Comment by lifthrasiir
7 hours ago
Note the first line:
> SoftBank said Tuesday it has sold its entire stake in U.S. chipmaker Nvidia for $5.83 billion as the Japanese giant looks to capitalize its “all in” bet on ChatGPT maker OpenAI.
They are switching gears, not exiting, folks.
I would have thought that the OpenAI bet is way more risky, because if someone comes along with a better model it could really hurt OpenAI. NVIDIA seems harder to dethrone imo.
They probably know more than us. Such as alternative chips or that the Chinese will go in-house sooner than we think. Nvidia’s moat is not as permanent as people think.
> They probably know more than us. Such as alternative chips or that the Chinese will go in-house sooner than we think. Nvidia’s moat is not as permanent as people think.
This is Softbank though, they're not really noted for great investment decisions (apart from Alibaba really early on).
Like, my prior is that when Softbank invest in something, the growth is done (but then I am, very much, a cynic).
1 reply →
Not to mention the US companies probably want to develop their own chips to get away from Nvidia too. I don’t really see why any company would want to base their entire business on the whims of nvidia.
Given how poorly Intel and AMD have fared, it suggests the hardware moat is much deeper than the software one.
2 replies →
In many ways OpenAI is transitioning towards an end-user facing product business. They have by far the strongest brand among consumers and are positioning themselves to take on Google/Meta in the ad business.
By proxy, having the strongest frontier model becomes less and less necessary for them and instead building a strong product by properly layering medium-strong models in a cost-efficient way is the priority.
I'd argue their brand might be too strong, ChatGPT has already begun to enter the same semantic space as "Velcro". Everyone I know seems to have tried it yet quickly you begin to realize that for most people ChatGPT == LLM, it seems everyone is using "ChatGPT" on completely different platforms.
In the end, regardless of technical understanding, people will always shop around on price if the feature set is similar enough I suppose.
> They have by far the strongest brand
Just like Blackberry, but instead of having to buy an new device to switch to the competition, the customer just needs to click on a different website.
1 reply →
This argument regarding brand loyalty gets repeated but it’s really weak to me. The immense majority of people don’t tie their identity to the software services they are using. Without network effects or an ecosystem locking customers will switch as soon as there is a less expensive and/or better alternative, as the history of software has shown countless of times.
4 replies →
> They have by far the strongest brand
They don't really, because of Gemini and Grok.
If it was OpenAI vs Claude then yeah, ChatGPT is known whereas Claude you have to be an enthusiast to know.
But everyone using Google runs into Gemini and everyone using X runs into Grok, and people talk. Now even the laypeople know there is more than one AI, and they're from big brands that they trust. Which means that people will window shop for the AI with the best performance per cost. Bye bye brand power.
2 replies →
It’s going be pretty hard to dethrone Google when even Apple is going to Google for AI with another huge billion $ contract.
2 replies →
It's hard to see how they become profitable enough to justify current valuations.
The numbers are just mind boggling even in the optimistic scenario.
4 replies →
It's SoftBank though, weird risk taking investments seems to be their jam.
That's how you get to have $6B to throw around on just one part of one your strategies though to be fair.
It's because they are playing with other (dumb) people's money.
Yea, I was thinking the same thing.
Probably so, but that doesn't mean their value can keep scaling without heavy diminishing returns. Softbank must assume they've taken 80%+ of the gains from this phase of NVIDIA's growth, and want to capture the next wave of growth.
I agree with you that OpenAI seems much more risky in terms of it's actual true viability as a business, but the risk:reward must be there for Softbank.
Without a doubt, in my opinion, OpenAI is a risky bet, but perhaps Softbank can make some money on OpenAI and then get out.
They want to profit from the IPO of OpenAI. Private investors get a free 20% - 30% gain not available to the retail investors.
Selling the _entire_ stake sounds really aggressive though? Is that normal?
Even if you're all-in on OpenAI, does it not make financial sense to have _some_ stake in Nvidia considering they are the only ones with an actual moat?
Unless there are CUDA alternative breakthroughs we will hear about in the next few days.
Amazon’s Idaho 2Gw data center it’s building for Anthropic has 0 Nvidia Cuda cores.
https://finance.yahoo.com/news/amazon-says-anthropic-will-us...
> Anthropic is using 500,000 of Amazon’s Trainium2 chips for its Claude AI models as part of the tech giant's Project Rainier
For researchers and academics, cuda is painful to avoid, but I'm not sure that it is for large companies, once the time comes to train and deploy large models.
Isn't it merely a matter of time before China has an alternative?
All you have to look at is ASIC miners. Once they had them, they were 10x faster than GPUs easily and made GPUs useless for those algos. Something very similar can happen soon.
3 replies →
They don't even need to invent anything. Just put in some work to polish what already exists: https://github.com/vosen/ZLUDA
2 replies →
lol, after wework you'd be an idiot to think it was nothing more than shilling their "winner" to potentially dump on the public. but i guess as long as you can make money on the way up.
They recognise that the larger bubble is in the datacenters.
Most of the hardware we are using was designed for computer graphics not AI. Now that China isn't buying Nvidia any longer and actively trying to get their own companies to produce hardware, what happens to all these datacenters when a company produces a device that has 80% of the performance of the current Nvidia hardware but 20% of its power consumption?
Even double the energy consumption is not that bad for half the price. at 20 cents per kWh over 5 years 1 kW load would be 8760. So from 30000 to 15000 you would still come ahead in cost.
Softbank is involved with Lutnick's Cantor & Fitzgerald in crypto deals:
https://www.reuters.com/business/media-telecom/cantor-nears-...
Masayoshi Son is BFF with Trump and Lutnick:
https://asia.nikkei.com/politics/international-relations/how...
There is more going on here than is publicly known. Investing in hot air is of course a specialty of the Trump admin.
Maybe they're trying to bail out OpenAI, but that's still not close to the amount they would need to do that. OpenAI needs like 1.3 trillion dollars.
the key is "looks to"
Softbank can also choose not to.
So averaging up then? That's even worse. More like degeneracy and gambling.