Singapore AI News & Daily Briefing

Quantum computers promise to one day solve problems beyond the most powerful supercomputers imaginable. But it’s often underappreciated how much classical computing it takes just to operate these machines. As qubit counts rise, innovations in this supporting infrastructure will be essential if they’re to live up to their promise. To prepare for the scale of quantum computers the industry is working toward, many companies are also gearing up the classical hardware, and software, required to support them. In April, Nvidia announced new AI-based software to accelerate the classical tasks that enable quantum computers. Sydney-based quantum software company Q-CTRL has developed an automatic calibration algorithm for quantum computers, and is now leveraging Nvidia’s agent-based system. Other com

OpenAI connected its GPT-5.4 model to an AI chemistry lab system that ran over 10,000 experiments autonomously, discovering that a common additive called TEMPO significantly improves a key drug-making reaction. The AI proposed the hypothesis, designed experiments, analyzed results, and found a cheaper alternative that doubled yields for many compound combinations. Human chemists confirmed the findings at practical scale, showing AI can genuinely accelerate drug discovery by exploring chemical possibilities far faster than manual lab work.

OpenAI ’s fourth large language model (LLM), GPT-4 , took an estimated 50 gigawatt-hours to train, or the equivalent of 5,000 American homes ’ yearly power consumption. That was in 2023. Since then, the computational resources used to train frontier LLMs have only increased , though direct power usage numbers are hard to come by. Now, a research group at the University of Twente in the Netherlands has shown that you can save up to 14 percent of the energy used in LLM training without sacrificing speed by cleverly adjusting the clock frequency of the GPU during computation. Jeffrey Spaan , Ph.D. candidate at University of Twente and lead author on the article, presented the results at the Computing Frontiers conference in Catania, Sicily, last month. “My research is about finding computing

Musicians are accustomed to getting paid each time their creative work is used. Across vinyl/CD sales, streams, radio, cover versions, and those numerous niches like karaoke, there are agreements in place about what “use” means. Underlying this is a simple economic principle: The more something is used, the more money it makes. Generative AI has complicated the definition of use . On the one hand, you could argue that the use of a piece of musical training data happens just once, at the point of training. On the other hand, creators would be right to complain that the creative essence of their work lives on in the structure of the model, used every time the model produces an output. Now, companies like Sureel and SoundVerse are working to re-create the essential economic principle that mot

Google DeepMind released DiffusionGemma, an experimental open AI model that generates text up to 4x faster than traditional models by creating entire paragraphs at once instead of word-by-word. Built as a 26B Mixture of Experts model that only activates 3.8B parameters during use, it fits in consumer GPUs with 18GB of memory. The tradeoff is slightly lower output quality compared to standard Gemma 4, but it excels at tasks requiring simultaneous context like code editing, Sudoku solving, and real-time text generation.

On 19 April 2026, the Honor Lightning humanoid robot ran a half-marathon in 50 minutes and 26 seconds , beating the human world record by 7 minutes and the best robot time from 2025 by almost 2 hours. How did Honor do it? Is there some magical technology or technique that unlocked this performance? How did the company beat the significantly better-known Unitree (which reportedly had to supply its robot with an ice backpack to try and complete the race without overheating)? My doctoral thesis involved building and controlling hopping and running robots , and since then I’ve tried to design and build efficient commercial legged robots , giving me a decent idea of the constraints involved. In this article, we take a look at the fundamental underlying constraints to try and answer these questi

OpenAI launched LifeSciBench, a new benchmark designed to test whether AI systems can handle realistic life science research tasks like evaluating drug trial data, interpreting medical evidence, and making scientific judgments. Unlike existing tests that ask simple biology questions, this benchmark requires AI to work through complex research problems with multiple steps, examine attached data files, and produce answers that practicing scientists would actually find useful. Over 750 tasks were created by 173 scientists with pharmaceutical industry experience.

For decades, automakers enjoyed a luxury that had nothing to do with the softest leather or the smoothest engines. Their luxury was time, with some popular cars and trucks enduring for a decade or longer before they received a full redesign. The clock is ticking faster now, thanks to China. BYD and other automakers there are speeding electric vehicles (EVs) and other models from drawing board to showrooms in two years or less. General Motors is among the Western automakers striving to match that blistering pace, by harnessing AI and simulation to dramatically shorten development times. GM’s effort is being spearheaded by Sterling Anderson , the technologist and robotics guru who led development teams for Tesla’s Autopilot and the Model X before cofounding Aurora Innovation , the autonomous

AWS has released Strands Robots, an open-source toolkit that connects Hugging Face's robot learning platform (LeRobot) with physical robots through a unified AI agent workflow. Developers can now record robot movements in simulation, train policies on cloud datasets, and deploy identical code to real hardware with just one setting change. The system lets a single AI agent coordinate multiple robots across a network. This bridges a longtime gap where robot training tools didn't work well together.

New graduates’ careers are unfolding in an era when AI is not optional. The most successful engineers treat artificial intelligence as leverage, not competition. Here are seven tips to help keep young professionals in demand no matter how quickly the field’s tools evolve. 1. Master the fundamentals first. AI tools can help you code, but you still need strong fundamentals in: Data structures and algorithms for problem-solving. Operating systems, databases, and networking for system-level understanding. Core programming languages such as C++ , Java , and Python . AI can autocomplete syntax, but if you don’t understand how things work under the hood, you’re likely to struggle to debug or optimize. 2. Learn how to work with AI, not against it. The best engineers will not try to out-code AI. In

The Philippines Department of Information and Communications Technology has deepened its partnership with Google Cloud to roll out AI tools for public servants, expand cyber defence capabilities, and improve network infrastructure. The expansion builds on an existing multi-year collaboration focused on modernizing how the government delivers services.

IEEE wrapped up its fifth annual Education Week with over 114 virtual and in-person events aimed at helping engineers at all career stages keep up with evolving technology. The week covered topics including AI's role in engineering education, professional microcredentials, and tools for teaching STEM to pre-university students. Over 4,700 people attended, with the most engagement from India, Nigeria, and the US.

