The popular AI framework Virtuals Protocol will host an in-person meet-up in Hong Kong on Friday, January 17.

March 16 — As electricity demand from AI data centers surges, the market is fiercely debating whether this will undermine Bitcoin mining or even threaten the Bitcoin network’s security. Cryptocurrency trader Ran Neuner said on social media that AI has become Bitcoin mining’s biggest competitor, as both rely heavily on electricity resources. He noted Bitcoin mining generates roughly $57 to $129 per megawatt-hour, while AI data centers can bring in $200 to $500 per megawatt-hour — up to eight times more than mining. As a result, more and more mining firms are pivoting to AI businesses. Neuner listed several mining companies that have embraced AI: - Core Scientific recently secured a $1 billion credit line for its AI hosting business; - MARA Holdings filed documents with the U.S. SEC, hinting it may sell Bitcoin to fund its AI transformation; - Hut 8 signed a $7 billion AI infrastructure deal with Google in December 2023; - Cipher Mining also cut its mining power to focus on

March 16 (Bitget Market Data) — Japan and South Korea’s key stock indexes closed mixed on Monday. South Korea’s KOSPI rose 62.61 points (1.14%) to finish at 5,549.85. By contrast, Japan’s Nikkei 225 slipped 68.46 points (0.13%) to close at 53,751.15.

On March 16th, per the Coinbob Popular Address Monitor, two Hyperliquid traders using a pyramid position management strategy—Paul Wei (0xdae) and “100% Win Rate Low Drawdown” (0x721)—have aligned on a take-profit range for BTC long positions. Both target a short-term BTC peak near $78k: - Paul Wei placed a take-profit order in the $75k-$78.1k range in the last hour. - “100% Win Rate Low Drawdown” set their order in the same range two days prior. For their next round of long position scaling: - “100% Win Rate Low Drawdown” placed 50 pyramid-style buy orders in the last hour, planning to add more positions in the $68k-$70.5k range once BTC dips below $70.5k. - Paul Wei, by contrast, is more conservative, targeting a re-entry around $66.5k. On leverage: - Paul Wei uses 2-3x leverage for wide-range rebalancing. - “100% Win Rate Low Drawdown” uses 20x leverage to chase higher returns in a narrow range—their initial capital has already tripled. Both traders employ a r

On March 16, Gnosis co-founder Friederike Ernst stated that the U.S. Digital Asset Market Structure Clarity Act could grant major financial institutions greater control over the crypto market. She noted certain provisions of the bill assume market activities need to go through centralized intermediaries, which could weaken blockchain users’ role as network participants and stakeholders. Ernst believes overreliance on institutional intermediaries would see users revert to “clients renting financial technology services” rather than actual network participants. However, she also noted the bill partially clarifies regulatory boundaries between the U.S. Securities and Exchange Commission (SEC) and Commodity Futures Trading Commission (CFTC), providing some protection for peer-to-peer transactions and self-custody. The CLARITY Act currently faces controversy in Congress, with the main disagreement centering on stablecoin governance issues. Galaxy Digital’s Head of Research Alex Thorn p

On March 16, official sources confirmed that Bittensor’s Templar Subnet (SN3) completed the largest decentralized large language model (LLM) pre-training in history on March 10. The model, **Covenant-72B**, boasts 720 billion parameters. It was pre-trained by the Templar team on Bittensor Subnet 3, using only public internet data (no centralized data centers required). Covenant-72B scored 67.1 on the zero-shot MMLU (Massive Multitask Language Understanding) benchmark—outperforming centralized baselines like LLaMA-2-70B and LLM360 K2 under identical test conditions. It is the largest fully permissionless collaborative LLM to date, with over 70 distinct nodes contributing computing power throughout the process. The team has released all model weights and checkpoints under the Apache License. The news appears to have fueled gains for Bittensor (TAO) and its subnet token: - TAO is up 54.8% over the past two weeks. - The Templar subnet token (τemplar) has surged 194% in the las

March 16 – Leading crypto outlet CoinCodex recently named its Top 8 ASIC Miners, with the ICERIVER? ALEO AE3 miner earning a spot. Per CoinCodex’s report, the ICERIVER? ALEO AE3 is an ASIC miner built exclusively for the zkSNARK algorithm, tailored for mining on the Aleo blockchain. It delivers a max hash rate of 2GH/s (±5%), making it a high-performance option for miners focused on Aleo’s privacy-focused network. Additional details from the ranking include: the AE3 runs at 3400W with an efficiency of 1.7 J/MH; it’s slated for release in November 2025; it emits 75dB of noise; and it requires a 200-250V power supply. Its stable, durable build and enterprise-grade specs make it ideal for professional mining setups.