billions of profits across the same time frame.
AMD has simply stopped competing with Nvidia on prices, it's nonsense to them.
Nowadays both Nvidia and AMD sell more GPUs to datacenters and super computers, where net margins are also twice as high than they do to gamers.
Essentially both AMD and Nvidia are abandoning the gaming aspect of their hardware, stopped competing price-wise, instead they prefer both to sit on higher margins in the gaming market and focus on GPGPU aspect for business, research, automotive, AI, etc, etc, where the real money and growth is.
The fact that this hardware is also able to crunch rendering calculations is more and more of a side effect.
Perhaps even "defective" (ex failed PCIe lanes) in some way that makes them useless. I always figured this was a "use the whole pig" situation.
https://en.wikipedia.org/wiki/Frontier_(supercomputer) https://en.wikipedia.org/wiki/LUMI
There's lots of demands for AMD hardware in datacenters and supercomputers.
AMD and Nvidia sell the same thing made at the same place in Taiwan.
Nvidia dove in like 1998. AMD sold it's moat in like, 2005.
That being said, they're just way, way behind. Nvidia can charge what they do simply because AMD isn't great at making GPUs. In gaming AMD never caught up to Nvidia's raytracing perf, and AMD's attempt to clone DLSS is significantly less performant. Meanwhile in AI/ML the industry consolidated around CUDA because OpenCL was never that great, and — Nvidia owning the standard, and having built around it for over a decade, gives them a really big advantage. Plus, every GPU sold in the last decade+ by Nvidia supports CUDA, whereas only the new server GPUs from AMD support ROCm.
Nvidia has executed incredibly well for decades: they pioneered shader programming, CUDA, raytracing, and DLSS. Arguably they even invented the GPU itself with the GeForce 256 in 1999 (as opposed to earlier "accelerator" cards). I struggle to think of anything AMD has done in the GPU space that matches even one of those accomplishments... Their only slight win is having OSS drivers for Linux, but honestly, that doesn't sell a whole lot of GPUs.
AMD caught up to Intel on the CPU side in part thanks to Intel's failures. Nvidia just hasn't fumbled.
Edit: Also, Intel is trying to edge their way into the GPU market as well with their Arc line of discrete GPUs, which is relatively affordable. They compete well on price, especially at the lower end: the A750 and A380 obliterate anything at their price points ($239 and $119, respectively). But, well, "at their price points" is a pretty critical part of that sentence. Their top-tier card, the A770, can barely eke out wins against the lower end of Nvidia's last-gen cards.
And re: 2008-2011 ATI comparisons from a sibling comment — by 2010 Nvidia's Fermi architecture was reliably beating ATI again, albeit with higher power draw, and in 2007 they had just launched CUDA which ended up being the most lucrative market segment for GPUs. In retrospect I think Fermi was much less of a flop than it seemed (and it was a fairly brief one compared to Intel's CPU woes).
Consumers need to wake up and smell the underlying meta[3].
[1] https://www.bls.gov/opub/btn/volume-12/measuring-shrinkflati...
[2] https://www.bls.gov/opub/mlr/2023/article/automotive-dealers...
The difference is AMD actually discounts old GPUs and CPUs, a 6950XT can be found for around 500$ or less.
Samsung is an enormous conglomerate of which Samsung Electronics is only one of many sub-companies.
The article (and the accounting world in general) misleadingly uses words like "loss" to mean "not as much as the last time we measured," unlike how most laypeople interpret the word to mean "expenses exceeded revenue so we are in the red."
In fact they are still in the black, only much less so than before. You can't "lose" something that wasn't yours to begin with.
Samsung Electronics reported a 95 per cent drop in first-quarter profits on Thursday, as tech demand remained sluggish in the face of inflation and higher interest rates.
The Korean semiconductor giant’s operating profit fell to Won640bn ($478.6mn) from Won14.12tn in the first three months of 2022, amid a sharp drop in global demand for memory chips. The results were consistent with estimates released earlier this month, when Samsung said it would make a “meaningful” cut in memory production to ease a glut.
On Thursday the company said it expected a “limited demand recovery” as major cloud service providers invest more conservatively and customers adjust their inventories.
Revenue 95% plunge would be somewhat different!
Margin 95% plunge would too.
In any case, hopefully next quarter their losses can crater by 2000%, i.e., go from a small profit to a larger one and get back to about where they were.
Samsung Electronics is heavily involved in the cyclical semiconductor business which is a business infamous for violent downswings and they seem to be taking a beating from the memory glut.
This entire article is devoid of any synthesis or value added information. If I saw this on DNYUZ I wouldn't give it a second look, but CNN has lesser aspirations .
If you look at the byline in this CNN article, you'll notice it says "Reuters".
Reuters offers a "news wire" service with articles of generic factoids and syndicated to various news outlets like CNN: https://en.wikipedia.org/wiki/News_agency
By design, news wire articles (business, politics, etc) don't have value-added synthesis or insightful analysis because the generic content is created quickly for fast dissemination.
It's not an "opinion piece" like a Matt Levine article or The Economist editorial/analysis.
Future Demand modeling doesnt really take collective availble attention into account. I think we are going to see demand mainly driven by machine to machine requiremnts. The human requirement are probably peaking.
Demand for chips will continue to grow (historical growth is about 5% a year) but I don't think the market can support a dozen new fabs around the world without taxpayer money. that's just too much supply and extremely expensive insurance for "national security".
Gee, we switched to manufacturing high end parts nobody buys ... and people stopped buying our wares, what to do indeed.
When a user is expected to have to repurchase a product in the future (e.g. phone upgrade), do they look at it as squeezing as much revenue out of a single user whilst they own a single device, or over lifetime ownership of many devices?
You can imagine a scenario where they squeeze an extra 20% revenue whilst that user owns their device (great for this quarter) at the cost of 100% of future revenue when the user moves to another brand on renewal.
If the average phone upgrade is 21 months [0], then this feels like a short enough timeframe for this to be felt in the numbers.
0: https://www.counterpointresearch.com/smartphone-users-replac...