Singapore AI News & Daily Briefing

Shopee is cutting about 8% of its developer workforce globally, including product and engineering staff in Singapore, as the e-commerce giant pivots more resources into AI. Affected employees are reportedly getting one month's salary per year of service plus two months' extra.

Unlike in the West, Southeast Asia's software industry is less threatened by AI because the market is still underpenetrated and relies on local relationships. Singapore dominates regional SaaS funding, and investor interest has shifted from pure-play SaaS to AI, but incumbent firms still have breathing room.

AI salaries in Singapore jumped 15-25% over the past year—five times faster than average wage growth—with fresh graduates landing S$70,000 to S$90,000 jobs. Chinese tech firms are offering PhDs up to S$200,000 to work in China, while a bachelor's degree and hands-on experience are enough for most local roles.

Quantum computers may seem futuristic, but they need massive classical computing support to function. Researchers are developing AI tools to automatically calibrate quantum hardware and decode errors in real time—tasks that currently require PhD-level expertise and take days or weeks. Companies like Nvidia, IBM, Google, and Q-CTRL are racing to automate these processes as qubit counts scale up.

OpenAI reports that groups linked to China are using AI to spread fake information about US tech policies, data centers, and ChatGPT itself. These influence operations aim to distort public debate and push false narratives. AI makes it cheap and easy to generate convincing disinformation, so be skeptical of online arguments about tech issues.

Researchers at the University of Twente have discovered a way to cut up to 14% of the energy used to train large AI models by adjusting GPU clock speeds with fine precision. By slowing memory clocks during heavy computation phases and vice versa, they achieved energy savings with only a 0.6% slowdown in training time. GPT-4 reportedly consumed enough electricity to power 5,000 American homes for a year during training.

London Stock Exchange Group shares how deploying OpenAI tools across its 27,000 employees cut product release cycles from months to weeks and boosted analyst productivity. The key was rethinking entire workflows, not just automating tasks. This real-world example shows how even heavily regulated industries can transform their pace with AI.

OpenAI has released a set of early-stage policy ideas for the 'Intelligence Age,' aiming to ensure advanced AI benefits everyone. They're offering research grants up to US$1 million and hosting discussions to gather feedback, kicking off a conversation about how to build institutions that share AI's prosperity widely.

OpenAI and Oracle are teaming up so that Oracle cloud customers can use their existing credit commitments to pay for OpenAI models. This removes a major purchasing hurdle for large companies looking to adopt AI. If your employer uses Oracle, you could soon access advanced AI tools through your existing cloud plan.

Google DeepMind released DiffusionGemma, an experimental AI model that generates text in blocks instead of word by word, making it up to 4x faster on personal GPUs. It trades some output quality for speed, perfect for real-time tasks like editing and coding on your own device. This means your computer could run advanced AI assistants without needing the cloud.

New engineers should focus on fundamentals, system design, and communication skills to stay relevant as AI tools become common. AI won't replace you if you learn to work with it—use it as a teammate but keep your own judgment sharp.

An astrophysicist is using OpenAI's Codex to invent new algorithms for simulating the chaotic plasma around black holes. By skipping tedious calculations of individual particle spirals, these AI-generated methods could unlock simulations that were previously impossible. This shows how AI can crack scientific problems that even supercomputers struggle with.

Chinese government-backed hackers are the biggest state-sponsored cyber threat to technology companies, according to a new CrowdStrike report. Criminal hacking groups drive 65% of hands-on-keyboard attacks on tech firms, but China-linked groups dominate government-ordered espionage campaigns against the sector.

IEEE recognized students and professors for community engineering projects, including an AI-powered emotion-recognition wearable for visually impaired people and an IoT system for Thai rice farmers that cuts water use by 63 percent and methane emissions by 7 percent annually.

Utilities are modernizing electrical grids to handle renewable energy, electric vehicles, and extreme weather. AI and sensors help manage power distribution in real-time while reducing outages - Georgia Power cut outages by 76 percent using these methods.

ZDNet puts Apple's upgraded Siri through its paces on Mac, testing its new AI capabilities powered by Apple Intelligence. Early impressions suggest Apple is making progress but still has room to improve before matching the most advanced AI assistants on the market.

Microsoft patched two dangerous zero-day vulnerabilities that a researcher called Nightmare Eclipse had exposed after a dispute over a vulnerability disclosure deal. The researcher had released proof-of-concept exploit code that could let attackers gain full system control, forcing Microsoft to act fast. One fix addresses a flaw in Windows' translation framework that could bypass security protections. Your Windows PC likely received this update — if not, do it now.

A single typo — an extra exclamation mark in the Linux kernel's firewall code — created a critical hole that lets untrusted users become root administrators. The bug sits in nf_tables, the system Linux uses to manage firewall rules, and works on Debian and Ubuntu. Patches exist but need to be applied. This is the third serious Linux privilege-escalation bug discovered in weeks.

A stunning 82% of Singapore companies have pulled back or shut down live AI customer service bots after deploying them—eight points above the global average. This happens even though Singapore leads Asia-Pacific in AI adoption with a 72% deployment rate. The problem: AI fails, support teams get overwhelmed, and most companies lack proper governance guardrails to keep AI in check. Only 27% of Singapore firms report mature governance despite 75% prioritising trust and security. The takeaway: deploying AI is easy; making it actually work is another story.

Singapore workers are among the world's fastest AI adopters, with 66 percent of AI users producing work they couldn't have done a year ago, versus 58 percent globally. Singapore ranks second globally on AI diffusion, and 78 percent of AI users recognize the urgency of adapting quickly. However, only 24 percent say their leadership is aligned on AI strategy — creating a 'Transformation Paradox' where employees are moving faster than their organizations. Microsoft is urging companies to redesign workflows and decision-making processes to capture the value workers are already creating.