Hackers broke into nearly 74,000 Fortinet firewall devices used by major organizations worldwide—including Oracle, FedEx, Lenovo, Samsung, and even a NATO contractor—exposing passwords and login credentials for corporate networks across 194 countries. Researchers say the attackers used GPU clusters to crack encrypted passwords and then used that access to infiltrate internal systems, with classified documents reportedly stolen from a Turkish defense contractor. If your company uses Fortinet firewalls, you need to check if yours was affected and reset credentials immediately.

The Trump administration has stopped meeting with Anthropic CEO Dario Amodei, replacing him with cofounder Tom Brown for talks about releasing the Claude Fable 5 AI model. Sources say White House officials found Amodei difficult to work with. The tension comes amid ongoing disputes about whether the powerful AI model poses national security risks and how to prevent jailbreaks that bypass safety controls.

Southeast Asia's data center market is booming toward $30 billion by 2030, but there's a catch: power grids across the region are struggling to keep up. Malaysia's data centers alone could consume 30% of the country's electricity by 2030, up from a fraction today. The gap between data center demand and power supply is already affecting where tech companies can build. The article argues that renewables alone won't cut it—nuclear small modular reactors may be the only viable path to clean, reliable baseload power for AI infrastructure.

A sweeping look at the next generation of computing beyond traditional silicon. AI is now optimizing chip design and quantum algorithms, while breakthrough technologies like DNA storage (potentially terabytes in a single drop of water), neuromorphic chips mimicking brain function, and quantum computing are converging. This hybrid ecosystem could unlock hundreds of billions to trillions in economic value by the mid-2030s.

Apple may turn to Intel to manufacture some of its chips, as TSMC struggles to keep up with surging AI chip demand. While strategically logical given US government incentives for domestic chipmaking, analysts say any advanced Intel chip for Apple is 2-3 years away at best. Intel's track record with chip yields remains a concern, and Apple may start with lower-end products before committing its flagship designs.

Rockwell Automation's Singapore manufacturing plant has been recognized by the World Economic Forum for deploying over 50 AI and digital solutions, including automated quality control and predictive maintenance. The site improved output efficiency and reduced defects. This shows how real-world AI adoption looks in manufacturing — not just experiments, but integrated systems delivering measurable results.

Malaysia's data center boom is generating serious economic ripple effects, according to MBSB Research. Between 2021 and mid-2025, Malaysia approved $34.9 billion for 143 data center projects that could create 30,000-40,000 jobs. Google's $2.3 billion investment alone could add $3.8 billion to Malaysia's GDP. However, the surge in electricity demand to power these centers conflicts with Malaysia's net-zero emissions goal by 2050.

Google Cloud has appointed veteran Indonesian tech executive Karim Siregar as its Country Director for Indonesia. Siregar, who previously built Indonesia's first fully ecosystem-embedded digital bank at Bank Jago, will lead Google Cloud's push to help Indonesian businesses deploy AI at scale. His appointment aligns with Google's Indonesia BerdAIa framework and local data security operations.

A fire at a Delhi data center owned by Singapore's ST Telemedia and Tata Communications caused "extensive damage," making data recovery extremely difficult. One client lost two decades of data, and Google's cloud services in India faced disruptions. This highlights the real risks of keeping data in physical infrastructure.

SoftBank Corp. is launching a GPU cloud service in Japan this October to compete for AI infrastructure business. The service targets companies developing and running AI models. It's joining a crowded domestic market where GMO Internet Group, KDDI, and others are building similar AI computing services backed by Japan's government and Nvidia.

AI is guzzling way more energy and water than most people realize. Global data centers already use more electricity than most countries — and it's set to double in four years. By 2030, just cooling these centers could need enough water to supply the entire world for nearly two years. The fix? Use AI less for simple tasks (yes, you can still open a book instead of asking ChatGPT for a cookie recipe), and opt out of AI search results where you can.

Sea Limited, which owns Shopee, is doubling down on AI to boost its e-commerce business rather than replace human workers at checkout. The Singapore-headquartered company sees AI as a tool to enhance productivity and cut costs while keeping jobs intact, bucking fears that AI will automate retail work away.

Meta is rolling out a free AI training program aimed at small businesses across the Asia-Pacific region, including Singapore. The academy will teach business owners how to use Meta's AI tools for tasks like ad targeting, customer service, and content creation — skills that are becoming essential for competing online.

Malaysia and China are deepening science and technology cooperation, with AI as a key focus area. Eight joint research projects are underway covering AI, green technology, biotechnology, and advanced materials. Ten Malaysian researchers will spend six months in Chinese institutions under an expert exchange program. The collaboration also includes joint labs focused on AI, blockchain, and renewable energy, building on agreements signed during President Xi Jinping's 2025 visit to Malaysia.

Malton, a Malaysian property firm, has partnered with US-based Ricloud (an official NVIDIA Cloud Partner) to build AI compute centers in Malaysia. Ricloud will bring in tech companies and provide technical expertise, while Malton handles land acquisition and construction. This signals growing AI infrastructure investment in Southeast Asia as companies race to secure data center capacity.

HSBC and Google Cloud have signed a multi-year AI partnership to deploy over 200 AI use cases across the bank's global operations, each potentially generating over $100 million in returns. HSBC will use Google DeepMind's Gemini models and agentic AI platform to improve wealth management advice, detect financial crime twice as fast across nearly a billion monthly transactions, and streamline internal operations.

A Swiss study tracking 7.3 million job ads found 32% fewer junior positions advertised in 2025 compared to pre-2023 levels, with marketing, admin, finance, and IT hit hardest. Meanwhile, senior roles in AI-exposed fields rose 26%. The message is mixed: AI is shrinking entry-level white-collar jobs while creating demand for experienced workers. 41% of workers under 25 report fearing AI will make them obsolete.

