Bittensor ($TAO) experienced a remarkable increase of 19.19% within the past day, driven by heightened interest in its AI-centric subnet ecosystem.
This surge coincided with a trending moment on social media from Templar, one of the most engaged subnets associated with $TAO, alongside an uptick in enthusiasm for decentralized artificial intelligence tokens.
Demand for Subnet Tokens Boosts $TAO Prices
At the time of writing, Bittensor’s primary token, $TAO, was valued at $284.75, reflecting a 19.19% rise over the last 24 hours.
Bittensor ($TAO) Price Performance. Source: BeInCrypto
The reason behind this price movement appears to be quite clear. Investors must possess $TAO to trade it for subnet tokens; thus, an increase in excitement surrounding Templar has led to significant buying pressure.
A community account known as Tao Telegraph pointed out that just one viral post about Templar was sufficient to elevate demand for $TAO. They speculated that simultaneous excitement across various subnets could greatly enhance this effect.
To be honest, $TAO‘s rise is largely due to this viral post by @tplr_ai.
Investors MUST own $TAO if they want to buy tao subnet tokens… there’s no alternative.
The hype around one subnet token this week has increased demand for $$ TA O ; just imagine if there were similar demands across multiple subnets… https://t.co/d8x4zGlpuy
— @TaoTelegraph — Bittensor Media (@taotelegraph), March 14, 2026
The top three subnets within the Bittensor ecosystem—SN3 Templar, SN4 Targon and SN39 Basilica—ranked among CoinGecko’s eight highest daily gainers. During this period,$ TA O exceeded $280 as well.
$ TA O surpassed $280 while three of its ecosystem’s subnets are among the top eight daily gainers.
🟢 SN3 – @tplr_ai
🟢 SN4 – @TargonCompute
🟢 SN39 – @basilic_ai pic.twitter.com / awtKkpW6ZG— CoinGecko (@coingecko), March 15 ,2026
Covenant-72B from Templar Marks AI Achievement Milestone
Beneath all the buzz lies a significant technical milestone achieved recently. On March 10th , Templar announced it had completed Covenant-72B—a large language model boasting an impressive count of 72 billion parameters and pre-trained solely on Bittensors’ Subnet Three.
The model underwent training using approximately1 .1 trillion tokens via standard internet connections without any centralized cluster or whitelist involvement . Anyone equipped with GPUs could freely participate .
To tackle bandwidth restrictions ,Templar employed SparseLoCo techniques where each participant executed local optimizer steps before compressing and sharing updates enabling decentralized training at such vast scales .
We have successfully completed history’s largest decentralized LLM pre-training run : Covenant -72B . It was permissionless on B ittensors’ Subnet Three .
With74 billion parameters and ~1 .1 trillion tokens processed through regular internet without any centralized infrastructure or whitelisting involved anyone possessing GPUs could join or exit freely .*n pic.twitter.com/W0Ks563Cld
— templar(@ tplr _ai)March10 ,2026
The team claims that Covenant -72B performs competitively against centralized models like LLaMA -270B .
Van de Poppe Adjusts Strategy Around Strength of$ TA O
*/
Trader Michael van de Poppe revealed he sold off10 .42 units of $ TA O ‘‘for approximately$3000 rotating proceeds into SEIand EIGENwhile maintaining nearly50 %of his portfolio allocated towards $ TA O ‘‘and NEAR.
*blockq uote*
*
“The intersection between AI & Crypto represents one key narrative driving upcoming markets,” Van de Poppe noted.*
*
He highlighted how recent movements involving $ TA O ‘‘ constituted what he termed as ‘1.8 sigma event’ within daily timeframes suggesting slight overextension which may lead into short-term corrections providing opportunities re-entry points.*
Whether$
{
{tag}}
}
can maintain these levels hinges upon ongoing activity surrounding various sub nets along with broader momentum seen throughoutAI sectors leading into next week.
This article titled **“**B i tt en so r ($ {tag} }) Surges20 %asTemplar ’ s ViralSubnetHypeFuelsBuyingFrenzy”** appeared first on BeInCrypto**.