London and Singapore-based Galaxy Data Center has secured $250 million in strategic financing to build AI-ready data centers across Southeast Asia. The company plans to deploy high-density, GW-scale AI data center campuses using green energy solutions. Singapore serves as its regional hub for coordinating multi-country expansion. This investment reflects the massive appetite for AI computing infrastructure as companies race to build out capacity for generative AI and cloud services.

Singapore's first cybersecurity chief David Koh is stepping down on July 1 after 11 years leading the Cyber Security Agency of Singapore. Gwenda Fong, currently deputy secretary at the Ministry of Digital Development and Information, will take over. Under Koh's leadership, Singapore built its national cybersecurity framework from scratch—including the 2018 Cybersecurity Act, the 2016 and 2021 national strategies, and international diplomatic initiatives. He also managed major breaches like the 2018 SingHealth attack and the 2025 UNC3886 espionage campaign on telecom infrastructure. For everyday Singaporeans, this transition matters because CSA protects the critical services you rely on daily—power, banking, transport, and telecom—from hackers who could disrupt your life or steal your personal data.

Work platform monday.com has appointed Chris Jordan as Regional VP for Southeast Asia and Country Manager for Singapore, establishing the city-state as its second Asia-Pacific headquarters after Sydney. The company has built a team of over 30 people in Singapore since October 2025, recording 23 percent year-on-year growth across Southeast Asia and serving more than 4,700 regional customers. Jordan brings over 20 years of APAC experience, including nearly a decade at Salesforce. Nearly half of Southeast Asian companies have moved beyond AI pilots, with Singapore leading at 56 percent enterprise AI adoption.

Singapore AI startup Acrab has raised over $350 million from early investors, including Temasek's venture arm Vertex, to build specialized computing hardware for AI agents. Founded in 2024, Acrab is developing a complete system including custom AI chips, local language model processing, and agent coordination software—all designed to run AI tasks on local devices instead of cloud servers. The idea is that AI agents will eventually work better on your own hardware than relying on distant data centers. For everyday users, this could mean faster, more private AI experiences on phones and computers.

Malaysia-founded respond.io has raised $62.5 million in Series B funding to expand its AI-powered customer messaging platform globally. The company helps businesses manage conversations across WhatsApp, TikTok, Instagram, and other channels using AI agents that work alongside human staff. With $35 million in annual recurring revenue and 169% year-on-year growth, the firm now serves over 10,000 businesses across 180 countries. The funding will fuel its push into North America and Europe while continuing to embed AI deeper into customer service workflows.

Interpol's latest Asia-Pacific cyberthreat report calls for stronger whole-of-society responses as criminals increasingly use AI to power scams. Over 6.5 billion cyber threats were detected in the region in 2024. Online scams and phishing remain the most common, but criminals now use AI-generated deepfakes, targeted spear-phishing, and automated messages at industrial scale. In Singapore, victims lost $913.1 million to scams in 2025, including a finance director who transferred $670,000 after a deepfake video call impersonated company executives. Interpol recommends better cloud security, continuous user education, and real-time threat intelligence sharing with industry partners.

Endowus, a Singapore-based wealth management platform backed by Vertex, has reached over $350 million in total financing. The article examines their financials—revenue doubled in 2023 to $15.4 million, though losses also widened to $23.7 million as the company chases profitability.

Plaud, an AI note-taking app maker, has hit $100 million in annual recurring revenue, a milestone that puts it in rare company among AI startups. Singapore plays a key role in its operations as a major R&D hub, though the company is headquartered elsewhere.

[ST Telemedia Global Data Centers (STT GDC)](https://www.sttelemediagdc.com/), a global data center provider, announced Tuesday the opening of STT Seoul 1, its first data center in South Korea. Developed and operated through a joint venture between STT GDC (60 percent) and Hyosung Heavy Industries (40 percent), the facility establishes a strategic foothold for STT GDC in one of Asia’s most advanced digital markets, extending STT GDC’s platform to support customers deploying artificial intelligence (AI) and cloud infrastructure across Northeast Asia, it said in a statement. According to the statement, South Korea is a strategically important market for high-performance digital infrastructure, underpinned by strong enterprise digitalization and growing demand for compute capacity. The openin

# London mayor Sadiq Khan launches 'S'pore-style' housing approach to build 'affordable homes' - Mothership.SG - News from Singapore, Asia and around the world %20AppleWebKit%2F537.36%20(KHTML%2C%20like%20Gecko)%20Chrome%2F148.0.0.0%20Safari%2F537.36&date=1781709905481) [](https://mothership.sg/) * [News](https://mothership.sg/category/news) * [Lifestyle](https://mothership.sg/category/nova) * [Politics](https://mothership.sg/category/parliament) * [Earth](https://mothership.sg/category/environment) * [World](https://mothership.sg/category/abroad) * [Features](https://mothership.sg/category/features) * [Careers](https://mothership.sg/careers) [![Image 6: Search](https://mothership.sg/assets/imag

Intel has committed an additional $2.6 billion to Vietnam, bringing its total investment there to $4.1 billion. The chipmaker originally walked away from a $3.3 billion Vietnam project in 2023 when the country couldn't offer the cash subsidies Intel wanted — that money went to Poland and Germany instead. Both those European projects have since been cancelled. Vietnam later created a new investment support fund to attract semiconductor manufacturers after adopting the global minimum tax, which neutralized old tax incentives. Intel now employs 6,500 people in Vietnam and generated $11.67 billion in exports in 2025.

