- Chain of Thought
- Posts
- Weekly AI Edge #14
Weekly AI Edge #14
Vana raises $25M to satisfy AI's data hunger
GM!
In this edition of our weekly AI Edge, we cover:
Bittensor leads the Crypto AI market with a 30% rise this week
AI’s getting ‘hangry’ for data!
Vana raises total funding to $25M
Our favourite tweets on X
BTW: We’ve been crafting our Crypto AI thesis for many weeks, and it's almost ready.
We explain why Crypto AI is the most compelling opportunity today, how the future might unfold, and ways to capture value.
It will be released in three parts. Part I will drop on Monday. But here’s a teaser for our loyal readers—more on this soon!
🦍 State of the Market..
Source: Coingecko
Alt season is here, baby!
The Crypto AI market cap surged to $31.5B this week, marking a whopping 35% increase (!!) over the last two weeks.
Crypto AI is one of the top-performing sectors this month, alongside memes, data availability protocols, and parallelized EVMs.
Leading the charge to Valhalla was TAO, skyrocketing 70% in just two weeks, fueling lots of Bittensor chatter and mindshare on Crypto Twitter.
Several tokens with market caps over $100M saw impressive gains: Delysium (AGI, +46%), Solidus Ai Tech (AITECH, +41%), and Spectral (SPEC, +34%).
Some of this is probably a spillover effect from TAO’s surge as traders move down the spectrum to chase lower-cap AI tokens for higher returns. There is clearly a risk-on sentiment in the market right now.
Price (7-day Change) | FDV | |
---|---|---|
Bittensor TAO | $543.82 (29.2%) | $11.4B |
Near NEAR | $5.65 (26.9%) | $6.6B |
Delysium AGI | $0.196 (46.9%) | $588M |
Spectral SPEC | $9.38 (34.7%) | $938M |
📊 Chart of the Week
TAO Emissions: Robust or Worrying?
Here’s a look at Bittensor’s emission schedule:
TAO follows a halvening schedule similar to BTC, with the first halvening set for November 16, 2025.
Right now, 35% of the total TAO is already in circulation (7.5M TAO). By the time of the next halvening, there will be 10.5M TAO in circulation—an increase of 3M TAO, or 40% of the current supply, over the next year.
With the large unlocks we’ve seen in the past year for projects like AltLayer, Arbitrum, Optimism, that has put a dampener on token prices, we can understand the anxiety.
As Emad points out:
To put this in context $TAO issuance over the next year or so is enough to pay for the 100,000 H100s @xai is using for that period, fastest supercomputer on earth
About 4x H100s @OpenAI used for GPT-4o
Web3 can easily provision mega compute
Let’s hope holders allocate it well
— Emad (@EMostaque)
12:17 PM • Sep 26, 2024
There are two ways to look at this—it’s a massive pool of resources Bittensor can use for AI development. If used wisely, Bittensor will thrive. If not… well, we’ve seen what happens.
🌴 AI is ‘Hangry’ for Data!
We’re approaching a wall regarding publicly scrapable data for AI training.
Rob from Volt Capital suggests that the problem may not be creating new data but accessing it.
We are hitting the limit on what can be publicly scraped.
Paths Forward:
1️⃣ coordinate to create new data
2️⃣ use synthetic data
3️⃣ generate data independentlyKey Distinctions:
> how it’s created (real vs synthetic)
> who creates it (partnerships vs solo)
> incentives… x.com/i/web/status/1…— Rob (@rsarrow)
2:40 PM • Sep 26, 2024
Platforms like Vana aim to solve this by enabling users to control and monetize their private data through dataDAOs. Users can submit their personal data, such as tweets, posts, and emails, to a dataDAO and be rewarded for it.
Synthetic data is already supplementing AI training. For example, Nous Research used synthetic data to train their Hermes LLMs, and Dria helps AI agents generate synthetic data for similar purposes. As real data becomes scarce, synthetic data will play a bigger role.
There’s also a need for more high-quality, organic data. Sapien AI believes that data labeling will become a major gig economy, with “frontier data”—high-quality, insightful information—becoming increasingly valuable. For instance, a doctor’s diagnostic process is an example of such valuable data.
As we near the limits of scrapable data, demand for private, high-quality, and synthetic data will grow.
🏆 Caught Our Eyes..
