Singapore AI News & Daily Briefing

Singapore is rolling out community-based AI training programs to help workers adapt to an AI-driven economy. The initiative, tied to the Smart Nation push, aims to equip everyday Singaporeans with practical AI skills rather than just technical expertise. This matters because nearly half of Singapore employees already fear job loss from AI, so accessible upskilling could be the difference between job security and redundancy.

Southeast Asian venture capital firms are split on how to approach AI investing. Manila-based Kickstart Ventures (backed by Ayala Corporation and BPI) has shifted its mandate toward AI since 2022, investing in San Francisco-based insurtech Lydia.AI and inference platform Featherless.AI. Singapore-based Analog Ventures has pivoted to AI-adjacent sectors after backing Blue Whale Energy, a virtual power plant for solar-enabled buildings in Singapore. Some VCs are moving capital to US AI hubs, raising concerns about SEA startups losing access to home-region funding.

Chinese tech giant Tencent is in talks to become the largest shareholder of Singapore-based AI agent startup Manus, according to sources. The move comes after Beijing ordered Meta to unwind its US$2 billion acquisition of Manus over concerns about Chinese technology being sold abroad. Tencent, along with Manus' original investors ZhenFund and HSG, plans to buy the company back from Meta for at least US$2 billion. Manus moved its operations from China to Singapore last year and gained fame as a general AI agent developer.

Apple has sued OpenAI for trade secret theft, accusing the ChatGPT maker of orchestrating a campaign to steal confidential information from Apple employees. Apple claims OpenAI poached over 400 former Apple workers and encouraged them to share details about upcoming products like smart glasses and AI devices. The lawsuit, filed in California, also names OpenAI's chief hardware officer Tang Tan, a former Apple executive who led iPhone design. The case highlights growing tensions between the two companies, which once partnered to bring ChatGPT to iPhones through Apple Intelligence but now find themselves competing directly in AI hardware.

Singapore's data centres face a 17% electricity tariff spike, but operators like CapitaLand Ascendas Reit and Keppel DC Reit are temporarily protected by fixed power contracts running until mid-2027. The catch: power makes up 50-60% of their operating costs, so when hedges expire, those costs get passed to tenants. Rather than fleeing Singapore, operators are likely to route new "spillover" compute capacity to cheaper locations like Johor and Batam, keeping Singapore as the premium low-latency hub.

Singapore workers in sectors deemed 'AI-resilient' — like healthcare, plumbing, security, and restaurant service — say AI helps with paperwork and administration but cannot replace human judgment, empathy, and physical tasks. Workers like healthcare assistant Radiana Ahmad use ChatGPT for translation while plumbers like Geraldine Goh use AI for quotes. DPM Gan Kim Yong previously noted that domestic-facing sectors like healthcare will remain important despite AI disruption. Experts say the productivity gap will grow between workers who use AI tools and those who do not.

Starbucks is building its own AI tools to replace software from Microsoft and IBM, aiming to cut $2 billion in costs under CEO Brian Niccol. The coffee giant proved a company can now build in-house what it used to buy from vendors — and the market noticed. IBM, ServiceNow, and Salesforce all dropped after the news, while Microsoft held steady since it also provides the cloud infrastructure Starbucks will build on. Notably, Starbucks had to pull an earlier AI inventory system due to accuracy problems, learning the hard way to fix the process first before automating.

OpenAI is scrapping Atlas, its AI-powered browser, just nine months after launch. The browser was plagued by slow performance, security flaws, and copyright headaches. It's now pivoting to ChatGPT Work, an office productivity suite. Atlas joins a growing list of failed OpenAI experiments, including its Sora video tool, raising questions about whether the company can actually ship useful products.

Meta pulled its Muse Image AI feature from Instagram just days after launch following widespread backlash. The tool, meant to generate AI photos for social media, automatically enrolled all users and drew criticism from SAG-AFTRA performers and privacy group Public Citizen over copyright and consent issues. Meta acknowledged the feature "missed the mark" and discontinued it, though Muse remains available on WhatsApp and the Meta AI app.

Swedish AI coding startup Lovable and Google Cloud have deepened their partnership, with Lovable set to increase its Google Cloud usage fivefold. The deal gives Lovable access to Anthropic's Claude and Google's Gemini models, plus a security tool from Wiz to catch code vulnerabilities in real time. Lovable, which recently raised US$330 million at a US$6.6 billion valuation, says users now create over 1 million new projects every week on its platform.