Top AI executives including OpenAI's Sam Altman, Google DeepMind's Demis Hassabis, and Anthropic's Dario Amodei gathered at the G7 summit in France to discuss AI deployment, amid European calls for tech sovereignty. The meeting comes after the U.S. government barred non-Americans from accessing Anthropic's most advanced models, highlighting how allies can be cut off from critical AI infrastructure. European leaders are pushing for homegrown AI alternatives to reduce dependence on American companies, with Canada's prime minister calling for diversified AI ecosystems. French President Macron has even required civil servants to use homegrown video conferencing tools instead of Zoom and Microsoft Teams.

Tokyo-based Sakana AI has raised $135 million in a Series B round led by Mitsubishi UFJ Financial Group, Khosla Ventures, and New Enterprise Associates. The company develops AI models tailored for Japan's needs, focusing on energy efficiency and sustainability, with partnerships in finance, defense, and manufacturing. Sakana AI was founded about two years ago and aims to address Japan's economic and demographic challenges through specialized AI technology.

Security giants Check Point and Illumio are teaming up to tackle a new breed of AI-powered cyber threats. The partnership combines Check Point's firewall defenses with Illumio's microsegmentation technology, which essentially creates walls inside corporate networks to stop hackers from moving freely if they break in. Research shows AI is now enabling attacks that used to take weeks to happen in minutes, leaving security teams almost no time to respond. The companies say this integrated approach gives enterprises a better shot at both blocking attacks and containing damage when breaches happen.

Chinese AI lab DeepSeek is raising its first funding round at a reported $45 billion valuation, up from $20 billion just weeks ago. The China Integrated Circuit Industry Investment Fund may lead the round, with Tencent and Alibaba also in talks. DeepSeek's models perform on par with top US models like GPT-4 and Claude but cost significantly less to run. The company releases open-weight models and has built its V4 to work with Chinese chips like Huawei's Ascend series, helping China build AI despite US chip export restrictions.

OpenAI, the maker of ChatGPT, has confidentially filed for a US stock market listing with the SEC. The company is working with Goldman Sachs and Morgan Stanley on a potential IPO as soon as autumn 2026. A March 2026 funding round valued the company at $852 billion. OpenAI also plans a share tender sale to let employees sell some stock before the IPO. The filing gives investors their first official peek into the finances of one of the world's most valuable private AI companies. Why it matters: If OpenAI goes public, regular people could finally invest in the company behind ChatGPT. It also signals AI companies are maturing into mainstream corporate structures.

Update on G7 leaders vow closer ties on AI as they hash out 'trusted partners' scheme. Coverage via Yahoo Singapore Tech.

Update on In US, EU mutual interest for Europe to use best AI models, von der Leyen says. Coverage via Yahoo Singapore Tech.

Analysts are examining whether Qualcomm can position itself to benefit from the rise of agentic AI—AI systems that independently complete complex tasks. The question is whether Qualcomm's mobile chip expertise translates to the new wave of AI computing demands, or if it risks being left behind as AI processing shifts.

Kuaishou's AI video tool Kling hit $240 million in annual revenue run rate by December 2025, doubling from $100 million in under a year. Over 60 million creators globally now use Kling to generate videos for ads, ecommerce, film, and animation.

Southwest Airlines has always prided itself on being different. For 54 years, the carrier built its identity around open seating, low fares and a scrappy servant's heart. But staying competitive required a reckoning with the technology foundations underneath. "We really needed to follow our customer," said Lauren Woods, the airline's Executive Vice President and Chief Information Officer. And following the customer meant reinventing nearly everything. **From Crisis to Catalyst** Woods stepped into the CIO role in early 2023, weeks after a December 2022 winter storm exposed the fragility of Southwest's crew scheduling systems and triggered mass cancellations. The crisis clarified what needed to change. "Once you're out of the crisis mode, it becomes: how do we not get into this situation ev

Alibaba Cloud is opening its first data centers in Brazil, France, and the Netherlands, with plans to expand to Mexico, Japan, South Korea, Malaysia, and Dubai next year. The company already operates 91 availability zones across 29 regions. As part of its push, it launched an AI Catalyst Program offering up to US$120,000 in cloud credits to selected AI startups globally. This expansion means more cloud infrastructure options for businesses worldwide, though the move also highlights intensifying competition among global cloud providers.

Companies are increasingly scrutinizing the costs of running AI operations as 'tokenmaxxing' — the practice of maximizing token usage — drives up spending. The article examines how enterprises are grappling with mounting AI bills as they deploy AI more broadly across operations. For everyday businesses considering AI tools, this signals that costs can add up quickly, potentially affecting pricing of AI-powered products and services.

OpenAI CEO Sam Altman is in talks with the White House about giving the US government a stake in the company, potentially through a sovereign wealth fund-style vehicle. Discussions began over a year ago with the Trump administration. Under OpenAI's restructuring plan, its nonprofit will retain control while converting its operating subsidiary into a public benefit corporation. The government stake idea comes as OpenAI restructures to allow employees and investors to hold equity while keeping nonprofit majority ownership.

Private equity giant Thoma Bravo has partnered with Google Cloud to roll out AI tools across its 580+ portfolio software companies, including cybersecurity firms Ping Identity, Darktrace, Sophos, and Proofpoint. Portfolio firms get access to Google's AI models, engineers, and sales channels via the Google Cloud Marketplace. The move could accelerate consolidation among cybersecurity vendors as large customers seek integrated AI-powered security platforms. Social engineering attacks using AI have surged 135% since ChatGPT's launch, making this partnership timely for fighting AI-driven threats.

OpenAI launched Lockdown Mode for ChatGPT, a security setting that disables web browsing and network features for high-risk users. The mode prevents prompt-injection attacks where hidden malicious instructions in files or webpages trick AI into sending out sensitive data.

