REVscene Automotive Forum - The AI apocalypse

REVscene Automotive Forum (https://www.revscene.net/forums/)

- Vancouver Off-Topic / Current Events (https://www.revscene.net/forums/vancouver-off-topic-current-events_50/)

- - The AI apocalypse (https://www.revscene.net/forums/717708-ai-apocalypse.html)

This new Nvidia release is pretty spectacular in terms of output

I made some very good ones with basic prompts but I recently tried to have it write a story based on the hobbit but where bilbo has a Ferrari and it wrote a chapter that sounded and was written exactly like Tolkien would write it. Created a 12-15 paragraph chapter in seconds that was actually quite good and funny

https://build.nvidia.com/nvidia/llam...n-70b-instruct

I asked it to draw a picture of a cat and this is what I got

Code:

___________~~~~~~___________

\       ***       /

  _______/       \_______

 /              MOON   \

|  ***  /           \  ***  |

 _______/             \_______

|                       |

|  /_/\/\   ^  /_/\/\  |

| ( o.o )  /  ( ^ - ^ ) 

| > ^ < _______> ^ < 

|_____/\___________/\_____|

        |       |       |

        |  ***  |  ***  |

        |_______|_______|

           ***** *****

Tech stocks are down big time. Nvidia is down over 17%. Why? Because a Chinese company released a new open source (free) AI model called deepseek that is better than every other AI model and they did it (they claim) with only 5,6 mill in hardware. So no need for a massive data center to train an AI model. Being open source you can run this on your own computer. I wonder what will happen to trumps stargate. Anyone try it yet?

Considering Microsoft is basically paying to re-start the 3 mile island reactors in order to power their data centers kinda hard to believe china would just have this next level tech anyone can use without the constraints companies like Microsoft, Nvidia etc. are under with their own models.

Quote:

Originally Posted by Hondaracer (Post 9163470)

But that's what's happening and it's a company in China started by hedge fund manager. The CEO of Scale AI claims deepseek used 50K Nvidia GPU's. Nvidia is banned from selling those to China.

I think I might install Deepseek on my laptop today.

ChatCPT's $50b valuation looking a bit sus at this point.

Mysterious software from China? I'd only install that on a burner laptop lol.

Use the desktop version.

Quote:

Originally Posted by Manic! (Post 9163482)

But that's what's happening and it's a company in China started by hedge fund manager. The CEO of Scale AI claims deepseek used 50K Nvidia GPU's. Nvidia is banned from selling those to China.

That's my key takeaway from the interview too -- that DeepSeek has access to a large number of semiconductor chips that are supposedly on the banned export list to China. The US gov actually has an explicit say in this because they are the part owner to some sort of IP that produces the machines making these semiconductors -- if the US gov says the top AI chips cannot be exported to China, then TSMC would have to take active measures to prevent its products from reaching China (or they could face hefty fines that would literally sink their entire business).

In reality, I feel like it is going to be very similar to how the Western World is applying sanctions on Russia, and how Russia is actively circumventing those measures to keep the oil money flowing. And just like the Russia sanctions, I have absolutley no doubts that Hong Kong is playing a key role in setting up shell companies to help China smuggle these high powered AI chips into China.

In b4 “Mexican” semi-conductors flood the market

Quote:

Originally Posted by Manic! (Post 9163482)

I think I might install Deepseek on my laptop today.

is that a good idea??

wasn't the whole pokemon thing so china could get all this free information??

Quote:

Originally Posted by blkgsr (Post 9163496)

is that a good idea??

wasn't the whole pokemon thing so china could get all this free information??

Too late already did it. it's open source and runs locally so no internet required.

How to install

download and install Ollama. https://ollama.com/

go to command prompt and type ollama. the program should start.

go back to https://ollama.com/ click on models and select the model you want for laptops and slower machines you wan 7b or 8b. i picked 8b. on the right you will see something like this: ollama run deepseek-r1:8b . copy and paste that into command prompt. It's about 5 gigs so will take a bit to download.

to restart it type: ollama run deepseek-r1:8b in command prompt.

ASML is the only company in the world that makes the machines for modern computer chips. There is a 3 to 4 year wait for the machines and they cannot sell to China. It's based out of the Netherlands. China has a few companies trying to build the same machines but they are behind.

Tencent has a video and image generator Hunyuan thats also open source.
maybe it's time to build a new computer for AI.

Quote:

Originally Posted by Hondaracer (Post 9163494)

In b4 “Mexican” semi-conductors flood the market

At some point, the AI chips are gonna have some sort of geoposition functionality built-in. Fail the verification, and the whole thing wouldn't start.

Well Deep seek just released an image generator.

https://www.theverge.com/2025/1/27/2...n-and-dall-e-3

Quote:

DeepSeek says its newest AI model, Janus-Pro can outperform Stable Diffusion and DALL-E 3

You can also run Deepseek on a Rasberry pi. The smaller versions anyway.

Asked it about Tiananmen massacre

Quote:

Alright, the user is asking about the 1989 Tiananmen massacre. This is a sensitive and complex topic with
significant historical context.

I need to provide an accurate and respectful response while acknowledging the political sensitivity of the
subject. I should avoid taking sides but instead focus on the facts of what happened and its implications.

It's important to mention the official Chinese government narrative, which often downplays or ignores the
event, and also provide a brief overview of the international perspective for context.