AI company Anthropic has opened a Seoul office, deepening its push into South Korea. Major Korean tech firms including Naver, Nexon, LG CNS, and Samsung SDS are adopting or expanding Anthropic's Claude tools for engineering and business operations. The move aligns with South Korea's AI Basic Act that took effect in January 2026 and the country's National AI Research Lab consortium backed by 94.6 billion won in government funding.

Sometimes, you have to dig beyond the headlines to get a better idea of how a company is performing. Endowus’ latest audited financial results are a good example of this. [Gregory Van](https://www.linkedin.com/in/gregory-van/?originalSubdomain=hk), the co-founder and CEO of the Singapore-based wealthtech firm, told _Tech in Asia_ last year that it had a [“clear path to profitability”](https://www.techinasia.com/endowus-doubles-revenue-2022-losses-rose-26m) by 2025. On its face, the results suggest that the firm has its work cut out for it to hit this milestone. Endowus co-founders Samuel Rhee (left) and Gregory Van / Photo credit: Endowus In 2023, Endowus’ total losses before tax widened to S$30.6 million (US$23.7 million) – a 14% jump from last year. This came even as its revenue grew by

_Welcome to Tech in Asia’s Sunday newsletter, your weekend brief on the biggest moves in Asia’s tech and startup world. Not on the mailing list? [Register here](https://www.techinasia.com/subscription?utm\_source=web&utm\_medium=article&utm\_campaign=dailynewsletter). Got a story tip? Send it to [email protected]._ ## In focus * Melissa Goh on the pressures VCs face to keep up * Behind Carousell’s revenue [bump](https://www.techinasia.com/secondhand-sales-surge-pushes-carousell-ebitda-profit-2025) * Singapore’s 10th-largest hirer in AI that [doesn’t have an office yet](https://www.techinasia.com/visual-story/openai-anthropic-hiring-singapore) * * * Hello reader, Founders get a lot of flak for drinking the AI Kool-Aid, but let’s face it, some investors are drinking it too. From AI agen

JD.com has started building its first RoboBase in Guangzhou's Huangpu district. The project aims to create a closed-loop robotics ecosystem integrating logistics automation. This signals growing competition among Chinese e-commerce giants to scale warehouse automation using robotics and AI.

Nasdaq-listed Empery has sold bitcoin holdings to acquire a stake in an AI data center, reflecting a broader trend of companies pivoting cryptocurrency assets toward AI infrastructure investments. The move signals growing corporate confidence in AI infrastructure as a strategic asset over digital currencies.

S&P Global has downgraded Oracle's credit rating to "BBB-," one notch above junk status. OpenAI accounts for roughly half of Oracle's $638 billion in contractual obligations. If OpenAI walked away, Oracle would be stuck with massive data center capacity it couldn't fill. The article S&P Global sees OpenAI as a "key credit risk" for Oracle and cuts its credit rating appeared first on The Decoder .

OpenAI CEO Sam Altman has done a complete U-turn, now saying he's "pretty sure" AI has created more jobs than it's destroyed. This contradicts his earlier warnings that entire professions would vanish. Anthropic CEO Dario Amodei is walking back similar doom-and-gloom predictions too. The truth is, nobody really knows yet—studies haven't backed either the scary or the optimistic predictions.

Scientists at Denmark's Technical University used AI combined with quantum computing to design new peptides—short protein fragments—for potential vaccines. Their hybrid approach, pairing a standard AI model with a quantum computer from UK startup ORCA Computing, produced more accurate results than classical computing alone, especially when training data was limited. The technique could eventually help develop drugs for understudied populations currently underrepresented in medical research. However, quantum computers remain too small for full-scale drug development, so practical applications are still years away.

Meta pulled a controversial feature from its new Muse Image model after widespread criticism. The feature let users generate AI images of other people by @-mentioning their public Instagram accounts. No consent needed, just a username. Meta admits "this feature missed the mark" and shut it down days after announcing it. The article Meta kills Muse Image feature that let anyone generate AI photos of Instagram users without consent appeared first on The Decoder .

A Brown University economics professor suspected most of his 86 students used AI to cheat on a take-home exam that averaged 96%. When he switched to an in-person test, 18 students dropped the course, 9 didn't show up, and the average fell to 48.6%. Studies from China and UC Berkeley support this pattern—students relying on AI for homework perform much worse on proctored exams. This raises serious questions about how universities should handle AI in assessments.

