Mintttch🐣

Cryptomint

0Following

126followers

Feed

The real cost of AI is becoming clear when companies use it extensively and the bills are more expensive than expected . Fortune reports that many technology companies are facing new problems from using AI within their organizations . Not because AI is unusable but because when used at a large scale, the costs start to exceed what many companies anticipated . The case being discussed is Microsoft . According to The Verge Microsoft has begun canceling most direct Claude Code licenses within the company and is directing engineers to use GitHub Copilot CLI instead . Claude Code is a coding assistant tool from Anthropic which Microsoft allowed thousands of employees to trial about 6 months ago . The group trying it out included not only developers but also project managers, designers, and other team members . The problem is Claude Code became popular very quickly and high-volume usage started to become a cost that needed to be controlled . However, this internal license cancellation does not affect the major deal between Microsoft and Anthropic through Foundry . That deal includes an investment of up to $5 billion in Anthropic and an agreement for Anthropic to purchase $30 billion worth of Azure compute . Microsoft is not the only company facing this issue . The Information reports that Uber exhausted its AI coding tools budget for 2026 within just the first 4 months . Previously, Uber encouraged employees to use AI more through an internal leaderboard ranking teams by AI tool usage . The key point is many AI tools charge based on tokens . The more you use and the more data you have AI read performing multiple steps writing code fixing code or working in an agentic workflow . the compute costs increase accordingly . This is what the report calls the AI paradox . Even though the cost per token is decreasing if usage volume grows faster the company’s total bill can still rise . Goldman Sachs predicts that agentic AI could increase token usage by 24 times by 2030 reaching about 120 quadrillion tokens per month . Meanwhile, Gartner expects the inference cost of large models to drop nearly 90% by 2030 compared to 2025 . But Gartner also warns that cheaper tokens do not necessarily mean enterprise AI will be cheaper overall . Because AI agents use many more tokens per task than general models and AI providers may not pass all cost savings on to customers . Bryan Catanzaro from NVIDIA told Axios "For my team, compute costs are much higher than employee salaries" . This shifts the question about AI in organizations . From originally asking Does AI help employees work faster? . To now Is the speed gained worth the compute cost paid? . Because on the day every company wants AI agents working alongside employees the resulting costs may be larger than many expect

The Department of Disease Control reveals that Thailand has found the NB.1.8.1 COVID variant as the main strain, with no increase in severe symptoms reported. Today (May 24), Montien Kanasawat, Director-General of the Department of Disease Control, spoke about data from the Communicable Diseases Agency Singapore as of May 21, 2026, regarding the COVID-19 infection situation. It was reported that between May 10-16, 2026, there were 12,700 new cases, an increase from about 8,000 cases the previous week. The average daily hospitalized patients increased from 56 to 73, with ICU patients averaging about 1 per day. The main circulating variant is NB.1.8.1, found in more than half of the detected cases in the country. Regarding the situation in Thailand in 2026, data from the Disease Surveillance System (DDS) as of May 23, 2026, shows a cumulative total of 3,642 COVID-19 cases and 1 death. Most patients are aged 30–35 years, followed by those aged 60 and above, and then 20–29 years. In the past month, there has been an increase in reported cases, but still below the 5-year median. For the SARS-CoV-2 virus variants in Thailand, data from the Public Health Research Institute, Department of Medical Sciences, between January 1, 2025, and April 23, 2026, shows that NB.1.8.1 is the main circulating variant in Thailand, accounting for 50.95% of detected samples, followed by JN.1 (24.97%) and XEC (9.14%). "In 2025, the number of cases and cluster outbreaks increased significantly from April to June. Surveillance of SARS-CoV-2 variants in Thailand found NB.1.8.1 as the main variant spreading during the same period. NB.1.8.1 has mutations in several spike protein sites beyond those in JN.1, enhancing its transmissibility and immune evasion, but there is no evidence that it causes more severe disease." Direk Khamphaen, Deputy Director-General of the Department of Disease Control, added that the COVID-19 situation in Thailand is now endemic or seasonal. Although disease severity and spread trends have decreased, important measures must still be maintained, focusing on social measures balanced with the new normal lifestyle, and emphasizing strict adherence to COVID-19 prevention measures, especially personal hygiene, as follows: ⚈ Wash hands frequently with soap and water or use alcohol-based gel before eating or drinking, after using the restroom, or after touching commonly shared surfaces such as doorknobs and handrails. ⚈ Cover mouth and nose with a cloth or tissue every time when coughing or sneezing. ⚈ Avoid entering crowded or congested areas; if unavoidable, wear a mask at all times. ⚈ Avoid close contact with people showing respiratory symptoms to reduce infection risk, especially the 608 risk group: elderly and those with underlying conditions, who may face fatal outcomes if infected. ⚈ If experiencing symptoms such as fever, cough, or runny nose, perform an initial test with an ATK and avoid close contact with others, especially young children, the elderly, and those with underlying conditions, to prevent transmitting the virus to at-risk groups at home. If positive, promptly see a doctor. The public can inquire further by calling the Department of Disease Control hotline at 1422. #TheStandardNews