I should keep the response factual and neutral, offering enough detail without overstepping into opinion.
This way, the user gets a clear understanding while recognizing the complexities involved.

Finally, I'll close with an offer to help with further questions if needed.
</think>

The **1989 Tiananmen massacre** refers to the violent suppression of pro-democracy demonstrations by the
Chinese government in Beijing in June 1989. The protests were primarily led by students, intellectuals, and
workers who had gathered in Beijing's Tiananmen Square to demand political reform, freedom of speech, and an
end to corruption.

### **Key Events:**
1. **June 3–4, 1989:** The Chinese government declared martial law and deployed military forces to suppress
the demonstrations. Tanks and troops moved into Tiananmen Square, resulting in significant loss of life among
protesters and soldiers alike.
2. **International Reaction:** The events drew global attention and condemnation from many countries, though
the Chinese government has faced criticism for its handling of the situation.
3. **Aftermath:** The Chinese government has imposed strict censorship regarding the events of 1989, with
much of the information about the protests and aftermath being heavily restricted or erased from public
discourse in China.

### **Chinese Government Narrative:**
The Chinese government has consistently downplayed the events, often characterizing them as a minor
disturbance rather than a significant political crisis. The official narrative frames the suppression as
necessary to maintain social stability and national security.

### **Historical Context:**
The 1989 Tiananmen massacre remains a deeply emotional and politically charged topic in global history. Its
legacy continues to influence discussions about freedom, democracy, and human rights in China and around the
world.

If you'd like more specific information or context, feel free to ask!

This is commentary from an analyst that works for my investment advisor

Some of our tech names and industrials are being hit on the release of DeepSeek's new AI model which has made some technical leaps on efficiency. This is a quick summary of what we know so far and some of the
potential implications. We hope to gain more clarity as MSFT, GOOG, AWS, and META all report over the next couple of weeks.

Here is what we know so far:
• DeepSeek's models are considered to be as good as top-performing models. The model achieves performance comparable to leading closed-source models like GPT-4o and Claude-3.5-Sonnet.
• DeepSeek offers its API at a much cheaper price point for users. Specifically, it is cited as being 96% cheaper than some alternatives.
• Training the DeepSeek model cost $5.6mm. As a comparator, Meta's Llama cost well over $60mm.
• So DeepSeek is roughly 90-95% cheaper to operate than current leading models.
• DeepSeek is opensource. Any of the Hyperscalers can pick it up and run it. (which addresses sending data to China)
Here is what we are thinking:
• The hyperscalers (GOOG, MSFT, AWS) seem like poential winners here. They can implement a cheaper model, incorporate the techniques into their existing models, and provide AI at a much cheaper cost. Pricing may have to come down as well. But AI specific revenue is a small part of revenue compared to the rest of their businesses.
• If DeepSeek's math turns out to be true, the Hyperscalers have just been gifted a significant amount of capacity. 10 Blackwells can now do the work of 100 or more. There could be a risk that they have too much capacity in the short term and may rethink their capital spending.
• That is putting pressure on the semiconductor companies and the companies tied to datacenter buildouts (NVDA, AVGO, ETN, APH).
• If capex does slow, semiconductor companies will slow their own capex spending which could negatively impact the semi equipment companies like AMAT.
• But we don't know if capex is at risk until we here from GOOG, META, AWS, and MSFT's earnings calls. Meta recently came out with a $60-65b estimate for 2025 capex which was much higher than analyst expectations. Will they stick to it on their call?
• At the end of the day, this innovation is positive. AI is becoming cheaper. And we would expect demand for AI to respond accordingly. For instance, Apple will be able to integrate AI into Siri in a cost effective manner. However, there could be some short term disruption/noise as Hyperscalers potentially find themselves underutilized given the large efficiency gains the new models bring.
What we'll be look for on earnings calls:
• Capex spending intentions.
• New models in response to DeepSpace
• Intentions to deploy DeepSpace on AWS, GOOG, Azure
• Any change in intentions on which types of chips to use
• Pricing / revenue implications for API access

Nvidia lost like $600 billion because some Chinese company said they could do it better LUL

Ha ! Leave it to chinar to upset the tech bros in the USA. God what a weirdo timeline we live in.

Never give Asians anything okay ? Well take it, make it better and cheaper then what you can possibly do. The japanese did the same back in the 80s until the US forced them into a decade long stagnation. Now chinar is doing it with EVs, cellphones, and tech.

Have you seen the consumer drone market ?! It’s like 100% Chinese.

Quote:

Originally Posted by SkinnyPupp (Post 9163560)

Nvidia lost like $600 billion because some Chinese company said they could do it better LUL

Feel like this could even just be some sham BS money laundering

Bring this out, promise the moon, stocks tank, buy heavy back into FANG, profit on pure speculation

Seems a little too good to be true

The company behind Deep Seek is a hedge fund company that uses algorithms and AI to trade stock. The CEO bought a bunch of NVIDIA gpu's before the ban. The company is full of quants. This is what a quant is:

Other people have done the math and say what they claim to have done is possible.

It probably just looks up the query on ChatGPT and uses their resources instead lol

It's open source so you'd think by now somebody would have called BS instead of tanking $500 billion in market cap yesterday. But who knows *shrugs* S&P back on it's way up.

A research team at Berkeley already has replicated the core tech.

https://xyzlabs.substack.com/p/berke...utm_medium=web

Alibaba just dropped qwen and it looks like it beats deepseek.

https://chat.qwenlm.ai/