DeepSeek FAQ @ Stratechery said:
DeepSeek FAQ
DeepSeek has completely upended people’s expectations for AI and competition with China. What is it, and why does it matter?
stratechery.com
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Budget AI Model DeepSeek Overtakes ChatGPT on App Store
- Thread starter MacRumors
- Start date
- Sort by reaction score
stratechery.com
This sort of development explains why Apple is smart to not jump into the LLM craze. We are still in its infancy and it’s way too early to declare which company has won or lost. Better to offer the OS that plugs into the LLM of choice and watch the dust settle while the rest fight it out amongst themselves.
In other news, seems like nobody is using Google Gemini.
In other news, seems like nobody is using Google Gemini.
It’s open sourced, so other people can freely build their own setups based on it. Not like it proprietary and reserved by any country exclusive ownership. There are many examples of open sourced being supported worldwide. This Sputnik moment caught an industry that thought all you have to do is spend billions on AI. If that was a tongue in cheek post excuse the explanation.
Last edited:
Thanks. Sucks as I live in Canada and he died back home in Australia and I can’t get back there to help my mum out. Feels surreal and guilt ridden.
Thank you.
It’s impossible now to try to live cash based, although I have one friend that still tries to. I run my own company and I use quickbooks online and receive all money digitally. It kind of sucks.
That’s the whole thing because of the chip ban it’s built on lower power and older chips. In fact US actions have forced the Chinese to come up with a more efficient LLM than the US can offer. Kind of ironic.
And giving China more free data of yours. Which is how they were able to do this dirt cheap originally
What and you think the original LLM didn’t steal vast amounts of data from across the web, much of it or people’s creative work. So it’s a nice irony that now they’ve had their work made use of.
I wonder if a lot if the companies who jumped in early on this are going to be the ones that crash and burn longer term.
The Q4 versions would be quantized from your screenshots. I'm able to run those, can even get the larger versions to run, they really get the fans going, but it is runnable.
I was talking about the full, large model. But the q versions definitely still work.
EDIT: Ok after looking closer, I'm not sure if that's the full model, it does show the large params. But is that the current R1?
OpenAI and MS? Sure, they've got competition now. But MS fell 2% and Nvidia fell 17%. They only way anyone is going to compete here is by doing more research and building better models-- all of which is going to require cutting edge hardware. Sure, DeepSeek did the job with 2,000 rather than 20,000 GPUs, but whoever is trying to top them is going to use their 20,000 to train 10 models.
DeepSeek succeeded with less resources, but those with the resources to throw at it will use their advantage.
Whenever there's a war, the weapons manufacturers make money.
a) you can run at home, but moving forward, surely a lot of consumer stuff won’t be on-device?
b) more complex professional/industrial uses, as per current supercomputers, surely rely on big data centre?
c) surely they don’t just “train” once, but are constantly updating?
d) server farms for cloud computing generally.
How data centers and the energy sector can sate AI’s hunger for power
The growth of data centers and AI rely on the availability of electric power. Opportunities for investors in power infrastructure and adjacent sectors are quickly emerging.
The average energy requirement of a query to ChatGPT is estimated at 3 to 9 watt hours (https://www.roboticstomorrow.com/st...en-depend-on-the-size-of-the-ai-model-/24021/)
That’s not insignificant, if you scale it up by the number of requests happening globally ever minute, especially as Google, Bing, Amazon et al start integrating AI responses by default into search results.
I’m no expert. But this has just been a big topic of conversation at the World Economic Forum, so it must be a “thing”.
AI's energy dilemma: Challenges, opportunities, and a path forward
AI’s energy demand from data centres is projected to grow but AI also helps reduce energy use significantly; coordinated efforts can manage the tension.
www.weforum.org
Chinese artificial intelligence startup DeepSeek could face curbs from the U.S. government as it upends the U.S. AI ecosystem, though enforcing restrictions on an open-source technology could be a challenge, experts said.
DeepSeek’s sudden rise has questioned the effectiveness of Washington’s efforts aimed at curbing China’s access to high-end tech over national security concerns.
For DeepSeek, which relied heavily on open-source code, the additional export restrictions that the U.S. government could impose are limited, said Lawrence Ward, a partner at U.S.-based law firm Dorsey & Whitney with a specialization in national security law.
In terms of the company’s potential use of certain Nvidia chips, it might face civil and criminal penalties, but enforcing those penalties may be “difficult if not impossible,” Ward said.
On Monday, U.S. lawmakers called for actions to slow down the Chinese tech startup, with some calling DeepSeek “a serious threat.”