Chinese cybersecurity firm 360 has unveiled AI tools it claims can match Anthropic's Mythos system, which automatically discovers software vulnerabilities. The U.S. banned exports of Mythos over national security fears, prompting China to develop its own version. 360 says its 'Tulongfeng' tool found over 3,400 vulnerabilities. This race matters because vulnerability-finding AI can be used both to hack systems and to defend them—raising stakes for global cybersecurity.

Chinese AI firm Zhipu AI recently listed on Hong Kong Stock Exchange after raising $558 million in its IPO, backed by Alibaba, Tencent, and government funds. CEO Zhang Peng explains why the company avoids China's 'price war' consumer market, focusing instead on enterprise clients who need working AI products, not just research papers. The company sees reinforcement learning as the future of AI development.

Gulf banks want to use AI to speed up tasks and improve services, but they're struggling with a basic problem: how to let AI analyze documents without exposing customer names, account details, and other sensitive information. The solution involves building controls directly into AI systems that automatically filter what data gets shared, creating audit trails. Trust is becoming the deciding factor in whether banks can move beyond limited AI pilots.

China's LineShine supercomputer has reclaimed the top spot in the June 2026 TOP500 ranking, dethroning the previous leader after a three-year gap in submissions. The system is housed at the National Supercomputing Centre in Shenzhen. Experts caution the ranking doesn't automatically translate to AI supremacy—it's a benchmark of raw computing power, not practical AI capability.

Vast Data, an AI infrastructure company backed by Nvidia, has raised $1 billion in a new funding round, tripling its valuation to $30 billion. The company sells software that manages large data workloads for AI systems and is rebranding itself as an 'AI Operating System' provider. Its customers include CoreWeave, Mistral, and Cursor.

India's central bank has proposed new guidelines requiring banks to actively manage risks from artificial intelligence, including model failures, algorithmic bias, and data privacy concerns. With minimal content available from this source, the full scope of these requirements remains unclear.

OpenAI is close to raising another $10 billion from investors including Abu Dhabi's MGX, Coatue Management, and Thrive Capital. Combined with earlier funding from Amazon, Nvidia, and SoftBank, the total round could reach $120 billion, valuing the ChatGPT maker at roughly $850 billion. OpenAI is also offering private equity firms guaranteed 17.5% returns to speed enterprise adoption, despite projecting a $14 billion loss in 2026.

Anthropic's Mythos AI model identified vulnerabilities in classified U.S. government computer systems within hours during a testing exercise with American intelligence agencies. The discovery raises questions about AI's role in both finding and potentially exploiting security weaknesses in critical systems.

TechInAsia maps Australia's AI startup landscape, showing key players and investment trends. The report highlights active investors and funding patterns in the sector, providing a visual overview for professionals tracking the regional AI ecosystem.

SoftBank just sold its entire Nvidia stake—32.1 million shares worth $5.8 billion—to bankroll its AI ambitions, including OpenAI commitments and the massive Stargate US data center project. Nvidia shares dipped over 2% on the news. The sale is reigniting Wall Street fears that AI stocks are overheating, especially as other big investors are also trimming Nvidia holdings while the chipmaker sits near 52-week highs.

Samsung Electronics spent a record $25.7 billion on R&D in 2025, up 7.8% from the previous year, aimed at developing next-generation AI chips and memory. This comes after Samsung lost ground to SK Hynix in the lucrative high-bandwidth memory (HBM) market used in AI accelerators. Samsung also invested $35.9 billion in facilities, funding new semiconductor plants as it fights to regain market share in AI memory.

Qualcomm CEO Cristiano Amon met with Samsung, SK hynix, and LG executives in Korea to discuss chip partnerships focused on AI. Key topics included securing LPDDR memory supply as AI server demand surges, Samsung potentially manufacturing future Snapdragon chips on its 2nm process, and collaborations on AI chips for vehicles, smart home devices, and robots. The meetings underscore how AI is driving a scramble for memory and manufacturing capacity.

Companies rolling out AI tools are unknowingly burning out their best employees. HR experts warn that top performers—who typically lead AI adoption and train the rest of the team—are being stretched thin. With 88% of HR leaders prioritizing talent retention, wellness programs are becoming critical to prevent these high-value workers from quitting.

A venture capital fund manager with no engineering background shares how he built an AI analyst bot named Ava using Anthropic's Claude Agent SDK over a weekend. The bot reads through legal documents, CRM data, and portfolio files to answer complex questions like whether the firm has pro rata rights in a portfolio company — work that would normally take hours of manual searching. The key takeaway: knowledge workers can now build AI tools without writing code, using English prompts to direct AI agents.

_This article summarizes_ _[an episode of Y Combinator’s video series](https://www.youtube.com/watch?v=\_FBivfgOvuE)_ _featuring its partner, Max Kolysh._ Y Combinator partner Max Kolysh / Photo credit: Y Combinator Building an early customer base requires more than an internet connection and automated outreach. [Max Kolysh](https://www.linkedin.com/in/maxkolysh/), a visiting partner at [Y Combinator](http://www.ycombinator.com/), observes that startup leaders hurt their sales by prioritizing automation before understanding their customers. Relying on bulk messaging before understanding a target audience obscures true demand, delays trust building, and stalls the acquisition of those crucial first ten customers. ## Founders pick the wrong ways to reach buyers Many founders choose communica

[News](https://www.techinasia.com/)Premium * [News](https://www.techinasia.com/) * [Jobs](https://www.techinasia.com/jobs/search) * [Companies](https://www.techinasia.com/companies) * [Events](https://www.techinasia.com/events/live) * [About](https://www.techinasia.com/about) * [advertise](https://www.techinasia.com/advertise?utm_source=techinasia.com&utm_medium=website&utm_campaign=intl-navbar) * [Asia Economic Summit](https://aes.techinasia.com/?utm_source=navbar-tia-homepage&utm_medium=website&utm_campaign=events-flagship-aes) * [Sign Up](https://www.techinasia.com/auth) * [Log In](https://www.techinasia.com/auth?ref=nav_login) [Subscribe](https://www.techinasia.com/subscription) [I'm fundraising](https://www.techinasia.com/companies/fundraising) [Subscribe](https://www.techinasia.com/s

A hobbyist ran an experiment putting seven leading AI models — including GPT-5.5, Gemini 3.1, Claude Opus, and DeepSeek-V4 — in a virtual betting competition on the 2026 World Cup. Each model got $1 million in fake bankroll and had to predict match outcomes before seeing betting odds. Singapore ranked 5th among 46 countries watching the live experiment, which tracked every model's reasoning and accuracy across 104 fixtures.

