“No need to go public:” Tether brushes off IPO talk post Circle debut

Solana’s DEX volume has reached $806.8 billion so far in 2025, with Jupiter still dominating. Solana’s (SOL) ecosystem has broken a major milestone in network activity. As of Wednesday, May 7, total decentralized exchange volume in 2025 on Solana was $806.8 billion, according to a data from Top Ledger. This represents a nearly 400% increase from the same period last year, when DEX volume was at $201 billion. Solana monthly DEX volume in USD | Source: Top Ledger January accounted for more than half of the DEX volume this year at $408 billion. The crpyto market rally, which started in November and culminated in January, caused a major explosion in activity. At the time, Solana’s price also contributed to rising dollar-denominated DEX metrics. By March, these figures went back to pre-rally levels. In terms of SOL, volumes saw a significant spike around January 20th, after which they returned to their previous levels. SOL-denominated daily DEX volumes on the Solana network | Source: DefiLlama Jupiter continues to dominate DEX volume on Solana In 2025, Jupiter exchange continued to dominate all other protocols in terms of volume, with $334.6 billion in volume. This means that 55% of all DEX volume routed through its aggregator. OKX was a distant second, with 5.21% of the volume or $32.2 billion. Interestingly, memecoin launchpad Pump.fun, facilitated $22.3 billion in volume, or 2.8% of the market. You might also like:Canada-listed SOL Strategies buys more Solana At the execution layer, Raydium remained in the lead, with $352.8 billion in volume. Meteora was second, with $113.7 billion or 14.17%. Following it was Orca with $103.9 billion in volume or 12.95%, and SolFi with $97.9 billion or 12.20%. DEX volume growth in 2025 coincided with rising adoption of Solana’s ecosystem. In the first quarter of 2025, Solana was firmly in second place after Ethereum. While Ethereum accounted for 30% of DEX volume on all chains, Solana controlled 23%. Read more:Solana upgrade raises block capacity by 20%, improves developer tools

Bitcoin treasury firm Strategy, formerly MicroStrategy, added another 15,355 BTC to its balance sheet over the past week. The company spent about $1.42 billion at an average price of $92,737 per Bitcoin, according to an SEC filing Monday. The purchase brings Strategy’s total Bitcoin (BTC) holdings to 553,555 BTC, worth over $52 billion at Bitcoin’s current price of just above $95,000.  The company’s cumulative average purchase price now stands at $68,459 per Bitcoin, reflecting a total investment of around $37.9 billion, including fees.  Strategy’s holdings account for more than 2.6% of Bitcoin’s fixed 21 million supply. Meanwhile, competition in corporate Bitcoin accumulation is heating up. Firms such as Cantor Fitzgerald, SoftBank, Bitfinex, and Tether recently announced a $3.6 billion Bitcoin venture, while new entrants like Twenty One Capital, led by Strike CEO Jack Mallers, are launching with a Bitcoin-native strategy.  Shares of Strategy’s MSTR closed 5.2% higher at $368.71 on Friday and are up another 1.42% in pre-market trading Monday. The stock has gained nearly 23% year-to-date, supported by Bitcoin’s price rebound and rising institutional interest in companies focused on Bitcoin. You might also like:Bitcoin poised to break $100k with ‘little difficulty’ amid capital inflows, analysts say Acquisition funding Proceeds from the sale of its Class A common stock and perpetual preferred stock funded the latest acquisitions. Between April 21 and April 27, Strategy sold approximately 4 million shares of MSTR for about $1.4 billion, while also selling over 435,000 STRK shares for $37.5 million.  Only $128.7 million worth of MSTR shares remain available for issuance under the company’s existing program. The move comes a week after Strategy added 6,556 BTC for $555 million, signaling the company’s continued aggressive approach to growing its Bitcoin treasury. You might also like:Founders Factory, Coinbase and more join forces to launch U.K. web3 accelerator

Bitcoin is showing renewed strength as on-chain data and macro conditions align to support a bullish outlook.  According to a May 13 post on the Bitfinex Alpha blog, the realized market cap of Bitcoin (BTC) has reached $889 billion after rising 2.1% in the past 30 days. The realized cap values each coin at the price at which it was last moved, unlike the regular market cap, which multiplies the current price by the total supply.  This approach better captures ongoing buying interest and more accurately represents the actual capital that has entered the market. Spot exchange-traded funds activity reflects that inflow. Net inflows into Bitcoin ETFs have exceeded $920 million over the last two weeks, indicating strong institutional interest. Meanwhile, on-chain data shows taht the number of coins held at a loss has drastically decreased, and over 3 million BTC have returned to profit territory. These changes suggest that market sentiment has improved and that sell-side pressure has decreased. You might also like:Falling Bitcoin dominance suggests altcoin season might be coming: MEXC COO After dropping 32% from its record high set earlier in this year, Bitcoin has made a strong comeback, regaining the $100,000 mark for the first time in more than three months. Bitfinex analysts credit the rally to a cooling of international trade tensions and the Federal Reserve’s more dovish stance. While broader risk markets have responded favorably to this macro backdrop, Bitcoin has outperformed equities, further confirming its status as a high-conviction asset for many investors. On the policy front, the U.S. continues to struggle with digital asset legislation. A key crypto bill, the Genius Act, failed in the Senate by a narrow vote, reflecting continued division on how to regulate the space. States like New Hampshire, on the other hand, are making progress with new legislation that permits investment in digital assets like Bitcoin.  All things considered, Bitcoin’s increasing realized cap and growing institutional presence suggest that this rally has deeper roots than many others and may still have room to grow. Read more:Tether adds 4,812 Bitcoin into Twenty One Capital’s treasury ahead of public listing

