Project hub

Nvidia

20 pieces · May 2024 to Oct 2026 · 3 flagship

Sectors

AI20 DePIN15 Privacy2 Robotics1 DeFi1 L1/L21

Search every mention All projects and tickers

Coverage per year

132024

52025

22026

Start here · the strongest pieces on Nvidia

Breakdown30 Jul 2024Aethir deep dive: decentralized GPU cloud for AI and gamingWalks through Aethir's architecture (containers, checkers, indexers), ATH staking security and compute reward model, arguing decentralized GPU supply can counter big tech's concentration of compute. Cites AI adding over $15T to the economy by 2030.Article30 Jan 2025DeepSeek: Implications for DePIN and the broader AI SectorLong-form article on DeepSeek R1's claimed sub-$5M training cost, its MoE and sparse activation design, and what it means for GPU scarcity and decentralized GPU networks like io.net. Covers closed vs open models and DePIN challenges.Thesis21 Aug 2024NEAR AI thesis: what Jensen Huang's touch signals31-part thread arguing NEAR is a leading crypto x AI base layer, combining sharded low-fee execution, chain signatures for agents acting across chains and the NEAR AI lab pursuing user-owned AGI. Expects GPU DePINs and agent apps to build on it.

Timeline · every piece, newest first

2026 2

23 Oct 2026Arcium Blackthorn supply gap: 600 to 4,500 rentable confidential GPUs worldwideArcium's survey of about 80 providers, cited in Four Pillars' Blackthorn deep dive, counts only 600 to 4,500 rentable confidential GPUs, about 0.04% of capable NVIDIA chips. Blackthorn drops the host enclave and binds the GPU to the live key exchange; Charité Berlin is reportedly its first production partner.ThesisPrivacyAI6 likes

14 Sep 2026SimArena at robo.innovate: why a browser simulator matters for CodecSimArena runs physics in WebAssembly (no GPU, MIT-licensed, 500 free credits a month, LeRobot export) and sits in TUM's robo.innovate alongside Nvidia, Universal Robots and NEURA. Every student building on it becomes a user.ThesisRoboticsAI20 likes

2025 5

01 Aug 2025Theoriq playbook: Alpha Protocol, AlphaSwarm and THQ for agent-driven DeFiBreaks down Theoriq's two layers (Alpha Protocol registry and AlphaSwarm multi-agent liquidity management) and the THQ stake, delegate and slash model. Notes the Kaito launchpad sale was heavily over-subscribed ahead of TGE.BreakdownAIDeFi29 likes

21 Jul 2025Big compute is cracking: summary of the DeAI case against centralized cloudSummarizes an article arguing AI demand is outrunning inelastic centralized compute: complexity doubles every 3-4 months, data center energy may exceed 945 TWh by 2030, and a 1,000-GPU cluster costs $2.8M monthly. io.net aggregates >35,000 GPUs at up to 90% lower cost, with zkML as the trust layer.BreakdownAIDePIN15 likes

03 May 2025TEEs explained: isolated compute for privacy-preserving AIExplains Trusted Execution Environments as isolated CPU zones shielding computation from the host OS, already deployed in major clouds and extending to GPUs like NVIDIA's H100, which is crucial for privacy-preserving AI and utility agents.ExplainerPrivacyAI26 likes

30 Jan 2025DeepSeek implications for DePIN: why open-source efficiency favors io.netTL;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.ThesisAIDePIN62 likes

30 Jan 2025DeepSeek: Implications for DePIN and the broader AI SectorLong-form article on DeepSeek R1's claimed sub-$5M training cost, its MoE and sparse activation design, and what it means for GPU scarcity and decentralized GPU networks like io.net. Covers closed vs open models and DePIN challenges.ArticleAIDePIN30 likes

2024 13

12 Sep 2024io.net vs Nvidia: a $233B daily move equals 179x IO FDVFrames io.net as a way to access scarce institutional GPUs at low cost, noting Nvidia's one-day $233B market cap swing equals 179x IO's FDV. Argues IO is undervalued.Market takeDePINAI33 likes

24 Aug 2024The race to AGI is a battle for compute, and how DePIN can counter itArgues AGI development is bottlenecked by finite compute, with Nvidia at about 80% of GPU supply and big clouds absorbing most of it. Says geo-distributed clustering by DePINs like io.net can even the playing field.ThesisAIDePIN6 likes