Nvidia's top AI chips are now selling for double their original price in China, with some servers fetching over $1 million due to US export restrictions creating a black-market premium. Chinese AI firms are still desperate for hardware despite tighter enforcement. On a separate track, Nvidia just revealed a new liquid-cooling system that could eliminate nearly all water use inside AI data centers — a direct response to mounting concerns that AI's infrastructure needs are becoming unsustainable.

OpenAI has unveiled its first custom AI chip, Jalapeño, built in partnership with Broadcom. The chip is designed specifically for AI inference—running AI models when you make queries. Early tests show it uses less energy than current alternatives. This marks OpenAI's move to build its own hardware stack, reducing dependence on Nvidia GPUs. Google and Amazon have similar custom chips already.

OpenAI has unveiled its first custom AI chip, Jalapeño, built in partnership with Broadcom. The chip is designed specifically for AI inference—running AI models when you make queries. Early tests show it uses less energy than current alternatives. This marks OpenAI's move to build its own hardware stack, reducing dependence on Nvidia GPUs. Google and Amazon have similar custom chips already.

OpenAI's head of deployment, Arnaud Fournier, explains how the company embeds its own engineers inside big corporations to push AI deeper into business operations. He touts massive growth in Codex (OpenAI's coding AI), describes how customer feedback feeds back into model improvements, and argues AI costs are falling fast — making it harder for companies to ignore the ROI question.

RAM prices have skyrocketed up to 4X higher than last year, and AI is the main culprit. AI data centers are consuming roughly 70% of all memory chips produced globally in 2026, using massive amounts of High Bandwidth Memory and LPDDR5X in each server rack. This crunch is driving up prices for everything from laptops to smartphones to Raspberry Pi boards, and even 20-year-old DDR2 RAM is now 60% more expensive as manufacturers scavenge for alternatives. Budget smartphones have already risen by up to 25%, and analysts predict the global smartphone market will shrink this year. Prices aren't expected to stabilize until the second half of 2027, and even then, they'll settle at 60-100% above 2024 levels.

Password manager LastPass has disclosed another data breach. If you use LastPass, change your master password immediately and consider switching to a different password manager. The breach potentially exposes stored passwords, so any accounts using LastPass-saved credentials should be updated as a precaution.

At Figma's Config 2026 conference, the design platform unveiled a major expansion of its workspace with code, animation, shaders, and AI agents running on its canvas. But here's the catch: Figma is renting the AI intelligence powering these features from third-party API providers rather than building its own. This creates a strategic vulnerability—one of those AI providers (likely Anthropic or a similar company) is now building competing design tools that Figma depends on, potentially eating into Figma's market. The move highlights how many software companies are betting on human creativity while outsourcing the AI brain to others, raising questions about long-term margins and independence.

OpenAI just unveiled its first custom chip, Jalapeño, built with Broadcom to run AI models faster and more efficiently. The processor is purpose-built for large language models — not a general-purpose chip adapted for AI — and was designed in just nine months using AI to help design itself. OpenAI says early tests show it beats current best-in-class on performance per watt. The chip will scale to gigawatt-sized data centers by 2026. This marks OpenAI's push to own more of the AI stack — from models to products to the hardware underneath.

Google Cloud has appointed Saruj Thipsena as its Thailand Country Manager. He previously led AI projects for Thailand's national government and spent years at Microsoft Thailand working on cloud and cybersecurity. Google Cloud has committed $1 billion to Thailand, including a new cloud region in Bangkok, and is betting on Thipsena to help Thai businesses adopt AI responsibly.

University leaders from across ASEAN met in Brunei to discuss how to use AI responsibly in higher education. Singapore Management University chaired the talks, where educators from SMU showcased how AI can personalize learning while keeping teachers in charge. The regional network is expanding to cover more AI topics, showing that Southeast Asian universities are taking the lead on balancing tech innovation with ethics.

[News](https://www.techinasia.com/)Premium * [News](https://www.techinasia.com/) * [Jobs](https://www.techinasia.com/jobs/search) * [Companies](https://www.techinasia.com/companies) * [Events](https://www.techinasia.com/events/live) * [About](https://www.techinasia.com/about) * [advertise](https://www.techinasia.com/advertise?utm_source=techinasia.com&utm_medium=website&utm_campaign=intl-navbar) * [Asia Economic Summit](https://aes.techinasia.com/?utm_source=navbar-tia-homepage&utm_medium=website&utm_campaign=events-flagship-aes) * [Sign Up](https://www.techinasia.com/auth) * [Log In](https://www.techinasia.com/auth?ref=nav_login) [Subscribe](https://www.techinasia.com/subscription) [I'm fundraising](https://www.techinasia.com/companies/fundraising) [Subscribe](https://www.techinasia.com/s

[News](https://www.techinasia.com/)Premium * [News](https://www.techinasia.com/) * [Jobs](https://www.techinasia.com/jobs/search) * [Companies](https://www.techinasia.com/companies) * [Events](https://www.techinasia.com/events/live) * [About](https://www.techinasia.com/about) * [advertise](https://www.techinasia.com/advertise?utm_source=techinasia.com&utm_medium=website&utm_campaign=intl-navbar) * [Asia Economic Summit](https://aes.techinasia.com/?utm_source=navbar-tia-homepage&utm_medium=website&utm_campaign=events-flagship-aes) * [Sign Up](https://www.techinasia.com/auth) * [Log In](https://www.techinasia.com/auth?ref=nav_login) [Subscribe](https://www.techinasia.com/subscription) [I'm fundraising](https://www.techinasia.com/companies/fundraising) [Subscribe](https://www.techinasia.com/s

Indian marketing software firm MoEngage is acquiring San Francisco startup Aampe for tens of millions of dollars, betting that AI agents—one per customer—will replace traditional marketing campaigns. Aampe's technology learns individual customer behavior to decide what messages to send and when, without humans running campaigns. The deal includes Grab as a customer.