SK hynix, the South Korean memory chipmaker, is being flooded with investment offers from big tech companies desperate to lock up AI memory supplies. Companies like Microsoft and Google are offering to fund dedicated production lines and even pay for manufacturing equipment like ASML machines. New deals include upfront payments of 10-30% of contract value and price floor clauses - changes from the old model where price protection was the main goal. SK hynix has no spare capacity left at its Yongin plant.

Databricks has crossed a $5.4 billion revenue run-rate with 65% year-on-year growth in Q4. The company raised over $7 billion including $5 billion in equity at a $134 billion valuation, with backing from JPMorgan, Goldman Sachs, Microsoft, and Qatar Investment Authority. Its AI products alone generate $1.4 billion. The funding will support Lakebase, a serverless database for AI applications, and Genie, a conversational AI tool for data queries.

Nvidia's new Vera CPU—its first processor built specifically for AI agents—will use memory chips from South Korea's SK Hynix. Nvidia CEO Jensen Huang announced the partnership after dinner with SK Group leadership in Seoul, saying the company had a big year with SK Hynix and is preparing for a "very, very large" second half. This matters because the memory-CPU partnership directly affects how fast and powerful AI systems will be, whether you're using smart assistants, automation tools, or cloud-based services. Better hardware means snappier AI experiences across the board.

Nvidia and LG Group are expanding their partnership to cover robotics, AI data centers, and mobility tech. Nvidia CEO Jensen Huang met LG's chairman in Seoul to outline collaboration across multiple LG affiliates, including robotics, data-center cooling, and AI software. LG Electronics already uses Nvidia's robotics platform and recently unveiled a home robot built on it.

South Korea's LG Innotek, a major iPhone camera module supplier, is expected to benefit from AI-driven iPhone demand as Apple expands its AI ecosystem. KB Securities maintained a buy rating with profit forecasts of $858 million this year and $1.06 billion in 2027. Higher premium iPhone mix should lift LG Innotek's camera business. The company is also developing FC-BGA semiconductor substrates at its Gumi plant for high-end servers, aiming to enter that market by 2026.

Banks and fintech lenders are increasingly using AI agents to automate loan processing—from reading documents and checking creditworthiness to monitoring compliance and flagging risks. A new guide breaks down how these AI systems work, which parts of the lending cycle they touch, and what benefits lenders can expect, like faster approvals and lower costs. The guide also covers common hurdles like outdated tech systems and staff resistance.

A Silicon Valley startup called Integral AI, founded by former Google researchers, is bringing AI-powered industrial robots to Japan. The company wants robots that can understand simple commands like 'make a coffee' and teach themselves the task without pre-programming. Japan supplies about 29% of global industrial robots. This signals a shift from AI that talks to AI that works with its hands.

Eyewear giant EssilorLuxottica and semiconductor equipment maker Applied Materials have partnered on augmented reality optics technology. HSBC projects the smart glasses market could reach $200 billion by 2040 as AR glasses evolve into mainstream consumer devices. This partnership combines EssilorLuxottica's optical expertise with Applied Materials' chip-making capabilities to solve the tough problem of making AR displays lightweight and comfortable enough for everyday wear.

Chinese AI startups Zhipu AI and MiniMax have released new flagship AI models designed for coding tasks, both preparing for IPOs in Hong Kong. Zhipu's GLM-4.7 model reportedly matched Anthropic's Claude performance on a coding benchmark, while MiniMax released its M2.1 model. Both companies claim their models offer competitive pricing for Asian enterprise teams. Zhipu's revenue grew over fivefold from 2022 to 2024, though it spent roughly seven times its revenue on R&D, raising questions about profitability.

At France's Vivatech tech conference, European governments are pushing for AI systems built on home soil rather than relying on American or Chinese platforms, citing data sovereignty and security concerns.

Anthropic CEO Dario Amodei admitted he doesn't know whether his company's Claude AI was used in a US missile strike on an Iranian elementary school that killed about 120 children in February. Claude is embedded in the Pentagon's Maven Smart System targeting platform, which reportedly helped strike 13,000 targets in just five weeks during US operations against Iran. Amodei claims 'human decision-making' was maintained, but experts warn the speed and volume of AI-assisted targeting makes meaningful human review nearly impossible. Anthropic is currently suing the Trump administration over being blacklisted by the Pentagon.

Nvidia researchers have developed a system where robots teach themselves new tasks by using AI coding agents. Eight robots in a fleet achieved up to 99 percent accuracy on difficult grasping tasks without humans manually programming each movement. The robots essentially write and refine their own training code, marking a step toward more autonomous robotic learning.

OpenAI researchers are developing ways to predict how often a new AI model will make mistakes once released to the public. Current safety tests don't fully reveal failure rates in real-world use. This new method could help catch problems before launch, though it's still a research proposal rather than an implemented tool.

OpenAI connected GPT-5.4 to an automated lab to improve a difficult chemical reaction used in drug discovery. The AI proposed using an oxidant called TEMPO, which boosted yields for 88% of tested compounds. Human chemists validated the results. This shows AI can go beyond theory to actually help scientists discover better chemical reactions, potentially speeding up drug development.

Over 200 human rights groups including Amnesty International are demanding governments and tech companies stop using AI for military targeting, calling it a 'death conveyor' operating at unprecedented speed. They point to Palantir's Maven Smart System (using Anthropic's Claude AI) reportedly identifying over 1,000 targets in just 24 hours during US-Iran operations. The coalition wants an immediate halt to AI tools in target selection and full transparency on how these systems are being used in combat. Human Rights Watch warns no international law currently covers AI in warfare, leaving serious legal gaps.

The Pentagon has revealed that Elon Musk's Grok AI was used to enable over 2,000 munitions strikes against Iran in just 96 hours during Operation Epic Fury, after Anthropic's Claude was pulled from the same military targeting program called Project Maven. A Pentagon AI chief confirmed under oath that Grok replaced Claude and handled the rapid targeting cycle. The disclosure has reignited debate over which AI companies will or won't power military operations, with Anthropic previously stating it restricts its AI from certain weapons uses.

