Microsoft’s recent earnings report answered a key question: Is the company’s heavy investment in AI paying off? The results show strong growth, with revenue increasing 18% to $90 billion for the quarter ending June 30, and net income rising 31% to $35.8 billion. The Microsoft Cloud division delivered $59.3 billion in revenue, up 27%, with Azure playing a major role. Following the announcement, Microsoft’s stock rose approximately 2% after hours, highlighting investor confidence in the company’s cloud and AI strategy.
Two major technology companies reported their AI expenditures on the same day with very different results. Microsoft's investment in AI was reflected positively, contributing to a rise in its cloud revenues. On the other hand, Meta’s AI spending resulted in a cash flow deficit. Meta managed to increase its revenue by 28% to $60.8 billion, surpassing market expectations, according to CNBC. Despite this growth, Meta’s profits declined by 14% to $15.8 billion, its earnings per share fell short of predictions, and its stock price dropped approximately 5%.
ICON plc has entered into a multi-year collaboration with Anthropic to integrate Anthropic's AI, Claude, into the clinical trial process. ICON, a global contract research organization, supports pharmaceutical companies by managing their clinical trials. As of June, the Dublin-based company employed around 40,200 people across 99 locations in 55 countries. Financial details of the partnership were not disclosed.
A penetration test at a major financial services firm, managing billions in client assets, revealed a serious vulnerability in AI-generated code. The assessment, conducted by incident response firm Sygnia, focused on a customer onboarding application largely developed with Claude AI. While the code demonstrated strong handling of government-issued IDs, identity verification, and payment data, the critical flaw discovered raises important security concerns for AI-assisted development in finance.
The number of software vulnerabilities detected by AI has roughly doubled compared to last year. Despite this surge, very few of these flaws are being actively exploited. According to the US National Vulnerabilities Database, 45,207 software weaknesses were recorded between January and July 27, nearing the total for all of 2025, which was already a record year. If this trend continues, the total for this year could be about twice as high as in 2025, Bloomberg reports.
On April 3, Starbucks informed NomadGo that it would discontinue the AI-driven inventory tool developed by the Redmond-based startup. Shortly after, NomadGo, a company of 30 employees, laid off a significant portion of its staff, including the technical team dedicated to Starbucks. The decision was made without informing Starbucks baristas in advance.
Anthropic revealed that their Claude Mythos Preview has identified significant mathematical vulnerabilities in two key cryptographic algorithms. The first vulnerability drastically reduces the effective key strength of HAWK, a post-quantum digital signature scheme currently being evaluated by NIST. The second flaw enhances the best-known attack against seven-round AES, a widely used symmetric encryption standard, by a factor of 200 to 800 times. These discoveries highlight overlooked weaknesses despite extensive expert review.
Starting Sunday, the European Commission officially gains enhanced authority to oversee the world’s leading AI developers under the new AI Act, now two years in effect. The EU’s AI Office, staffed by just 36 personnel, will have the power to demand documentation, perform in-depth evaluations, and access cutting-edge AI models directly. Non-compliance can lead to significant fines, potentially reaching 3% of a company's global revenue.
The advancement of enterprise software has resolved many operational challenges over the last decade. Real-time supply chain tracking and automated global payroll systems now function seamlessly across multiple tax areas. Despite these breakthroughs, organizing business trips often remains a cumbersome task, with employees navigating outdated interfaces and manual approval processes. This inefficiency persists even though automation and AI have transformed other business operations.
The US Food and Drug Administration (FDA) announced Dexcom as the first company chosen to participate in its Technology-Enabled Meaningful Patient Outcomes (TEMPO) pilot program. This initiative permits certain digital health devices to be offered to Medicare beneficiaries prior to formal FDA authorization through a policy known as enforcement discretion. Rather than issuing a full approval, this program allows earlier access to promising medical technologies.
