SoftBank profits drop 18%, stock falls 4% in Tokyo

fortune.com Aug 7, 2026

SoftBank profits drop 18%, stock falls 4% in Tokyo

Curated by Shubham · TechCapsules · Aug 7, 2026

AI-summarised brief · reviewed before publication

SoftBank Group Corp. reported an 18% decline in fiscal first-quarter profit to 347.3 billion yen, driven by higher costs that offset investment gains. Despite the profit drop, quarterly sales rose nearly 11% to 2 trillion yen. Chief Financial Officer Yoshimitsu Goto confirmed strong performance in the Arm business and announced an additional $20 billion investment in OpenAI, with more planned for the current fiscal year. The Tokyo-based conglomerate, founded by Masayoshi Son, continues its aggressive strategy in artificial intelligence, energy, and robotics through its Vision Fund. Its portfolio includes major stakes in ByteDance, Intel, PayPay, and TSMC. While the company does not provide annual forecasts, shares fell 4% in Tokyo trading following the earnings release. The results reflect the volatile nature of SoftBank’s high-risk, high-reward investment model in fledgling technology sectors.

💡 Why It Matters

· The additional $20 billion commitment to OpenAI underscores SoftBank’s pivotal role in funding the global AI infrastructure race.
· This move signals that major institutional capital is prioritizing generative AI dominance over short-term profitability metrics.

AI Capsules Investment

Read full article on fortune.com

ChatGPT brings unlimited text chats to free users

OpenAI has lifted text‑chat limits for all ChatGPT users as the platform surpasses one billion weekly users. The upgrade introduces GPT‑5.6 Luna as the default model for Free and Go accounts, replacing GPT‑5.5, and adds a “Think” button that lets users request higher reasoning power for complex queries. Plus and Pro subscribers receive GPT‑5.6 Sol, optimized for quick tasks, and a thinking slider to adjust depth of analysis. Internal tests show factual errors dropped 62% for Luna and 68% for Sol.

💡 By removing text limits, OpenAI expands access to advanced AI reasoning, enabling broader experimentation and productivity across its user base.

businessworld.in · Aug 7

India Risks Becoming A ‘Digital Colony’ Without Sovereign AI Models: Nasscom Chairperson

At the Nasscom FutureForge 2026 conference, Nasscom chairperson Srikanth Velamkanni warned that India risks becoming a ‘digital colony’ unless it develops sovereign AI models and deeptech capabilities. He traced India’s historic strength to low‑capital, high‑margin software services built on a large, educated workforce, but argued that the next era demands AI‑driven productivity gains of ten to one‑hundred times, which will make software cheaper and shift competition to deep innovation. Velamkanni noted that Indian IT and ITeS firms are beginning to invest in and acquire deeptech startups, citing HCLTech’s stake in Sarvam as an example. He emphasized that deeptech ventures require capital, R&D and the ability to define problems rather than merely solve briefs, and that success could create millions of new jobs.

💡 Sovereign AI models will keep India’s data and digital infrastructure under national control, shaping its future tech sovereignty.

appleinsider.com · Aug 7

Adobe is hell-bent on becoming the creative engine behind AI

Adobe and OpenAI launched a new ChatGPT plugin that bundles more than 70 Adobe creative tools into a single interface activated by typing @Adobe. Users with an Adobe account can install the plugin in ChatGPT and generate images, short videos, PDFs and other assets, with free access for basic features and expanded capabilities for paid ChatGPT or Adobe subscriptions. The integration builds on earlier single‑tool plugins and follows Adobe’s broader strategy of embedding its software across AI platforms, including upcoming plugins for Anthropic’s Claude, Google Gemini and Slack. Adobe hopes the all‑in‑one offering will lock creators into its ecosystem as the market shifts toward freelance, project‑based work and AI‑driven content production, countering recent revenue pressures and competition from lower‑cost alternatives.

💡 By embedding its suite directly into ChatGPT, Adobe positions itself as the default creative engine for the next wave of AI‑generated content, aiming to retain relevance amid a fragmented, price‑sensitive creator economy.

opalesque.com · Aug 7

The AI singularity is closer than you think

Nvidia has cemented its status as a standout growth firm in the global technology sector, with revenue climbing from roughly $60 billion to over $120 billion across the last two fiscal years, according to senior market analyst Antonio Di Giacomo of XS.com. The surge reflects the company’s dominance in graphics processing units and expanding artificial‑intelligence workloads, positioning it at the forefront of the industry’s rapid evolution. Di Giacomo’s commentary appears in a subscription‑only newsletter that requires active login and cookie consent. The piece is interspersed with unrelated snippets about Leavenworth’s Bavarian tourism, private‑market fund trends, hedge‑fund consolidation and trading recommendations, underscoring the fragmented nature of the source material. Analysts view the doubling of revenue as a benchmark for AI‑driven hardware demand worldwide.

💡 The revenue jump shows AI‑centric chips are becoming essential infrastructure, speeding adoption across multiple sectors.

gizmodo.com · Aug 7

This Startup Will Pay You $50 to Let AI Read Your Mind

In June, Meta unveiled Brain2Qwerty, a system that translates brain scans into text to aid communication for people with neurodegenerative diseases. Parallel to this effort, San Francisco startup Conduit, founded in 2024 by Oxford graduate Rio Popper and AI researcher Clem von Stengel, is developing non‑invasive “telepathy at scale” – a wearable helmet with electrodes that captures neural signals for AI interpretation. To train its models, Conduit pays participants $50‑$55 for two‑hour sessions in which they converse with large language models while wearing the four‑pound device, limiting each person to 32 sessions. The company requires participants to type without looking at their fingers and encourages free‑form dialogue to maximize data. Former OpenAI alignment researcher Naomi Bashkansky recently joined Conduit, underscoring the venture’s growing talent pool.

💡 Conduit’s paid data‑collection scheme accelerates the creation of consumer‑grade mind‑reading interfaces, potentially reshaping how humans interact with AI beyond voice and text.

windowslatest.com · Aug 7

Microsoft admits it force-installed AI-powered OneDrive on Windows 11, says it’s not…

Microsoft confirmed that the OneDrive Photos app was force‑installed on Windows 11 devices as part of a recent OneDrive update, extending the web and mobile photo experience to the desktop. The app appears on both consumer and enterprise PCs, even though it only works with personal Microsoft accounts, leaving IT admins unable to remove it without uninstalling the entire OneDrive client. Microsoft said the rollout to managed devices was unintentional and that the AI‑powered “People” feature, which groups faces, is disabled by default and does not train models on local photos; any cloud‑processed data is deleted after 30 days. The company plans to release an update that will let users uninstall OneDrive Photos separately.

💡 It forces a consumer‑focused AI feature onto business machines, creating admin headaches and raising privacy concerns about unintended data handling.