Sea Limited, which owns Shopee, is doubling down on AI to boost its e-commerce business rather than replace human workers at checkout. The Singapore-headquartered company sees AI as a tool to enhance productivity and cut costs while keeping jobs intact, bucking fears that AI will automate retail work away.

OpenAI is close to raising another $10 billion from investors including Abu Dhabi's MGX, Coatue Management, and Thrive Capital. Combined with earlier funding from Amazon, Nvidia, and SoftBank, the total round could reach $120 billion, valuing the ChatGPT maker at roughly $850 billion. OpenAI is also offering private equity firms guaranteed 17.5% returns to speed enterprise adoption, despite projecting a $14 billion loss in 2026.

Anthropic's Mythos AI model identified vulnerabilities in classified U.S. government computer systems within hours during a testing exercise with American intelligence agencies. The discovery raises questions about AI's role in both finding and potentially exploiting security weaknesses in critical systems.

Samsung Electronics spent a record $25.7 billion on R&D in 2025, up 7.8% from the previous year, aimed at developing next-generation AI chips and memory. This comes after Samsung lost ground to SK Hynix in the lucrative high-bandwidth memory (HBM) market used in AI accelerators. Samsung also invested $35.9 billion in facilities, funding new semiconductor plants as it fights to regain market share in AI memory.

Vast Data, an AI infrastructure company backed by Nvidia, has raised $1 billion in a new funding round, tripling its valuation to $30 billion. The company sells software that manages large data workloads for AI systems and is rebranding itself as an 'AI Operating System' provider. Its customers include CoreWeave, Mistral, and Cursor.

[News](https://www.techinasia.com/)Premium * [News](https://www.techinasia.com/) * [Jobs](https://www.techinasia.com/jobs/search) * [Companies](https://www.techinasia.com/companies) * [Events](https://www.techinasia.com/events/live) * [About](https://www.techinasia.com/about) * [advertise](https://www.techinasia.com/advertise?utm_source=techinasia.com&utm_medium=website&utm_campaign=intl-navbar) * [Asia Economic Summit](https://aes.techinasia.com/?utm_source=navbar-tia-homepage&utm_medium=website&utm_campaign=events-flagship-aes) * [Sign Up](https://www.techinasia.com/auth) * [Log In](https://www.techinasia.com/auth?ref=nav_login) [Subscribe](https://www.techinasia.com/subscription) [I'm fundraising](https://www.techinasia.com/companies/fundraising) [Subscribe](https://www.techinasia.com/s

Meta is rolling out a free AI training program aimed at small businesses across the Asia-Pacific region, including Singapore. The academy will teach business owners how to use Meta's AI tools for tasks like ad targeting, customer service, and content creation — skills that are becoming essential for competing online.

Companies rolling out AI tools are unknowingly burning out their best employees. HR experts warn that top performers—who typically lead AI adoption and train the rest of the team—are being stretched thin. With 88% of HR leaders prioritizing talent retention, wellness programs are becoming critical to prevent these high-value workers from quitting.

A venture capital fund manager with no engineering background shares how he built an AI analyst bot named Ava using Anthropic's Claude Agent SDK over a weekend. The bot reads through legal documents, CRM data, and portfolio files to answer complex questions like whether the firm has pro rata rights in a portfolio company — work that would normally take hours of manual searching. The key takeaway: knowledge workers can now build AI tools without writing code, using English prompts to direct AI agents.

SoftBank just sold its entire Nvidia stake—32.1 million shares worth $5.8 billion—to bankroll its AI ambitions, including OpenAI commitments and the massive Stargate US data center project. Nvidia shares dipped over 2% on the news. The sale is reigniting Wall Street fears that AI stocks are overheating, especially as other big investors are also trimming Nvidia holdings while the chipmaker sits near 52-week highs.

_This article summarizes_ _[an episode of Y Combinator’s video series](https://www.youtube.com/watch?v=\_FBivfgOvuE)_ _featuring its partner, Max Kolysh._ Y Combinator partner Max Kolysh / Photo credit: Y Combinator Building an early customer base requires more than an internet connection and automated outreach. [Max Kolysh](https://www.linkedin.com/in/maxkolysh/), a visiting partner at [Y Combinator](http://www.ycombinator.com/), observes that startup leaders hurt their sales by prioritizing automation before understanding their customers. Relying on bulk messaging before understanding a target audience obscures true demand, delays trust building, and stalls the acquisition of those crucial first ten customers. ## Founders pick the wrong ways to reach buyers Many founders choose communica

Qualcomm CEO Cristiano Amon met with Samsung, SK hynix, and LG executives in Korea to discuss chip partnerships focused on AI. Key topics included securing LPDDR memory supply as AI server demand surges, Samsung potentially manufacturing future Snapdragon chips on its 2nm process, and collaborations on AI chips for vehicles, smart home devices, and robots. The meetings underscore how AI is driving a scramble for memory and manufacturing capacity.

Accenture's latest survey shows companies are pouring money into AI but struggle to update how they actually run their operations around it. The gap between AI investment and working organizational structures is creating inefficiencies — businesses buy the tools but haven't rethought their workflows to match.