Meta is rolling out AI search powered by Facebook posts, which means answers to your questions could come from years-old, unverified social media updates. Early tests show the feature confidently gives wrong information. Think of it like asking a stranger at a party for directions—sometimes helpful, often unreliable.

Pramaana Labs, a startup focused on making AI systems mathematically provably correct, raised $27 million in seed funding from Khosla Ventures. The company targets high-stakes fields like law, drug discovery, and tax preparation, where AI errors carry real financial and legal consequences. Their 'formal verification' approach mathematically proves AI outputs are reliable, rather than just testing them statistically.

Investigations by The Atlantic have confirmed that roughly 21 million copyrighted songs — including hits by Taylor Swift and Bad Bunny — were used to train AI music generators like Suno and Udio without permission. Major labels Sony, UMG, and Warner are suing for up to $150,000 per song, with the total potential damages running into billions. Courts have yet to rule on whether AI training on copyrighted material constitutes fair use. Researchers at the University of Tennessee have developed a tool called HarmonyCloak that lets artists add inaudible noise to recordings, making them unlearnable by AI while sounding normal to human ears.

Major tech companies like Microsoft, Amazon, Google, Meta, and Oracle are spending 70% more on AI infrastructure yearly, but their cash flow is only growing 23%. If this continues, they won't be able to fund AI data centers from operations alone by late 2026. Some are already seeking outside investment to cover costs.

General Motors is using AI and simulation to slash car development time from four years down to two. Former Tesla Autopilot chief Sterling Anderson is leading the charge, using physics-based AI models that can predict how design changes affect a vehicle in under a minute instead of 15 hours. The electric GMC Hummer already benefited from this approach, and the tech is even being applied to NASA's lunar rover.

Companies like Sureel and SoundVerse are building systems to track how AI models use musicians' work so artists can get paid royalties. Warner Music acquired Sureel, which labels music files with usage instructions and tracks AI training. The challenge is figuring out how much each piece of music influenced an AI output, since unlike streaming, AI training happens once but its effects last forever. Some experts argue simpler negotiated deals might work better than complex attribution systems.

Anthropic is expanding its Project Glasswing cybersecurity initiative to 150 additional partners across 15+ countries. The program gives vetted organizations early access to Claude AI models to hunt for software vulnerabilities before malicious actors can exploit them. Partners have already uncovered more than 10,000 critical security flaws since the April pilot. The company is also extending Mythos access to European organizations.

Plaud, an AI note-taking app maker, has hit $100 million in annual recurring revenue, a milestone that puts it in rare company among AI startups. Singapore plays a key role in its operations as a major R&D hub, though the company is headquartered elsewhere.

Chinese AI lab DeepSeek is raising its first funding round at a reported $45 billion valuation, up from $20 billion just weeks ago. The China Integrated Circuit Industry Investment Fund may lead the round, with Tencent and Alibaba also in talks. DeepSeek's models perform on par with top US models like GPT-4 and Claude but cost significantly less to run. The company releases open-weight models and has built its V4 to work with Chinese chips like Huawei's Ascend series, helping China build AI despite US chip export restrictions.

SK hynix, the South Korean memory chipmaker, is being flooded with investment offers from big tech companies desperate to lock up AI memory supplies. Companies like Microsoft and Google are offering to fund dedicated production lines and even pay for manufacturing equipment like ASML machines. New deals include upfront payments of 10-30% of contract value and price floor clauses - changes from the old model where price protection was the main goal. SK hynix has no spare capacity left at its Yongin plant.

Databricks has crossed a $5.4 billion revenue run-rate with 65% year-on-year growth in Q4. The company raised over $7 billion including $5 billion in equity at a $134 billion valuation, with backing from JPMorgan, Goldman Sachs, Microsoft, and Qatar Investment Authority. Its AI products alone generate $1.4 billion. The funding will support Lakebase, a serverless database for AI applications, and Genie, a conversational AI tool for data queries.

South Korea's LG Innotek, a major iPhone camera module supplier, is expected to benefit from AI-driven iPhone demand as Apple expands its AI ecosystem. KB Securities maintained a buy rating with profit forecasts of $858 million this year and $1.06 billion in 2027. Higher premium iPhone mix should lift LG Innotek's camera business. The company is also developing FC-BGA semiconductor substrates at its Gumi plant for high-end servers, aiming to enter that market by 2026.

A Silicon Valley startup called Integral AI, founded by former Google researchers, is bringing AI-powered industrial robots to Japan. The company wants robots that can understand simple commands like 'make a coffee' and teach themselves the task without pre-programming. Japan supplies about 29% of global industrial robots. This signals a shift from AI that talks to AI that works with its hands.

Eyewear giant EssilorLuxottica and semiconductor equipment maker Applied Materials have partnered on augmented reality optics technology. HSBC projects the smart glasses market could reach $200 billion by 2040 as AR glasses evolve into mainstream consumer devices. This partnership combines EssilorLuxottica's optical expertise with Applied Materials' chip-making capabilities to solve the tough problem of making AR displays lightweight and comfortable enough for everyday wear.

OpenAI CEO Sam Altman is in talks with the White House about giving the US government a stake in the company, potentially through a sovereign wealth fund-style vehicle. Discussions began over a year ago with the Trump administration. Under OpenAI's restructuring plan, its nonprofit will retain control while converting its operating subsidiary into a public benefit corporation. The government stake idea comes as OpenAI restructures to allow employees and investors to hold equity while keeping nonprofit majority ownership.