S&P Global has downgraded Oracle's credit rating to just above junk status. The reason: OpenAI accounts for nearly half of Oracle's $638 billion in data center contracts. If OpenAI walked away, Oracle would be stuck with massive unused capacity it can't easily fill. This signals that even tech giants face real financial risk when they bet heavily on AI partnerships.

Communities across the US are pushing back against AI data centers as they demand massive amounts of electricity, straining local power grids and raising environmental concerns. This grassroots resistance could reshape where and how AI companies build their infrastructure. Energy constraints are emerging as a key factor limiting AI expansion.

Anthropic has upgraded Claude Code with a built-in browser that lets its AI read, click, and type directly on websites without switching tools. Write actions like form submissions are filtered through safety classifiers, and sensitive tasks like purchases or account creation still require human approval. This marks another step toward AI that can autonomously navigate and complete tasks online for you.

Anthropic analyzed 1.2 million Claude Cowork sessions and found half of all usage goes to mundane office tasks nobody wants to do—things like compiling status reports, building onboarding checklists, or putting together slide decks. Developers barely use Cowork since they prefer Claude Code for coding work. This shows AI is increasingly being used to handle the boring "work around the work" rather than core professional tasks.

Researchers claim to have built a Go-playing engine that defeats established AI like KataGo without using neural networks, training data, or statistical learning. Instead, it relies purely on geometric calculations and topological analysis of the board. If verified, this challenges the assumption that massive training data and deep learning are always necessary for AI to master complex games like Go. The findings remain unverified by independent researchers.

AI is enabling entrepreneurs to launch businesses at record rates, though details are limited in this source.

Reed Jobs, son of Steve Jobs, runs Yosemite Ventures, an oncology-focused VC firm. In a new interview, he explains how AI is reshaping cancer drug discovery—helping scientists target proteins previously considered "undruggable" and potentially cutting clinical trial costs in half by using AI-generated control groups. One portfolio company doubled pancreatic cancer survival rates from 12 to 24 months. The firm now manages 25 companies across two funds.

Vancouver residents are fighting back against plans to build new AI data centres in their city. Hundreds have protested over concerns about water usage (the centres would consume over 33 million litres annually), electricity demands, and job losses from automation. A city councillor is now pushing for a temporary moratorium on new data centres until proper regulations are in place. The federal government, however, is moving forward with plans to invest heavily in AI infrastructure across Canada, saying these data centres are necessary for businesses to grow. This highlights a growing tension between AI expansion and community concerns about resources and livelihoods.

A new Straits Times feature explores how Singapore workers in "AI-resilient" sectors—healthcare, plumbing, security, and eldercare—are actually embracing AI tools to handle paperwork and admin tasks, while keeping the hands-on, people-facing parts of their jobs firmly human. Workers like Radiana Ahmad, a healthcare assistant, use ChatGPT for translation but say AI cannot replace the physical care and emotional connection they provide. The piece comes as Deputy PM Gan Kim Yong highlights that domestic-facing sectors will remain key employers even as AI reshapes the economy. Experts say while these jobs are harder to automate, workers who adapt to AI tools will have a growing productivity edge over those who do not.

An opinion piece argues that America's AI boom could face serious obstacles due to data center infrastructure challenges including power grid limitations, water consumption for cooling systems, and local opposition to large-scale facilities. The piece warns that environmental and regulatory hurdles may slow AI expansion significantly.

ByteDance is launching Seedance 2.5, the first AI video model capable of generating native 30-second clips in a single pass with impressive consistency in character, lighting, and motion. Early testing shows one-shot videos that hold together remarkably well, even maintaining small details across the full duration. The model accepts up to 50 reference inputs and supports region-level edits, targeting advertisers and creators who currently need to stitch shorter clips together.

A Pennsylvania state senator is facing scrutiny over his involvement in a $300 million data center deal. While he claims he had no role in securing the contract, his consulting firm still received payment. The deal highlights how data center contracts, particularly those tied to AI infrastructure, can attract political attention and potential conflicts of interest.

Singapore workers in sectors deemed 'AI-resilient' — like healthcare, plumbing, security, and restaurant service — say AI helps with paperwork and administration but cannot replace human judgment, empathy, and physical tasks. Workers like healthcare assistant Radiana Ahmad use ChatGPT for translation while plumbers like Geraldine Goh use AI for quotes. DPM Gan Kim Yong previously noted that domestic-facing sectors like healthcare will remain important despite AI disruption. Experts say the productivity gap will grow between workers who use AI tools and those who do not.