Source: Vana
Fundraising
Project Updates
Sentient releases their whitepaper on OML for AI entrepreneurship, discussing open vs closed model deployment paradigms
Chasm Network’s Scout Node Sale sells out in 6 minutes upon launch
Kuzco's Sam Hogan announces inference.net, a wholesaler of open-source LLM inference tokens
Hyperspace introduces Play, an AI research product using Groq and Meta's Llama3.1
Allora Network announces v0.5.0 release with improvements to user experience and network performance
CrunchDAO announces their token launch on Solana
BasedAI introduces their privacy-first decentralized AI solution
Gaia introduces Gaia Domain Names (GDN) for AI projects on the Gaia network
Prime Intellect teases an upcoming platform for contributing compute power to decentralized training
Incentives / Rewards Programs
Fluence Network launches a staking program for decentralized cloud computing
Bittensor
xponentcrisis shares insights from the Bittensor Asia Meetup at Token 2049
τaonomics announces upcoming Ethereum smart contracts on Bittensor
Tensorplex Labs introduces Tensorplex Delegate for TAO holders, offering 15% APR and weight assignment features
Macrocosmos releases V1.0.0 for subnet 25 Protein Folding
Sami Kassab discusses Ω Focus, a one-click Bittensor mining solution that pays users for focusing on priority tasks
Seth Bloomberg analyzes Bittensor's network effects and shares his concerns.
Wayfinder
Kalos demonstrates AIWayfinder's capabilities in creating complex smart contracts He also demos creating a smart contract using AIWayfinder, integrating Chainlink VRF and deploying it to Base
🔥 On X..
No crypto AI companies in YC - can someone fix this please?
Hearing the latest Y Combinator batch has 69 AI companies and no crypto companies.
Great sign that now is the absolute best time to be building consumer crypto apps. Popular sentiment in Silicon Valley is always dead wrong.
— Nick Tomaino (@NTmoney)
8:56 PM • Sep 23, 2024
Sama outlines the bright future ahead for AI
The Intelligence Age:
— Sam Altman (@sama)
5:00 PM • Sep 23, 2024
OpenAI CTO leaves. Is this a sign of trouble in the company?
I shared the following note with the OpenAI team today.
— Mira Murati (@miramurati)
7:34 PM • Sep 25, 2024
DMD releases a 26-page report on Bittensor
Over the last month I've been writing an in-depth report on Bittensor. The goal is to explain its unique design and why I believe TAO will be one of the best performers this cycle.
The report can be accessed here:
A thread to summarise my findings🧵
— DMD (@Daveeemor)
4:39 PM • Sep 23, 2024
Matthew Sigel shares Novogratz's explanation of AI and crypto intersection
Great Explanation of Intersection of AI & Crypto
by @novogratz
— matthew sigel, recovering CFA (@matthew_sigel)
4:12 PM • Sep 13, 2024
Jeff Amico explains why decentralized training is the future of AI
Why decentralized training is the future
Part 1: State of model training today
Part 2: Recent breakthroughs in distributed training
Part 3: Power of edge compute
Part 4: Remaining challengesPart 1 below, including why training is so expensive 👇 x.com/i/web/status/1…
— Jeff Amico (@_jamico)
5:00 PM • Sep 23, 2024
Bagel Network discusses prompt-based unlearning techniques for enhancing AI privacy
Prompt-Based Unlearning: Pioneering Selective Forgetting in AI
In our recent blog post on machine unlearning, we explore how cutting-edge techniques are reshaping AI privacy by enabling systems to "forget" sensitive information without compromising their overall functionality.… x.com/i/web/status/1…
— Bagel 🥯 (@bagel_network)
5:09 PM • Sep 24, 2024
Fish | Datura demonstrates renting H100 GPUs from Bittensor miners within one minute
Thanks @opentensor for having me on novelty search today.
Demo of renting H100 GPUs from bittensor miners within 1 minute here:— Fish | Datura (@fish_datura)
11:09 PM • Sep 26, 2024
Varun Mathur introduces a new agent interface to reduce AI costs
A breakthrough new agent interface
How do you reduce the price of advanced AI while scaling and increasing the intelligence of the system ? Are $20/month subscription plans, like for OpenAI's o1 model today, the only option for billions of us on this planet ?
At @HyperspaceAI… x.com/i/web/status/1…
— Varun (@varun_mathur)
10:22 PM • Sep 12, 2024
That’s it for this week! If you have specific feedback or anything interesting you’d like to share, please just reply to this email. We read everything.
Cheers,
Teng Yan & Joshua
Did you like this week's edition? |
This newsletter is intended solely for educational purposes and does not constitute financial advice. It is not an endorsement to buy or sell assets or make financial decisions. Always conduct your own research and exercise caution when making invest
Reply