DeepSeek implications for DePIN: why open-source efficiency favors io.net
Quote374 words
TL;DR of his DeepSeek article: the $5M training cost claim is doubtful, but efficiency gains help open-source AI and demand for decentralized compute. io.net, which owns no GPUs, has little H100 depreciation risk unlike DePINs that bought hardware.
Key claims
DeepSeek's rumoured $5M training cost is likely marketing
Cheaper models raise demand for decentralized compute
DePINs that bought H100s face margin risk if mid-tier GPUs suffice
Will DeepSeek change AI forever? And what does it mean for DePIN infra plays?
DeepSeek’s rumoured training cost of only $5M has sent shock waves through the US tech sector and crypto markets.
In my article I dive deep into the implications of it all. Keep reading here for shorter TL;DR.
So what is DeepSeek?
DeepSeek is a pretty performant LLM built by a Chinese AI quant fund. It’s fully open-source and that can be run at a great price.
What about the training cost?
The $5m total training cost part of the story is hard to believe and is probs just (highly effective) marketing. The fact that various optimisations to make it much cheaper to run, while maintaining high performance, is a big milestone for open-source AI though and likely a net positive for the industry (as well DePIN and crypto more broadly).
DeepSeek definitely doesn’t render everything else obsolete tho and the long-term impact on the AI industry will remain measured.
However, DeepSeek does exemplify the drive towards cheaper training/inference setups not dependant on scarce top-tier GPU resources from which many smaller companies or independent researchers are priced out from in the first place.
By providing builders a cheap to operate, yet highly performant model that can be run on local infrastructure and fine-tuned to serve specific use cases, the demand for decentralized compute infrastructure like @ionet’s GPU network, will likely even see an uptick in demand.
Especially as @ionet is uniquely positioned as a marketplace for efficient compute resource allocation, that does NOT own its GPUs, there is also little downside risk from top tier GPUs like Nvidia’s H100 losing demand and ultimately value.
With other DePIN networks that have spent millions to acquire such GPUs, that’s different, as a shift towards models that can be run on mid-tier GPU infra, could impact margins and ultimately even the viability of their business models.
So imo quite bullish for well designed DePIN protocols such as @ionet, more of a risk for others, but overall a net positive for our industry and a catalyst for open-source AI.
Check full article for more in-depth information on the tech that underlies DeepSeek, context on the cost difference to other models, and the reported performance benchmarks.
QUOTING @zerokn0wledge_
t.co/DrC2hzna5e
t.co/DrC2hzna5e
How the call played out · since 30 Jan 2025
Prices as of 23 Oct 2026This piece takes a view on $IO, so that move is read against the call.
−94%
Moved against the call · over 21 months
- At publication
- $2.68
- Now
- $0.159
- vs BTC
- −92%
- After 1 month
- −54%
- After 3 months
- −70%
- After 1 year
- −95%
- High since
- 0.0% 30 Jan 2025
- Low since
- −96% 06 Apr 2026
Since then
Zero has written 16 more pieces on io.net and $IO since this one.
- io.net IDE tokenomics: #1 revenue DePIN at 5.3x MC/ARRBullish call on $IO
- DeAI as crypto's endgame: compute, agents and autonomous economiesOn io.net
- Why decentralized compute solves the startup GPU cost squeezeOn io.net
- Big compute is cracking: summary of the DeAI case against centralized cloudOn io.net
Daily closes in USD from DefiLlama and Yahoo Finance. Only a bullish or bearish thesis is read as a call; explainers, maps and mentions are context. Not financial advice.