Altcoins see widespread pullback, TAC drops over 40% in 24 hours.

Cancore, a trading platform built on the Canton Network, has closed a $7.2 million seed funding round. The platform is designed to bridge assets on the Canton Network with liquidity on public blockchains. Participating in the round are Paper Ventures, dao5, ArkStream Capital, X Ventures, MH Ventures, and Proof of Talk. Wasif Rashid, co-founder and CEO of Cancore, stated that no formal lead investor was named in the round, though Paper Ventures and dao5 secured the largest allocations. The funding was executed via a Simple Agreement for Future Tokens (SAFT), with a fully diluted valuation of $60 million corresponding to Cancore’s planned CNRX token issuance, Rashid added.

According to a Reuters report, Moody’s Ratings said Tuesday that massive borrowing to fund artificial intelligence (AI) infrastructure is pushing the private credit sector beyond traditional corporate lending. However, a large volume of loans will mature in the coming years, leaving weaker borrowers facing potential refinancing difficulties. Specifically, AI financing needs have expanded beyond data center construction to areas including land, power, semiconductors, graphics processing units (GPUs), and tensor processing units (TPUs). Moody’s projects that total capital expenditures of the U.S.’s five largest hyperscale cloud service providers and CoreWeave will reach nearly $1 trillion by 2027, with funding sources including operating cash flow, debt, and financing structures such as joint ventures and special purpose vehicles (SPVs). Moody’s warned that loans maturing between 2027 and 2029 will test weaker borrowers’ ability to refinance amid higher interest rates. Of the 607 borrowing firms tracked by Moody’s, around 17% use payment-in-kind (PIK) interest—a practice that defers cash interest payments and adds accrued interest to debt, further increasing their debt burden. Additionally, Moody’s flagged risks related to “circular financing”: as cross-investments, lending, and procurement between tech firms, chipmakers, and AI developers grow, these interconnected exposures could deteriorate simultaneously if the industry enters a downturn.

Renowned trader Killa posted that he is still holding his long Bitcoin position entered around $76,400, with no changes to his current trading plan. To capture significant market moves, one must patiently endure extended sideways ranges and dull price action; the difference between profitable and losing traders often lies in patience and the ability to stick to a predefined plan amid shifting market sentiment. Killa noted he will not move his stop-loss up to the entry price, instead holding the position until either his target is hit or the stop-loss is triggered. Frequent position changes by traders are precisely why many suffer trading losses; the key is whether the market structure has changed or if the trading plan needs adjustment. Both factors remain unchanged, so he will not panic, second-guess himself, or waver, and will continue executing his original strategy.

Crypto venture capital firm Dragonfly has announced that Rob Hadick and Tom Schmidt have been promoted to managing partners, and will lead the firm alongside Haseeb Qureshi and Bo. Dragonfly noted that the promotions recognize the critical roles the two have played in the company's growth, and mark a further refinement of its partnership structure.

US President Trump said that soon, we will officially ban the term "artificial intelligence (AI)" in all relevant documents.

Oracle (ORCL.N) has launched an AI data platform designed for large-scale enterprise artificial intelligence applications. The company announced an expansion of its cloud AI chip collaboration with NVIDIA (NVDA.O), planning to scale the deployment of AI servers equipped with NVIDIA’s next-generation Vera Rubin chips to the multi-gigawatt level. It will also introduce Google TPU chips, along with AI inference chips from SambaNova Systems and Cerebras Systems, to enrich its computing power supply. Oracle further plans to use its self-developed Acceleron network chips in some NVIDIA servers to replace a portion of NVIDIA’s networking products. The firm stated that different AI workloads have varying computing power needs, and a diversified chip supply helps customers select suitable infrastructure based on performance, throughput and cost. NVIDIA dominates the AI training chip market, while competition in the AI inference market is more diverse.