HN Distilled

Essential insights from Hacker News discussions

DeepSeek R2 launch stalled as CEO balks at progress

Original Article

Hacker News Discussion

Conflicting Reports on the Reason for DeepSeek R2's Stalled Launch

There is significant confusion and disagreement within the discussion regarding the actual R2 launch status. Initial reports from Reuters suggested the CEO was "not satisfied with its performance," but later parts of the article and subsequent analysis by users point to a lack of GPU capacity due to export restrictions as the primary reason. This discrepancy has led to skepticism about the sourcing and accuracy of the reporting.

  • "The title of the article is 'DeepSeek R2 launch stalled as CEO balks at progress' but the body of the article says launch stalled because there is a lack of GPU capacity due to export restrictions, not because a lack of progress." (teruakohatu)
  • "The article says this: 'June 26 (Reuters) - Chinese AI startup DeepSeek has not yet determined the timing of the release of its R2 model as CEO Liang Wenfeng is not satisfied with its performance,' But yes, it is strange how the majority of the article is about lack of GPUs." (Thorrez)
  • "I am pretty sure that the information has no access to / sources at Deepseek. At most they are basing their article on selective random internet chatter amongst those who follow Chinese ai." (chvid)
  • "The phrasing for quoting sources is extremely codified, it means the journalists have verified who the sources are (either insider or people with access with insider information)." (tonfa)
  • "First, nobody is training on H20s, it's absurd. Then their logic was, because of high inference demand of DeepSeek models there are high demand of H20 chips, and H20s were banned so better not release new model weights now, otherwise people would want H20s harder. Which is... even more absurd. The reasoning itself doesn't make any sense. And the technical part is just wrong, too." (rfoo)

Strategic Value of Open-Sourcing and Brand Building

A key theme is understanding DeepSeek's motivation for open-sourcing their models. While the immediate financial returns may not be the primary driver, participants suggest that open-sourcing is a strategic move for building brand recognition, attracting talent, and establishing a geopolitical and cultural presence. There's also speculation about whether this open-sourcing will continue or if DeepSeek will eventually shift towards a more proprietary model, similar to OpenAI.

  • "Releasing the model has paid off handsomely with name recognition and making a significant geopolitical and cultural statement." (teruakohatu)
  • "But will they keep releasing the weights or do an OpenAI and come up with a reason they can't release them anymore?" (teruakohatu)
  • "If they continue to release the weights + detailed reports what they did, I seriously don't understand why. I mean it's cool. I just don't understand why. It's such a cut throat environment where every little bit of moat counts. I don't think they're naive. I think I'm naive." (ngruhn)
  • "If you’re not appearing, you’re disappearing. Now they are firmly on the map, which presumably helps with hiring, doing deals, influence. If they stop publishing something, they run the risk of being labelled a one-hit wonder who got lucky." (senko)

Future Monetization Beyond Inference Pricing

Several users believe that DeepSeek, and similar AI companies, are not solely focused on making money through direct inference pricing for their models. Instead, the long-term vision is posited to involve leveraging advanced AI for other revenue-generating activities, such as conducting research that could lead to highly profitable ventures, or utilizing the collected usage data.

  • "I don't think any of these companies are aiming at long term goal of making money from inference pricing of customers." (Davidzheng)
  • "What is DeepSeek aiming for if not that, which is currently the only thing they offer that cost money? They claim their own inference endpoints has a cost profit margin of 545%, which might be true or not, but the very fact that they mentioned this at all seems to indicate it is of some importance to them and others." (diggan)
  • "Well it's certainly helpful in the interim that they can recoup some money from inference. I'm just saying that with systems with more intelligence in the future can be used to make money in much better ways than charging customers for interacting with it. For instance it could conduct research on projects which can generate massive revenue if successful." (Davidzheng)
  • "maybe they benefit from the usage data they collect?" (coderatlarge)

Skepticism Towards "GPU Capacity" as a Stated Reason

There is a recurring skepticism that the lack of GPU capacity is the genuine reason for the R2 launch delay. Some users suggest this explanation is a "bullshit" excuse, especially given that alternative deployment strategies exist, such as offering the model in other regions or enabling specific features. The possibility of waiting for domestically sourced Chinese silicon is also raised as a potential, albeit speculative, factor.

  • "The lack of GPU capacity sounds like bullshit though, and it's unsourced. It's not like you can't offer it as a secondary thing, sort of like O-3 or even just turning on the reasoning." (impossiblefork)
  • "maybe they’re just waiting to see if they can run on chinese sourced silicon? just speculating" (coderatlarge)
  • "R1 is the most anti-H20 model released ever." (rfoo)
  • "MLA uses way more flops in order to conserve memory bandwidth, H20 has plenty of memory bandwidth and almost no flops. MLA makes sense on H100/H800, but on H20 GQA-based models are a way better option." (terafo)

The Problematic Nature of Modern Journalism and Sourcing

A significant portion of the discussion revolves around the perceived unreliability of modern journalism, particularly concerning news from China. Participants express concern about media outlets fabricating or misrepresenting information for clicks, engaging in self-referential sourcing, and a general decline in journalistic integrity. This leads to a call for greater critical evaluation of news sources.

  • "This is why we need to be critical of journalists nowadays. No longer are they the Fourth Column, protecting society and democracy by providing accurate information." (FooBarWidget)
  • "Not just 'China news', unfortunately." (chrisweekly)
  • "That sounds to me like you are excusing a bad reality based on a nonexistent ideal. Saying 'there are bad journalists' is a huge understatement. There are many, perhaps even the majority. Ask yourself why society at large has stopped trusting mainstream media, it's not just because there are a 'few' bad apples but because the bad apples are widespread and systemic." (FooBarWidget)
  • "It also doesn't follow that my critique of journalists/journalism means tearing down journalism altogether. It can also mean that society as a whole needs to hold journalism accountable, and demand that they return to the role of the Fourth Column." (FooBarWidget)
  • "How it is written it could be 3 anonymous and random guys from Reddit who heard about DeepSeek online." (rvnx)