AI company Anthropic has opened a Seoul office, deepening its push into South Korea. Major Korean tech firms including Naver, Nexon, LG CNS, and Samsung SDS are adopting or expanding Anthropic's Claude tools for engineering and business operations. The move aligns with South Korea's AI Basic Act that took effect in January 2026 and the country's National AI Research Lab consortium backed by 94.6 billion won in government funding.

_Welcome to Tech in Asia’s Sunday newsletter, your weekend brief on the biggest moves in Asia’s tech and startup world. Not on the mailing list? [Register here](https://www.techinasia.com/subscription?utm\_source=web&utm\_medium=article&utm\_campaign=dailynewsletter). Got a story tip? Send it to [email protected]._ ## In focus * Melissa Goh on the pressures VCs face to keep up * Behind Carousell’s revenue [bump](https://www.techinasia.com/secondhand-sales-surge-pushes-carousell-ebitda-profit-2025) * Singapore’s 10th-largest hirer in AI that [doesn’t have an office yet](https://www.techinasia.com/visual-story/openai-anthropic-hiring-singapore) * * * Hello reader, Founders get a lot of flak for drinking the AI Kool-Aid, but let’s face it, some investors are drinking it too. From AI agen

JD.com has started building its first RoboBase in Guangzhou's Huangpu district. The project aims to create a closed-loop robotics ecosystem integrating logistics automation. This signals growing competition among Chinese e-commerce giants to scale warehouse automation using robotics and AI.

Nasdaq-listed Empery has sold bitcoin holdings to acquire a stake in an AI data center, reflecting a broader trend of companies pivoting cryptocurrency assets toward AI infrastructure investments. The move signals growing corporate confidence in AI infrastructure as a strategic asset over digital currencies.

Reed Jobs, son of Steve Jobs, runs Yosemite Ventures, an oncology-focused VC firm. In a new interview, he explains how AI is reshaping cancer drug discovery—helping scientists target proteins previously considered "undruggable" and potentially cutting clinical trial costs in half by using AI-generated control groups. One portfolio company doubled pancreatic cancer survival rates from 12 to 24 months. The firm now manages 25 companies across two funds.

Meta pulled its Muse Image AI feature from Instagram just days after launch following widespread backlash. The tool, meant to generate AI photos for social media, automatically enrolled all users and drew criticism from SAG-AFTRA performers and privacy group Public Citizen over copyright and consent issues. Meta acknowledged the feature "missed the mark" and discontinued it, though Muse remains available on WhatsApp and the Meta AI app.

Sometimes, you have to dig beyond the headlines to get a better idea of how a company is performing. Endowus’ latest audited financial results are a good example of this. [Gregory Van](https://www.linkedin.com/in/gregory-van/?originalSubdomain=hk), the co-founder and CEO of the Singapore-based wealthtech firm, told _Tech in Asia_ last year that it had a [“clear path to profitability”](https://www.techinasia.com/endowus-doubles-revenue-2022-losses-rose-26m) by 2025. On its face, the results suggest that the firm has its work cut out for it to hit this milestone. Endowus co-founders Samuel Rhee (left) and Gregory Van / Photo credit: Endowus In 2023, Endowus’ total losses before tax widened to S$30.6 million (US$23.7 million) – a 14% jump from last year. This came even as its revenue grew by

Vancouver residents are fighting back against plans to build new AI data centres in their city. Hundreds have protested over concerns about water usage (the centres would consume over 33 million litres annually), electricity demands, and job losses from automation. A city councillor is now pushing for a temporary moratorium on new data centres until proper regulations are in place. The federal government, however, is moving forward with plans to invest heavily in AI infrastructure across Canada, saying these data centres are necessary for businesses to grow. This highlights a growing tension between AI expansion and community concerns about resources and livelihoods.

Swedish AI coding startup Lovable and Google Cloud have deepened their partnership, with Lovable set to increase its Google Cloud usage fivefold. The deal gives Lovable access to Anthropic's Claude and Google's Gemini models, plus a security tool from Wiz to catch code vulnerabilities in real time. Lovable, which recently raised US$330 million at a US$6.6 billion valuation, says users now create over 1 million new projects every week on its platform.

A Brown University economics professor suspected most of his 86 students used AI to cheat on a take-home exam that averaged 96%. When he switched to an in-person test, 18 students dropped the course, 9 didn't show up, and the average fell to 48.6%. Studies from China and UC Berkeley support this pattern—students relying on AI for homework perform much worse on proctored exams. This raises serious questions about how universities should handle AI in assessments.