There are no easy solutions to restrict the use of an open-source model, especially one that is being widely tested and used by organizations and individuals, said Paul Triolo, partner at Albright Stone Group.
One option would be for the Commerce Department to craft rules that require tech giants such as Apple and Google to take down DeepSeek’s app, restricting its downloads in the U.S. market, Triolo said, adding that it would, however, be challenging to pull the app off other platforms such as Github.
Last edited:
DeepSeek just blew up the AI industry’s narrative that it needs more money and power | CNN Business
The story of AI in the 2020s has gone something like this:
www.cnn.com
New YorkCNN —
The story of AI in the 2020s has gone something like this:
- Sam Altman: Look, a toy that can write your book report.
- VCs: This will fix everything!
- Doomers: This will ruin everything.
- Tech: We need money!
- Everyone else: Could we maybe not destroy the environment over this?
- Tech: Let’s restart Three Mile Island.
- Tech: We need money!
- Wall Street: Where’s our return?
- Tech: (Chants) More power! More power! More power!
Suddenly, all that money and computing power that the Sam Altmans, Mark Zuckerbergs and Elon Musks have been saying are crucial to their AI projects — and thus America’s continued leadership in the industry — may end up being wildly overblown.
Consumer, no.
More likely that all enterprise will run their own private compute in house for privacy and security
No big models are read only. You can tune a model and you can add RAG for up to date information. Mostly they're using agents now to add things like real time news from the web. But the base model just stays as is it.
It's not insignificant but it's about 10x each Google search. The thing is one ChatGPT request often gets what you need to know where as you might be searching Google hundreds of times a day for any and everything - the usage changes somewhat.
The blockchain certainly uses a lot more energy - and mining for crypto exponentially more energy.
As LLM's become more ubiquitous into everything we run from computer games (albeit tiny models, like we have on our iPhones now) to massive models so will the hardware that runs it in efficency. In ten years the nm of the processors will be reduced and we'll be able to run the same commands at 1/100th of the electricity costs - and in the mean time all data centers running it should be carbon neutral and where possibly generating their own electricity from solar and wind anyway.
There's a small part of me of course that wonders if some cards might be back doored out of Taiwan...
That’s a very valid point which is why I’m not a huge fan of AI from many perspectives. It makes things better but as a creative professional it’s a very fine line to walk with stealing creative works for the sake of training AI. Sadly the faster this all moves the less protections that will be in place and it will be a massive cleanup disaster from a legal perspective down the road.
Regarding “everyone else”. That’s pretty loaded. The same people telling (brainwashing) that the world will come to an end soon because of climate change are largely benefitting from all of this money being funneled around. As well using their private jets and supporting foreign countries that produce the worlds products which cause infinitely more carbon emissions.DeepSeek just blew up the AI industry’s narrative that it needs more money and power | CNN Business
The story of AI in the 2020s has gone something like this:
New YorkCNN —
The story of AI in the 2020s has gone something like this:
And finally, in the year 2025, here comes DeepSeek to blow up the industry’s whole narrative about AI’s bottomless appetite for power, and potentially break the spell that had kept Wall Street funneling money to anyone with the words “harnessing artificial intelligence” in their pitch deck.
- Sam Altman: Look, a toy that can write your book report.
- VCs: This will fix everything!
- Doomers: This will ruin everything.
- Tech: We need money!
- Everyone else: Could we maybe not destroy the environment over this?
- Tech: Let’s restart Three Mile Island.
- Tech: We need money!
- Wall Street: Where’s our return?
- Tech: (Chants) More power! More power! More power!
Suddenly, all that money and computing power that the Sam Altmans, Mark Zuckerbergs and Elon Musks have been saying are crucial to their AI projects — and thus America’s continued leadership in the industry — may end up being wildly overblown.
If it was really that dire things would be VERY VERY different. China for example doesn’t give a crap about any of it and they have the worlds largest carbon footprint.
TL;DR:
90+ percent of this is about money and only money. Unless every person stop buying literally everything from cell phones to clothing….you are part of the “problem” you’ve been brainwashed into thinking is gonna end it all. Is it a problem? Kind of?. Is the world gonna end tomorrow or next year or even next decade because we don’t have enough windmills and solar panels. NOPE!
Hold on thats not right at all. China is actively changing, they even call it the battle for blue sky, they are proactively reducing their pollution foot print, and making the air they breathe better and healthier. Remember they are also transitioning from gas to EV and they lead the market in solar power.
10 years ago, one could say China didnt care about pollution, not today. China changed because things were dire.