Private equity giant Thoma Bravo has partnered with Google Cloud to roll out AI tools across its 580+ portfolio software companies, including cybersecurity firms Ping Identity, Darktrace, Sophos, and Proofpoint. Portfolio firms get access to Google's AI models, engineers, and sales channels via the Google Cloud Marketplace. The move could accelerate consolidation among cybersecurity vendors as large customers seek integrated AI-powered security platforms. Social engineering attacks using AI have surged 135% since ChatGPT's launch, making this partnership timely for fighting AI-driven threats.

Banks and fintech lenders are increasingly using AI agents to automate loan processing—from reading documents and checking creditworthiness to monitoring compliance and flagging risks. A new guide breaks down how these AI systems work, which parts of the lending cycle they touch, and what benefits lenders can expect, like faster approvals and lower costs. The guide also covers common hurdles like outdated tech systems and staff resistance.

Chinese AI startups Zhipu AI and MiniMax have released new flagship AI models designed for coding tasks, both preparing for IPOs in Hong Kong. Zhipu's GLM-4.7 model reportedly matched Anthropic's Claude performance on a coding benchmark, while MiniMax released its M2.1 model. Both companies claim their models offer competitive pricing for Asian enterprise teams. Zhipu's revenue grew over fivefold from 2022 to 2024, though it spent roughly seven times its revenue on R&D, raising questions about profitability.

Singapore AI startup Acrab has raised over $350 million from early investors, including Temasek's venture arm Vertex, to build specialized computing hardware for AI agents. Founded in 2024, Acrab is developing a complete system including custom AI chips, local language model processing, and agent coordination software—all designed to run AI tasks on local devices instead of cloud servers. The idea is that AI agents will eventually work better on your own hardware than relying on distant data centers. For everyday users, this could mean faster, more private AI experiences on phones and computers.

Endowus, a Singapore-based wealth management platform backed by Vertex, has reached over $350 million in total financing. The article examines their financials—revenue doubled in 2023 to $15.4 million, though losses also widened to $23.7 million as the company chases profitability.

Nvidia's new Vera CPU—its first processor built specifically for AI agents—will use memory chips from South Korea's SK Hynix. Nvidia CEO Jensen Huang announced the partnership after dinner with SK Group leadership in Seoul, saying the company had a big year with SK Hynix and is preparing for a "very, very large" second half. This matters because the memory-CPU partnership directly affects how fast and powerful AI systems will be, whether you're using smart assistants, automation tools, or cloud-based services. Better hardware means snappier AI experiences across the board.

Nvidia and LG Group are expanding their partnership to cover robotics, AI data centers, and mobility tech. Nvidia CEO Jensen Huang met LG's chairman in Seoul to outline collaboration across multiple LG affiliates, including robotics, data-center cooling, and AI software. LG Electronics already uses Nvidia's robotics platform and recently unveiled a home robot built on it.

OpenAI, the maker of ChatGPT, has confidentially filed for a US stock market listing with the SEC. The company is working with Goldman Sachs and Morgan Stanley on a potential IPO as soon as autumn 2026. A March 2026 funding round valued the company at $852 billion. OpenAI also plans a share tender sale to let employees sell some stock before the IPO. The filing gives investors their first official peek into the finances of one of the world's most valuable private AI companies. Why it matters: If OpenAI goes public, regular people could finally invest in the company behind ChatGPT. It also signals AI companies are maturing into mainstream corporate structures.

Kuaishou's AI video tool Kling hit $240 million in annual revenue run rate by December 2025, doubling from $100 million in under a year. Over 60 million creators globally now use Kling to generate videos for ads, ecommerce, film, and animation.

OpenAI launched Lockdown Mode for ChatGPT, a security setting that disables web browsing and network features for high-risk users. The mode prevents prompt-injection attacks where hidden malicious instructions in files or webpages trick AI into sending out sensitive data.

At France's Vivatech tech conference, European governments are pushing for AI systems built on home soil rather than relying on American or Chinese platforms, citing data sovereignty and security concerns.

Top AI executives including OpenAI's Sam Altman, Google DeepMind's Demis Hassabis, and Anthropic's Dario Amodei gathered at the G7 summit in France to discuss AI deployment, amid European calls for tech sovereignty. The meeting comes after the U.S. government barred non-Americans from accessing Anthropic's most advanced models, highlighting how allies can be cut off from critical AI infrastructure. European leaders are pushing for homegrown AI alternatives to reduce dependence on American companies, with Canada's prime minister calling for diversified AI ecosystems. French President Macron has even required civil servants to use homegrown video conferencing tools instead of Zoom and Microsoft Teams.

Companies are increasingly scrutinizing the costs of running AI operations as 'tokenmaxxing' — the practice of maximizing token usage — drives up spending. The article examines how enterprises are grappling with mounting AI bills as they deploy AI more broadly across operations. For everyday businesses considering AI tools, this signals that costs can add up quickly, potentially affecting pricing of AI-powered products and services.

Singapore's first cybersecurity chief David Koh is stepping down on July 1 after 11 years leading the Cyber Security Agency of Singapore. Gwenda Fong, currently deputy secretary at the Ministry of Digital Development and Information, will take over. Under Koh's leadership, Singapore built its national cybersecurity framework from scratch—including the 2018 Cybersecurity Act, the 2016 and 2021 national strategies, and international diplomatic initiatives. He also managed major breaches like the 2018 SingHealth attack and the 2025 UNC3886 espionage campaign on telecom infrastructure. For everyday Singaporeans, this transition matters because CSA protects the critical services you rely on daily—power, banking, transport, and telecom—from hackers who could disrupt your life or steal your personal data.