Tether has invested in pan-African blockchain solutions platform Shiga Digital, marking another step in its strategy to expand across the African continent. Fresh from another strategic investment in Chile-based crypto platform Orionx, with eyes on Latin America, Tether has announced plans to tap into its massive resources once again to bring Tether (USDT)-powered solutions to users across Africa. With Shiga Digital, Tether is looking to offer a solution to the problem of cross-border payments and global liquidity. The collaboration will leverage Shiga Digital’s suite of products, including virtual accounts, over-the-counter services, treasury management and foreign exchange. The services are targeted for enterprises across the continent, and USDT will be key to allowing users access payments in foreign currencies. Shiga Digital and Tether are looking to democratize traditional finance via the USDT integration.  You might also like:Tether allocates 10,500 BTC for SoftBank’s investment into Twenty One Capital Blockchain adoption is on the rise and not many crypto companies boost the traction and impact of Tether, issuer of world’s largest stablecoin by market share. Tether’s market cap currently stands at over $153 billion.  In early May, Tether published a report showing it held nearly $120 billion in U.S. Treasuries, with quarterly operating profit exceeding $1 billion. “At Tether, we believe stablecoins are the heartbeat of financial transformation,” said Paolo Ardoino, chief executive officer of Tether. “By collaborating with innovators like Shiga Digital, we aim to deliver financial access and efficiency to African enterprises. Together, we are not just imagining a future powered by blockchain technology, we are building it.” Tether’s growing presence in Africa extends beyond investment and cross-border payment solutions. The company is also committed to accelerating blockchain adoption through partnerships in education and innovation. In February, Tether signed a memorandum of understanding with the Republic of Guinea. The initiative aims to support the country’s digital transformation and economic growth by leveraging blockchain infrastructure. You might also like:Morocco arrests key suspect in French crypto kidnapping spree