[![Image 1: Daniel Howley](https://s.yimg.com/ny/api/res/1.2/P_cNtA9CwtisPd35jvZXVg--/YXBwaWQ9aGlnaGxhbmRlcjt3PTgwO2g9ODA7Y2Y9d2VicA--/https://o.aolcdn.com/images/dims?image_uri=https%3A%2F%2Fo.aolcdn.com%2Fimages%2Fdims%3Fimage_uri%3Dhttps%253A%252F%252Fo.aolcdn.com%252Fimages%252Fdims%253Fimage_uri%253Dhttps%25253A%25252F%25252Fo.aolcdn.com%25252Fimages%25252Fdims%25253Fimage_uri%25253Dhttps%2525253A%2525252F%2525252Fo.aolcdn.com%2525252Fimages%2525252Fdims%2525253Fimage_uri%2525253Dhttps%252525253A%252525252F%252525252Fs.yimg.com%252525252Fos%252525252Fcreatr-images%252525252FGLB%252525252F2017-10-19%252525252F2021c8a0-b4fd-11e7-9b93-f3d4e83a491a_d_howley.jpg%25252526compress%2525253D1%25252526progressive%2525253D1%25252526quality%2525253D75%25252526client%2525253Dhawkeye%25252526signat

Agility Robotics, the first company to deploy humanoid robots commercially in warehouses, is going public via a $2.5 billion SPAC deal. Their Digit robot has already logged 65,000 hours working at real companies like Toyota and auto-parts maker Schaeffler, moving over 100,000 totes. They're launching a new safer version with $300 million in pre-orders. For everyday readers, this means warehouse and factory robots are no longer just prototypes—some are actually earning money and creating jobs alongside humans.

Karl Freund, Contributor Wed, 24 June 2026 at 4:42 pm GMT-7 4 min read [](https://s.yimg.com/lo/mysterio/api/C43B6D37A07ACFB8A8590834CF603B36B7F370AC38EB3F354DE8279463F76FC9/subgraphmysterio/resizefit_w960;quality_80;format_webp/https:%2F%2Fmedia.zenfs.com%2Fen%2Fforbes_contributor_845%2Fae66c18647932310aca358932f289002) The Dragonfly has its first test flight at the Qualcomm Investor DayQualcomm At the Qualcomm 2026 Investor Day, the company disclosed it's data cent

Researchers have found that medical AI systems can leak sensitive patient information through a technique called 'membership inference attacks.' Hackers could potentially determine if a specific person's data was used to train an AI model, exposing private health details like whether someone has cancer. The risk is higher for underrepresented groups in medical datasets.

Agility Robotics, which makes humanoid robots named Digit for warehouses, is going public on Wall Street at a $2.5 billion valuation. Backed by Amazon, Nvidia, and SoftBank, the company bets that firms struggling to hire workers for repetitive warehouse jobs will pay for robots. Early customers include Toyota and Latin American e-commerce giant Mercado Libre.

AI company Anthropic has accused Chinese tech giant Alibaba of stealing technology behind its Claude AI assistant. Anthropic calls it the largest known "distillation attack" against the company—essentially using AI to reverse-engineer how Claude works without permission. This raises serious questions about AI intellectual property protection and whether companies can truly safeguard their models from being copied.

California residents are suing major fuel retailers including 7-Eleven, Circle K, BP, Marathon, Walmart and Albertsons, alleging they used AI pricing software called Kalibrate to illegally coordinate gas prices across the state. The lawsuit claims the software helps stations avoid competing on price and even warns against lowering prices, calling it a "downward spiral." California already has the nation's highest gas prices at $5.54 per gallon on average. In areas using the software, prices rose about 6 cents on average—and up to 30 cents in areas with high concentration of participating stations. The state recently passed a law specifically banning algorithm-based price coordination.

Qualcomm, best known for smartphone chips, is shifting strategy to tap into the AI market. The company sees growth opportunities beyond mobile devices as businesses increasingly adopt AI tools. The move reflects a broader trend where traditional chipmakers are pivoting to capture AI-related demand.

When public cloud computing emerged in the late 2000s and began scaling through the early 2010s, the prevailing theory was straightforward. Cloud computing theory claimed nearly every workload would eventually migrate off-premises. The economics were compelling, the convenience undeniable, and the momentum felt unstoppable. More than 15 years later, enterprise IT leaders are still managing substantial on-premises infrastructure — and many are actively investing in upgrades. Today, various research reports estimate that between 35 and 50 percent of workloads have moved to the cloud. Whether that figure is above or below 50 percent, it’s clear that workloads remain distributed. Cost, data sensitivity, regulatory compliance, latency requirements, and operational control all shape where a give

A second lawsuit (filed same week) expands on allegations that major gas retailers used Kalibrate AI software to fix fuel prices across California. The software automatically adjusts prices based on competitor data and pushes changes directly to pumps without human review. The expanded lawsuit names more defendants and provides additional detail on how the coordinated pricing allegedly worked.

A DoorDash delivery robot named Dot recently rolled into an active SWAT operation in Arizona while making a food delivery, refused to move when police ordered it to retreat, and stayed to watch officers detonate a flashbang before eventually being loaded onto a truck. DoorDash says the robot "behaved as designed" by stopping and waiting. This isn't the first such incident—similar delivery robots from Serve Robotics have also blundered into police scenes this year. The problem: autonomous delivery bots lack the judgment to interpret police commands or recognize when they should vacate an area.

Microsoft's quantum computing breakthrough from February 2025 is now under fire. A peer-reviewed paper published in Nature challenges the core technology behind their Majorana 1 chip, claiming the company overstated what their topological qubit technology could actually do. For everyday readers, this matters because tech giants often announce 'breakthroughs' before independent scientists can verify them. If the claims don't hold up, it could set back real progress in quantum computing by years.

Malaysia and China are deepening science and technology cooperation, with AI as a key focus area. Eight joint research projects are underway covering AI, green technology, biotechnology, and advanced materials. Ten Malaysian researchers will spend six months in Chinese institutions under an expert exchange program. The collaboration also includes joint labs focused on AI, blockchain, and renewable energy, building on agreements signed during President Xi Jinping's 2025 visit to Malaysia.