Work platform monday.com has appointed Chris Jordan as Regional VP for Southeast Asia and Country Manager for Singapore, establishing the city-state as its second Asia-Pacific headquarters after Sydney. The company has built a team of over 30 people in Singapore since October 2025, recording 23 percent year-on-year growth across Southeast Asia and serving more than 4,700 regional customers. Jordan brings over 20 years of APAC experience, including nearly a decade at Salesforce. Nearly half of Southeast Asian companies have moved beyond AI pilots, with Singapore leading at 56 percent enterprise AI adoption.

Tokyo-based Sakana AI has raised $135 million in a Series B round led by Mitsubishi UFJ Financial Group, Khosla Ventures, and New Enterprise Associates. The company develops AI models tailored for Japan's needs, focusing on energy efficiency and sustainability, with partnerships in finance, defense, and manufacturing. Sakana AI was founded about two years ago and aims to address Japan's economic and demographic challenges through specialized AI technology.

Alibaba Cloud is opening its first data centers in Brazil, France, and the Netherlands, with plans to expand to Mexico, Japan, South Korea, Malaysia, and Dubai next year. The company already operates 91 availability zones across 29 regions. As part of its push, it launched an AI Catalyst Program offering up to US$120,000 in cloud credits to selected AI startups globally. This expansion means more cloud infrastructure options for businesses worldwide, though the move also highlights intensifying competition among global cloud providers.

Singapore workers are among the world's fastest AI adopters, with 66 percent of AI users producing work they couldn't have done a year ago, versus 58 percent globally. Singapore ranks second globally on AI diffusion, and 78 percent of AI users recognize the urgency of adapting quickly. However, only 24 percent say their leadership is aligned on AI strategy — creating a 'Transformation Paradox' where employees are moving faster than their organizations. Microsoft is urging companies to redesign workflows and decision-making processes to capture the value workers are already creating.

London and Singapore-based Galaxy Data Center has secured $250 million in strategic financing to build AI-ready data centers across Southeast Asia. The company plans to deploy high-density, GW-scale AI data center campuses using green energy solutions. Singapore serves as its regional hub for coordinating multi-country expansion. This investment reflects the massive appetite for AI computing infrastructure as companies race to build out capacity for generative AI and cloud services.

Interpol's latest Asia-Pacific cyberthreat report calls for stronger whole-of-society responses as criminals increasingly use AI to power scams. Over 6.5 billion cyber threats were detected in the region in 2024. Online scams and phishing remain the most common, but criminals now use AI-generated deepfakes, targeted spear-phishing, and automated messages at industrial scale. In Singapore, victims lost $913.1 million to scams in 2025, including a finance director who transferred $670,000 after a deepfake video call impersonated company executives. Interpol recommends better cloud security, continuous user education, and real-time threat intelligence sharing with industry partners.

Intel has committed an additional $2.6 billion to Vietnam, bringing its total investment there to $4.1 billion. The chipmaker originally walked away from a $3.3 billion Vietnam project in 2023 when the country couldn't offer the cash subsidies Intel wanted — that money went to Poland and Germany instead. Both those European projects have since been cancelled. Vietnam later created a new investment support fund to attract semiconductor manufacturers after adopting the global minimum tax, which neutralized old tax incentives. Intel now employs 6,500 people in Vietnam and generated $11.67 billion in exports in 2025.

Malaysia-founded respond.io has raised $62.5 million in Series B funding to expand its AI-powered customer messaging platform globally. The company helps businesses manage conversations across WhatsApp, TikTok, Instagram, and other channels using AI agents that work alongside human staff. With $35 million in annual recurring revenue and 169% year-on-year growth, the firm now serves over 10,000 businesses across 180 countries. The funding will fuel its push into North America and Europe while continuing to embed AI deeper into customer service workflows.

[ST Telemedia Global Data Centers (STT GDC)](https://www.sttelemediagdc.com/), a global data center provider, announced Tuesday the opening of STT Seoul 1, its first data center in South Korea. Developed and operated through a joint venture between STT GDC (60 percent) and Hyosung Heavy Industries (40 percent), the facility establishes a strategic foothold for STT GDC in one of Asia’s most advanced digital markets, extending STT GDC’s platform to support customers deploying artificial intelligence (AI) and cloud infrastructure across Northeast Asia, it said in a statement. According to the statement, South Korea is a strategically important market for high-performance digital infrastructure, underpinned by strong enterprise digitalization and growing demand for compute capacity. The openin

Analysts are examining whether Qualcomm can position itself to benefit from the rise of agentic AI—AI systems that independently complete complex tasks. The question is whether Qualcomm's mobile chip expertise translates to the new wave of AI computing demands, or if it risks being left behind as AI processing shifts.

Update on In US, EU mutual interest for Europe to use best AI models, von der Leyen says. Coverage via Yahoo Singapore Tech.

Southwest Airlines has always prided itself on being different. For 54 years, the carrier built its identity around open seating, low fares and a scrappy servant's heart. But staying competitive required a reckoning with the technology foundations underneath. "We really needed to follow our customer," said Lauren Woods, the airline's Executive Vice President and Chief Information Officer. And following the customer meant reinventing nearly everything. **From Crisis to Catalyst** Woods stepped into the CIO role in early 2023, weeks after a December 2022 winter storm exposed the fragility of Southwest's crew scheduling systems and triggered mass cancellations. The crisis clarified what needed to change. "Once you're out of the crisis mode, it becomes: how do we not get into this situation ev

Security giants Check Point and Illumio are teaming up to tackle a new breed of AI-powered cyber threats. The partnership combines Check Point's firewall defenses with Illumio's microsegmentation technology, which essentially creates walls inside corporate networks to stop hackers from moving freely if they break in. Research shows AI is now enabling attacks that used to take weeks to happen in minutes, leaving security teams almost no time to respond. The companies say this integrated approach gives enterprises a better shot at both blocking attacks and containing damage when breaches happen.