“请”和“谢谢”只是礼貌用语，还是它们正在改变 ChatGPT 的学习和行为方式，并每天给 OpenAI 的人工智能造成数百万美元的损失？ 目录说“请”可能会花费数百万美元人工智能查询和隐藏的基础设施负载你的语气告诉了人工智能什么 说“请”可能会花费数百万美元 这是我们大多数人从小就被教导的道理。说“请”。说“谢谢”。礼貌无代价。但随着人工智能的发展，这句古老的箴言可能不再适用。对聊天机器人保持礼貌，实际上可能需要付出代价。 OpenAI 首席执行官 Sam Altman 在 X 上的一次简短交流中透露了一个关于 AI 系统运作机制的有趣细节。当被问及用户在 ChatGPT 查询中添加“请”和“谢谢”等额外词语会给 OpenAI 带来多少成本时，Altman 回答说：“数千万美元的花费是值得的。谁也说不准。” 数千万美元花得值——谁知道呢– Sam Altman (@sama) 2025 年 4 月 16 日 我们在 ChatGPT 中输入的每个单词都会经过庞大的数据中心进行处理，在那里它会被分解成标记，经过复杂的计算，最终变成回复。即使是一些简单的寒暄也会被同样对待。它们需要强大的计算能力。 这意味着电力、冷却系统以及每个请求需要更多时间。这些额外的令牌在数百万次对话中累积起来，就形成了实际的能源和基础设施成本。 根据 TechRadar 母公司 Future 在 2024 年 12 月进行的一项调查，51% 的美国人工智能用户和 45% 的英国人工智能用户经常使用人工智能助手或聊天机器人。 其中，美国人更注重礼貌。在美国，67% 的用户表示他们与人工智能交谈时彬彬有礼。其中，82% 的用户表示，无论对方是否是人类，他们觉得这样做是正确的。 另外18%的人则有不同的动机。他们表示，保持礼貌只是为了防止人工智能真的爆发叛乱——虽然可能性不大，但他们不想冒险站到错误的一边。 pic.twitter.com/37wyhZOm9k— Cubiqq (@cubiqq) 2025年4月16日 Then there’s the remaining 33% of American users who don’t bother with niceties. For them, the goal is to get answers, fast. They either find politeness unnecessary or believe it slows them down. Efficiency, not etiquette, shapes the way they interact. AI queries and the hidden infrastructure load ChatGPT 的每一次响应都由消耗电力和水的计算系统驱动。看似简单的来回交互背后隐藏着巨大的资源消耗，尤其是在用户数量不断增长的情况下。 高盛的一份报告估计，每次 ChatGPT-4 查询大约消耗 2.9 瓦时的电力，几乎是一次谷歌搜索的十倍。 据 Epoch AI 称，GPT-4o 等较新的模型已经提高了效率，将每次查询的能耗降低至约 0.3 瓦时。然而，当每天进行数十亿次查询时，即使是微小的差异也会迅速累积起来。 OpenAI 的运营成本反映了这一规模。据报道，根据多个行业消息来源引用的内部估计，该公司每天花费约 70 万美元来维持 ChatGPT 的运行。 这笔费用背后的主要原因是其庞大的用户群。从2024年12月到2025年初，每周用户数量从3亿跃升至4亿以上，部分原因是受到吉卜力风格艺术提示等病毒式功能的推动。随着使用量的激增，对电网和物理基础设施的需求也随之增加。 国际能源署预测，到 2030 年，数据中心将推动发达经济体电力需求增长 20% 以上，而人工智能被认为是这一增长的主要驱动力。 水是这个等式的另一个部分，经常被忽视。《华盛顿邮报》的一项研究发现，撰写一封 100 字的人工智能生成电子邮件大约需要 0.14 千瓦时的电力，足以点亮 14 个 LED 灯泡一小时。 产生相同的响应会消耗 40 到 50 毫升的水，主要用于冷却处理数据的服务器。 从规模上看，这种消耗水平引发了更广泛的担忧。弗吉尼亚州是美国数据中心密度最高的州，2019年至2023年间，该州的用水量增长了近三分之二。根据《金融时报》的调查，仅2023年，该州的总用水量就至少达到了18.5亿加仑。 随着数据中心在全球范围内不断扩张，尤其是在电力和土地价格低廉的地区，预计当地水资源和能源供应的压力将会越来越大。其中一些地区可能无法承受长期影响。 你的语气告诉了人工智能什么 在经过大量人类对话训练的人工智能系统中，用户提示的语气会强烈影响回应的语气。 使用礼貌的语言或完整的句子通常会使答案更具信息量、更符合语境、也更尊重他人。这种结果并非偶然。 在幕后，像 ChatGPT 这样的模型是在大量人类写作数据集上进行训练的。在微调过程中，它们会根据人类反馈进行一个称为强化学习的过程。 In this stage, real people evaluate thousands of model responses based on criteria such as helpfulness, tone, and coherence. When a well-structured or courteous prompt leads to a higher rating, the model begins to favor that style. Over time, this creates a built-in preference for clarity and respectful language patterns. Real-world examples reinforce this idea. In one informal Reddit experiment, a user compared AI responses to the same question framed with and without the words “please” and “thank you.” The polite version often triggered longer, more thorough, and more relevant replies.  Hackernoon 上发表的另一项分析发现，不礼貌的提示往往会产生更多事实错误和偏见内容，而适度礼貌的提示则在准确性和细节之间取得了最佳平衡。 这种模式在各种语言中都适用。在一项涉及英语、中文和日语的跨语言测试中，研究人员发现粗鲁的提示会全面降低模型的性能。 极度礼貌并不一定能带来更好的答案，但适度的礼貌通常能提高质量。研究结果还暗示了文化差异，表明“恰当”的礼貌程度会因语言和语境的不同而有所差异。 话虽如此，礼貌并非万能良方。最近的一项即时工程评估测试了 26 种提升 AI 输出的策略。其中之一就是添加“请”这样的词语。 结果表明，虽然这类短语有时有所帮助，但它们并不能持续提高 GPT-4 的正确率。在某些情况下，添加额外的单词会引入噪音，导致回答不够清晰或准确。 A more detailed study conducted in March 2025 examined politeness at eight different levels, ranging from extremely formal requests to outright rudeness.  研究人员使用 BERTScore 和 ROUGE-L 等基准来衡量摘要任务的结果。无论语气如何，准确性和相关性都保持相当一致。 然而，回复的长度各不相同。当提示非常突然时，GPT-3.5 和 GPT-4 会给出较短的答案。LLaMA-2 的表现有所不同，在礼貌程度适中时，回复最短，而在礼貌程度极高时，回复则较长。 礼貌程度似乎也会影响人工智能模型处理偏见的方式。在刻板印象检测测试中，过于礼貌和敌意的提示都会增加出现偏见或拒绝反应的可能性。中等礼貌程度的表现最佳，最大限度地减少了偏见和不必要的审查。 在测试的模型中，GPT-4 完全拒绝的可能性最小，但所有模型都表现出类似的模式——似乎有一个最佳点，音调可以帮助模型准确响应而不损害平衡。 最终，我们说什么以及我们如何说决定了我们能得到什么。无论我们想要得到更好的答案、更少的偏见，还是仅仅想要更深思熟虑的互动，我们的措辞都至关重要。 And while politeness might not always boost performance, it often brings us closer to the kind of conversation we want from the machines we’re increasingly talking to.