21 Aug 2024NEAR AI thesis: what Jensen Huang's touch signals31-part thread arguing NEAR is a leading crypto x AI base layer, combining sharded low-fee execution, chain signatures for agents acting across chains and the NEAR AI lab pursuing user-owned AGI. Expects GPU DePINs and agent apps to build on it.ThesisAIL1/L224 likes

09 Aug 2024io.net and Leonardo AI: GPU partnership and DePIN demand caseArgues $IO is underappreciated after io.net expanded its Leonardo AI partnership to add L40S GPUs alongside A100s. Cites over 318,000 verified GPUs and 28,000 cluster-ready GPUs against enterprise cloud spend above $79B in Q2 2024.ThesisDePINAI107 likes

30 Jul 2024Aethir deep dive: decentralized GPU cloud for AI and gamingWalks through Aethir's architecture (containers, checkers, indexers), ATH staking security and compute reward model, arguing decentralized GPU supply can counter big tech's concentration of compute. Cites AI adding over $15T to the economy by 2030.BreakdownDePINAI39 likes

27 Jul 2024Tokenized GPUs as an RWA: Compute Labs GNFTs on io.netArgues tokenizing H100 GPUs as fractionalized GNFTs gives retail income-generating exposure to AI compute beyond NVDA stock, while decentralizing io.net's supplier base.ThesisRWADePINAI25 likes

21 Jul 2024AI market sizing and the DePIN GPU case: io.net and AethirCites the AI market at $184B in 2024 heading to $826B by 2030 (Statista) and UBS's $420B 2027 revenue forecast, then argues io.net and Aethir can win GPU demand. Aethir commits 1,000 H100s to io.net, which has about 28,000 cluster-ready GPUs.ThesisAIDePIN46 likes

18 Jul 2024io.net deep dive: GPU aggregation, Aethir H100s and network metricsDetails io.net's aggregation of consumer, datacenter and Aethir H100 capacity, with 28,000 cluster-ready and over 294,000 verified GPUs. Cites 378,000+ inferences, 1.9M onchain transactions and about $5.25M in block-reward revenue to providers.BreakdownDePINAI74 likes

17 Jul 2024Cloud compute market sizing and why Aethir fits the GPU shortageSizes cloud infrastructure (AWS 31%, Azure 25%, Google 11%, over $400B annual by year-end) and ties GPU scarcity (Nvidia 80% share, 4-11 month delivery) to the case for Aethir, which reports $36M ARR and a $30B+ TAM.ThesisAIDePIN40 likes

06 Jun 2024Distributed GPU computing and io.net: House of Chimera report TL;DRDigests the House of Chimera GPU computing report: AI demand versus GPU supply, consumer versus datacenter GPU economics and io.net's staking model. Adds his own caveat that utilization is only 1-5% versus Akash's 25%, so demand must catch up with supply.BreakdownAIDePIN25 likes

30 May 2024Decentralized GPU networks in AI: io.net case study on training vs inferenceExplains the difference between AI model training and inference, then applies it to io.net's GPU supply. Compares io.net H100 pricing at $1.19 per hour against AWS at about $3.06 and Google Cloud at about $2.48.ExplainerAIDePIN23 likes

25 May 2024io.net H100 GPUs via Aethir at $1.19 per hour vs AWS and Google CloudExplains what Nvidia H100 GPUs are and prices io.net's Aethir-supplied capacity at $1.19 per hour against AWS and Google Cloud A100 pricing. Notes the A100 comparison is not like for like.ExplainerAIDePIN88 likes

15 May 2024What 1,000 H100s mean for io.net: GPUs vs CPUs and training capacityExplains GPU versus CPU architecture and what the H100 is, then sizes the 1,000 H100 addition to io.net: about 18.4M CUDA cores and 80TB of memory, enough to train models in the hundreds of billions of parameters. Notes GPT-3 training on 256 A100s took weeks.ExplainerAIDePIN88 likes

Often covered with

io.net71 Aethir26 Google Cloud8 AWS9 DeepSeek10 OpenAI13 Render18 Arcium26

Working together

Research, campaigns and builds for deep-tech teams.

View CV A1 Research