A stunning 82% of Singapore companies have pulled back or shut down live AI customer service bots after deploying them—eight points above the global average. This happens even though Singapore leads Asia-Pacific in AI adoption with a 72% deployment rate. The problem: AI fails, support teams get overwhelmed, and most companies lack proper governance guardrails to keep AI in check. Only 27% of Singapore firms report mature governance despite 75% prioritising trust and security. The takeaway: deploying AI is easy; making it actually work is another story.

# London mayor Sadiq Khan launches 'S'pore-style' housing approach to build 'affordable homes' - Mothership.SG - News from Singapore, Asia and around the world %20AppleWebKit%2F537.36%20(KHTML%2C%20like%20Gecko)%20Chrome%2F148.0.0.0%20Safari%2F537.36&date=1781709905481) [](https://mothership.sg/) * [News](https://mothership.sg/category/news) * [Lifestyle](https://mothership.sg/category/nova) * [Politics](https://mothership.sg/category/parliament) * [Earth](https://mothership.sg/category/environment) * [World](https://mothership.sg/category/abroad) * [Features](https://mothership.sg/category/features) * [Careers](https://mothership.sg/careers) [![Image 6: Search](https://mothership.sg/assets/imag

Update on G7 leaders vow closer ties on AI as they hash out 'trusted partners' scheme. Coverage via Yahoo Singapore Tech.

CrowdStrike has launched a new security tool called Continuous Identity for AI Agents that keeps AI workers in check as they access sensitive company data. Instead of giving AI agents permanent passwords that never expire (old-school security), the system constantly checks who's using each agent, what device they're on, and the risk level before letting them access anything — and cuts access the moment it's no longer needed. Built on technology from their SGNL acquisition, it assigns every AI agent a verifiable digital ID. This matters because companies are now deploying fleets of autonomous AI agents with system-level access to important data, and old security models weren't designed for that.

Jeff Bezos says AI will create more jobs for humans rather than replace workers entirely. The Amazon founder argues that automation historically shifts labor toward new tasks, and AI will do the same. It's a common refrain from tech leaders, though critics point out past predictions about technology creating jobs didn't always pan out equally.

Meta's head of product for AI workplace tools is leaving the company, according to a Reuters report. This executive led Meta's push to embed AI features into workplace products. The departure raises questions about the company's AI product roadmap during a competitive phase in the market.

Healthcare data has become one of the most traded commodities on underground cybercrime markets, with ransomware-linked data sales making up over a third of all activity. Patient records are especially valuable because, unlike credit cards, they cannot be cancelled or reissued — meaning stolen health data can be exploited for identity theft, fraud, and extortion for years. The findings highlight how hospitals and clinics, often running outdated IT systems and relying on third-party vendors, remain sitting targets as both cybercriminals and defenders increasingly deploy AI to scale their operations.

Senator Bernie Sanders has proposed giving Americans direct ownership stakes in AI companies, arguing that public investment in AI technology should benefit citizens directly rather than just shareholders. The plan would allow taxpayers to receive equity or dividends from AI ventures funded by government initiatives.

AI-powered shopping tools now find exactly what you want in seconds—but shoppers are increasingly hitting the brakes at checkout. A new study shows 70% of consumers use AI more for product search, yet 53% distrust AI-generated reviews and brand content. The problem: people don't doubt the AI recommendation, they doubt whether the retailer behind it is real. Even as AI gets better at predicting what you want, trust in online transactions hasn't kept pace. Major brands like Klarna see 700% more traffic from AI searches, but most consumers won't let an AI spend even $1 on their behalf without approval.

VivaTech, one of Europe's biggest tech conferences, wrapped its 10th edition in Paris with AI on everyone's minds. Around 200,000 visitors—including Amazon founder Jeff Bezos and French President Emmanuel Macron—heard visitors repeatedly name AI as both their biggest excitement and biggest worry. One French student captured the mood: 'It has double edges. It can have a lot of benefits for humanity, but at the same time it could take us to places that we don't want to go.' Highlights included humanoid robots, privacy-focused earpieces, and even a personal survival capsule for disaster scenarios.

Google has sued a group called Outsider Enterprise for allegedly running a massive AI-powered scam operation. The scheme used cheap $88-a-week software with over 290 ready-made phishing templates to impersonate banks, retailers, and government agencies. In just two weeks, they sent 2.5 million scam texts and operated roughly 1 million fake websites. The FBI estimates the operation stole at least 3.87 million credit cards and caused $1.9 billion in losses. Google says it's fighting back with AI tools that now block over 10 billion scam messages monthly, and is coordinating with the FBI and major US carriers to shut down the operation.

Data company Databricks used its annual summit to announce Genie One, an AI 'agent' designed to act as a coworker that learns your business and automates workflows. It also agreed to buy security firm Panther to help companies monitor threats across their AI systems. The company launched several other products focused on real-time data processing and customer management. Over 20,000 organizations—including Mastercard, AT&T, and adidas—already use Databricks, and 70% of Fortune 500 companies rely on its platform.

Jeff Bezos has spoken out against the common fear that AI will destroy jobs. He argues AI could actually cause a labor shortage by boosting productivity and creating more demand. It's a contrarian take from a major tech figure, though it's heavy on opinion without hard data.

The EU's cybersecurity agency ENISA is meeting with AI company Anthropic in San Francisco. This comes after a US export control rule forced Anthropic to cut off access to its most powerful AI models for people outside the US. The meeting was set up by Anthropic before the new restrictions, but the timing matters as governments worldwide grapple with how to control advanced AI while staying engaged with AI companies.

A coalition of major tech companies is boosting its funding for carbon removal projects by $915 million and bringing in Anthropic as a new member. It's a climate commitment by AI players, showing how companies are spending on sustainability even as they face regulatory pressure elsewhere.