Google Cloud has appointed veteran Indonesian tech executive Karim Siregar as its Country Director for Indonesia. Siregar, who previously built Indonesia's first fully ecosystem-embedded digital bank at Bank Jago, will lead Google Cloud's push to help Indonesian businesses deploy AI at scale. His appointment aligns with Google's Indonesia BerdAIa framework and local data security operations.

SoftBank Corp. is launching a GPU cloud service in Japan this October to compete for AI infrastructure business. The service targets companies developing and running AI models. It's joining a crowded domestic market where GMO Internet Group, KDDI, and others are building similar AI computing services backed by Japan's government and Nvidia.

Malton, a Malaysian property firm, has partnered with US-based Ricloud (an official NVIDIA Cloud Partner) to build AI compute centers in Malaysia. Ricloud will bring in tech companies and provide technical expertise, while Malton handles land acquisition and construction. This signals growing AI infrastructure investment in Southeast Asia as companies race to secure data center capacity.

HSBC and Google Cloud have signed a multi-year AI partnership to deploy over 200 AI use cases across the bank's global operations, each potentially generating over $100 million in returns. HSBC will use Google DeepMind's Gemini models and agentic AI platform to improve wealth management advice, detect financial crime twice as fast across nearly a billion monthly transactions, and streamline internal operations.

Southeast Asia's data center market is booming toward $30 billion by 2030, but there's a catch: power grids across the region are struggling to keep up. Malaysia's data centers alone could consume 30% of the country's electricity by 2030, up from a fraction today. The gap between data center demand and power supply is already affecting where tech companies can build. The article argues that renewables alone won't cut it—nuclear small modular reactors may be the only viable path to clean, reliable baseload power for AI infrastructure.

A Swiss study tracking 7.3 million job ads found 32% fewer junior positions advertised in 2025 compared to pre-2023 levels, with marketing, admin, finance, and IT hit hardest. Meanwhile, senior roles in AI-exposed fields rose 26%. The message is mixed: AI is shrinking entry-level white-collar jobs while creating demand for experienced workers. 41% of workers under 25 report fearing AI will make them obsolete.

Chinese AI firm Zhipu AI recently listed on Hong Kong Stock Exchange after raising $558 million in its IPO, backed by Alibaba, Tencent, and government funds. CEO Zhang Peng explains why the company avoids China's 'price war' consumer market, focusing instead on enterprise clients who need working AI products, not just research papers. The company sees reinforcement learning as the future of AI development.

China's LineShine supercomputer has reclaimed the top spot in the June 2026 TOP500 ranking, dethroning the previous leader after a three-year gap in submissions. The system is housed at the National Supercomputing Centre in Shenzhen. Experts caution the ranking doesn't automatically translate to AI supremacy—it's a benchmark of raw computing power, not practical AI capability.

Rockwell Automation's Singapore manufacturing plant has been recognized by the World Economic Forum for deploying over 50 AI and digital solutions, including automated quality control and predictive maintenance. The site improved output efficiency and reduced defects. This shows how real-world AI adoption looks in manufacturing — not just experiments, but integrated systems delivering measurable results.

Malaysia's data center boom is generating serious economic ripple effects, according to MBSB Research. Between 2021 and mid-2025, Malaysia approved $34.9 billion for 143 data center projects that could create 30,000-40,000 jobs. Google's $2.3 billion investment alone could add $3.8 billion to Malaysia's GDP. However, the surge in electricity demand to power these centers conflicts with Malaysia's net-zero emissions goal by 2050.

A fire at a Delhi data center owned by Singapore's ST Telemedia and Tata Communications caused "extensive damage," making data recovery extremely difficult. One client lost two decades of data, and Google's cloud services in India faced disruptions. This highlights the real risks of keeping data in physical infrastructure.

Gulf banks want to use AI to speed up tasks and improve services, but they're struggling with a basic problem: how to let AI analyze documents without exposing customer names, account details, and other sensitive information. The solution involves building controls directly into AI systems that automatically filter what data gets shared, creating audit trails. Trust is becoming the deciding factor in whether banks can move beyond limited AI pilots.

TechInAsia maps Australia's AI startup landscape, showing key players and investment trends. The report highlights active investors and funding patterns in the sector, providing a visual overview for professionals tracking the regional AI ecosystem.

Chinese cybersecurity firm 360 has unveiled AI tools it claims can match Anthropic's Mythos system, which automatically discovers software vulnerabilities. The U.S. banned exports of Mythos over national security fears, prompting China to develop its own version. 360 says its 'Tulongfeng' tool found over 3,400 vulnerabilities. This race matters because vulnerability-finding AI can be used both to hack systems and to defend them—raising stakes for global cybersecurity.

India's central bank has proposed new guidelines requiring banks to actively manage risks from artificial intelligence, including model failures, algorithmic bias, and data privacy concerns. With minimal content available from this source, the full scope of these requirements remains unclear.

AI is guzzling way more energy and water than most people realize. Global data centers already use more electricity than most countries — and it's set to double in four years. By 2030, just cooling these centers could need enough water to supply the entire world for nearly two years. The fix? Use AI less for simple tasks (yes, you can still open a book instead of asking ChatGPT for a cookie recipe), and opt out of AI search results where you can.

A hobbyist ran an experiment putting seven leading AI models — including GPT-5.5, Gemini 3.1, Claude Opus, and DeepSeek-V4 — in a virtual betting competition on the 2026 World Cup. Each model got $1 million in fake bankroll and had to predict match outcomes before seeing betting odds. Singapore ranked 5th among 46 countries watching the live experiment, which tracked every model's reasoning and accuracy across 104 fixtures.