S&P Global has downgraded Oracle's credit rating to just above junk status. The reason: OpenAI accounts for nearly half of Oracle's $638 billion in data center contracts. If OpenAI walked away, Oracle would be stuck with massive unused capacity it can't easily fill. This signals that even tech giants face real financial risk when they bet heavily on AI partnerships.

Singapore's data centres face a 17% electricity tariff spike, but operators like CapitaLand Ascendas Reit and Keppel DC Reit are temporarily protected by fixed power contracts running until mid-2027. The catch: power makes up 50-60% of their operating costs, so when hedges expire, those costs get passed to tenants. Rather than fleeing Singapore, operators are likely to route new "spillover" compute capacity to cheaper locations like Johor and Batam, keeping Singapore as the premium low-latency hub.

Scientists at Denmark's Technical University used AI combined with quantum computing to design new peptides—short protein fragments—for potential vaccines. Their hybrid approach, pairing a standard AI model with a quantum computer from UK startup ORCA Computing, produced more accurate results than classical computing alone, especially when training data was limited. The technique could eventually help develop drugs for understudied populations currently underrepresented in medical research. However, quantum computers remain too small for full-scale drug development, so practical applications are still years away.

OpenAI CEO Sam Altman has done a complete U-turn, now saying he's "pretty sure" AI has created more jobs than it's destroyed. This contradicts his earlier warnings that entire professions would vanish. Anthropic CEO Dario Amodei is walking back similar doom-and-gloom predictions too. The truth is, nobody really knows yet—studies haven't backed either the scary or the optimistic predictions.

Anthropic analyzed 1.2 million Claude Cowork sessions and found half of all usage goes to mundane office tasks nobody wants to do—things like compiling status reports, building onboarding checklists, or putting together slide decks. Developers barely use Cowork since they prefer Claude Code for coding work. This shows AI is increasingly being used to handle the boring "work around the work" rather than core professional tasks.

S&P Global has downgraded Oracle's credit rating to "BBB-," one notch above junk status. OpenAI accounts for roughly half of Oracle's $638 billion in contractual obligations. If OpenAI walked away, Oracle would be stuck with massive data center capacity it couldn't fill. The article S&P Global sees OpenAI as a "key credit risk" for Oracle and cuts its credit rating appeared first on The Decoder .

Meta pulled a controversial feature from its new Muse Image model after widespread criticism. The feature let users generate AI images of other people by @-mentioning their public Instagram accounts. No consent needed, just a username. Meta admits "this feature missed the mark" and shut it down days after announcing it. The article Meta kills Muse Image feature that let anyone generate AI photos of Instagram users without consent appeared first on The Decoder .

Communities across the US are pushing back against AI data centers as they demand massive amounts of electricity, straining local power grids and raising environmental concerns. This grassroots resistance could reshape where and how AI companies build their infrastructure. Energy constraints are emerging as a key factor limiting AI expansion.

Starbucks is building its own AI tools to replace software from Microsoft and IBM, aiming to cut $2 billion in costs under CEO Brian Niccol. The coffee giant proved a company can now build in-house what it used to buy from vendors — and the market noticed. IBM, ServiceNow, and Salesforce all dropped after the news, while Microsoft held steady since it also provides the cloud infrastructure Starbucks will build on. Notably, Starbucks had to pull an earlier AI inventory system due to accuracy problems, learning the hard way to fix the process first before automating.

OpenAI is scrapping Atlas, its AI-powered browser, just nine months after launch. The browser was plagued by slow performance, security flaws, and copyright headaches. It's now pivoting to ChatGPT Work, an office productivity suite. Atlas joins a growing list of failed OpenAI experiments, including its Sora video tool, raising questions about whether the company can actually ship useful products.

Singapore's data protection watchdog (PDPC) wants companies to tell you explicitly when they use your personal data to train AI — no more hiding it in fine print. Banks using your transactions, social media platforms using your posts, voice recordings from apps — all would need AI-specific notifications with clear opt-out options. This is Singapore's push to keep privacy rules relevant in the AI age.

OpenAI CEO Sam Altman and SpaceX CEO Elon Musk are back at each other again on social media. Altman joked that his company's latest model (5.6) must be the best because Musk is 'obsessed' with him again. The renewed spat comes after Apple sued OpenAI for trade secrets theft. The two tech leaders co-founded OpenAI in 2015 but have been feuding for years over the company's direction, with Musk now running rival AI firm xAI.