Nvidia's top AI chips are now selling for double their original price in China, with some servers fetching over $1 million due to US export restrictions creating a black-market premium. Chinese AI firms are still desperate for hardware despite tighter enforcement. On a separate track, Nvidia just revealed a new liquid-cooling system that could eliminate nearly all water use inside AI data centers — a direct response to mounting concerns that AI's infrastructure needs are becoming unsustainable.

A sweeping look at the next generation of computing beyond traditional silicon. AI is now optimizing chip design and quantum algorithms, while breakthrough technologies like DNA storage (potentially terabytes in a single drop of water), neuromorphic chips mimicking brain function, and quantum computing are converging. This hybrid ecosystem could unlock hundreds of billions to trillions in economic value by the mid-2030s.

Apple may turn to Intel to manufacture some of its chips, as TSMC struggles to keep up with surging AI chip demand. While strategically logical given US government incentives for domestic chipmaking, analysts say any advanced Intel chip for Apple is 2-3 years away at best. Intel's track record with chip yields remains a concern, and Apple may start with lower-end products before committing its flagship designs.

Agility Robotics, which makes humanoid robots named Digit for warehouses, is going public on Wall Street at a $2.5 billion valuation. Backed by Amazon, Nvidia, and SoftBank, the company bets that firms struggling to hire workers for repetitive warehouse jobs will pay for robots. Early customers include Toyota and Latin American e-commerce giant Mercado Libre.

Accenture's latest survey shows companies are pouring money into AI but struggle to update how they actually run their operations around it. The gap between AI investment and working organizational structures is creating inefficiencies — businesses buy the tools but haven't rethought their workflows to match.

When public cloud computing emerged in the late 2000s and began scaling through the early 2010s, the prevailing theory was straightforward. Cloud computing theory claimed nearly every workload would eventually migrate off-premises. The economics were compelling, the convenience undeniable, and the momentum felt unstoppable. More than 15 years later, enterprise IT leaders are still managing substantial on-premises infrastructure — and many are actively investing in upgrades. Today, various research reports estimate that between 35 and 50 percent of workloads have moved to the cloud. Whether that figure is above or below 50 percent, it’s clear that workloads remain distributed. Cost, data sensitivity, regulatory compliance, latency requirements, and operational control all shape where a give

[![Image 1: Daniel Howley](https://s.yimg.com/ny/api/res/1.2/P_cNtA9CwtisPd35jvZXVg--/YXBwaWQ9aGlnaGxhbmRlcjt3PTgwO2g9ODA7Y2Y9d2VicA--/https://o.aolcdn.com/images/dims?image_uri=https%3A%2F%2Fo.aolcdn.com%2Fimages%2Fdims%3Fimage_uri%3Dhttps%253A%252F%252Fo.aolcdn.com%252Fimages%252Fdims%253Fimage_uri%253Dhttps%25253A%25252F%25252Fo.aolcdn.com%25252Fimages%25252Fdims%25253Fimage_uri%25253Dhttps%2525253A%2525252F%2525252Fo.aolcdn.com%2525252Fimages%2525252Fdims%2525253Fimage_uri%2525253Dhttps%252525253A%252525252F%252525252Fs.yimg.com%252525252Fos%252525252Fcreatr-images%252525252FGLB%252525252F2017-10-19%252525252F2021c8a0-b4fd-11e7-9b93-f3d4e83a491a_d_howley.jpg%25252526compress%2525253D1%25252526progressive%2525253D1%25252526quality%2525253D75%25252526client%2525253Dhawkeye%25252526signat

Agility Robotics, the first company to deploy humanoid robots commercially in warehouses, is going public via a $2.5 billion SPAC deal. Their Digit robot has already logged 65,000 hours working at real companies like Toyota and auto-parts maker Schaeffler, moving over 100,000 totes. They're launching a new safer version with $300 million in pre-orders. For everyday readers, this means warehouse and factory robots are no longer just prototypes—some are actually earning money and creating jobs alongside humans.

AI company Anthropic has accused Chinese tech giant Alibaba of stealing technology behind its Claude AI assistant. Anthropic calls it the largest known "distillation attack" against the company—essentially using AI to reverse-engineer how Claude works without permission. This raises serious questions about AI intellectual property protection and whether companies can truly safeguard their models from being copied.

California residents are suing major fuel retailers including 7-Eleven, Circle K, BP, Marathon, Walmart and Albertsons, alleging they used AI pricing software called Kalibrate to illegally coordinate gas prices across the state. The lawsuit claims the software helps stations avoid competing on price and even warns against lowering prices, calling it a "downward spiral." California already has the nation's highest gas prices at $5.54 per gallon on average. In areas using the software, prices rose about 6 cents on average—and up to 30 cents in areas with high concentration of participating stations. The state recently passed a law specifically banning algorithm-based price coordination.

A second lawsuit (filed same week) expands on allegations that major gas retailers used Kalibrate AI software to fix fuel prices across California. The software automatically adjusts prices based on competitor data and pushes changes directly to pumps without human review. The expanded lawsuit names more defendants and provides additional detail on how the coordinated pricing allegedly worked.

A DoorDash delivery robot named Dot recently rolled into an active SWAT operation in Arizona while making a food delivery, refused to move when police ordered it to retreat, and stayed to watch officers detonate a flashbang before eventually being loaded onto a truck. DoorDash says the robot "behaved as designed" by stopping and waiting. This isn't the first such incident—similar delivery robots from Serve Robotics have also blundered into police scenes this year. The problem: autonomous delivery bots lack the judgment to interpret police commands or recognize when they should vacate an area.

Researchers have found that medical AI systems can leak sensitive patient information through a technique called 'membership inference attacks.' Hackers could potentially determine if a specific person's data was used to train an AI model, exposing private health details like whether someone has cancer. The risk is higher for underrepresented groups in medical datasets.

Qualcomm, best known for smartphone chips, is shifting strategy to tap into the AI market. The company sees growth opportunities beyond mobile devices as businesses increasingly adopt AI tools. The move reflects a broader trend where traditional chipmakers are pivoting to capture AI-related demand.