One of the biggest changes AI has brought to my work is that I can do more. I am no longer as daunted by the mundane work involved in getting something started.

A research idea, teaching session or presentation comes with preparation: organising notes, building a structure, condensing information and revising the wording. These tasks are necessary, but the effort can become a barrier.

AI makes that preparation feel more manageable.

Across clinical practice, research, teaching and healthcare innovation, I use it to synthesise evidence, challenge ideas, review documents and prepare for discussions.

The benefit goes beyond completing an existing task faster. It changes how willing I am to begin. But being able to produce more makes deciding what deserves my attention more important.

Start with a specific task

Recently, I asked AI to tighten a healthcare innovation presentation to ten minutes. My instruction was simple: less is more.

Producing material was easy. Deciding what deserved the audience’s attention was harder.

The task involved taking material developed for an innovation discussion and reducing it to fit the time available. AI could help reorganise the argument, simplify explanations and suggest shorter headings.

However, shortening a deck requires more than removing words. Which points support the discussion? Which can be dropped? What should the audience understand or decide afterwards?

Those questions provide the standard against which to assess the output.

This is why defining the task matters. “Improve this presentation” leaves considerable room for interpretation. A specific audience, purpose and time limit give the work clearer boundaries.

For founders and operators, the equivalent might be an investor update or management paper. Before generating the document, identify the decision it needs to support.

A small error in this article

Preparing this article provided a concrete example of why checking matters.

My original instructions about the presentation were “Tighten for 10 mins presentation” and “Less is more”.

AI rewrote them as “Tighten for a 10-minute presentation. Less is more.” It improved the grammar but retained quotation marks, presenting edited wording as a direct quotation.

When I subsequently asked for verification and a word count, the review identified the discrepancy. The revised opening paraphrases the request.

The meaning had barely changed. Yet quotation marks make a specific claim: these were the words used.

Also Read: The sovereign threshold: Why airtight AI decisions are paralysing executive teams

When AI transforms source material, it can preserve the general meaning while changing a detail that matters. In research, a more consequential example would be turning “associated with” into “causes”: a small wording change with a substantial change in meaning.

Asking AI to check its work helped surface the quotation issue. However, another AI response is not independent evidence. The original wording made the discrepancy verifiable.

Use summaries to begin the review

Another recent task involved asking AI about the study design and appropriate critical appraisal tool for a national linked-data study of out-of-pocket healthcare costs in Australia.

The question was more specific than “summarise this paper”. I wanted to understand how to assess the evidence.

AI can explain terminology and compare appraisal approaches. Selecting an appropriate tool, however, depends on what the researchers actually did.

Was an intervention evaluated? How were participants selected? How were variables measured? Which biases could affect the findings?

The answers must come from the methods.

The practical lesson is to use the summary as an entry point, then return to the source for claims that determine the appraisal.

The same applies to market reports and customer research. Understanding the conclusion is insufficient if the method cannot support it.

Challenge polished but empty answers

Some output needs correction because it says too little.

I have asked AI to make writing less like a “motherhood statement”: something agreeable that offers limited practical direction.

Consider: “Healthcare organisations should embrace innovation to improve care.”

What should the reader do after reading that sentence? Which problem deserves attention? What would change? How would improvement be assessed?

The sentence is not necessarily false. It is insufficiently specific to guide action.

AI can make this harder to notice because fluent language gives an argument the appearance of completion.

A useful test is whether the paragraph still works after replacing the organisation’s name with another. If it does, it may need more context.

For a founder, “we use AI to improve efficiency” raises similar questions. Which task changes? Who benefits? What happens when the output is wrong?

Also Read: The AI told me to raise my prices by 40 percent but it had never been to a wet market

More capacity still requires judgement

I have not stopped reviewing or editing because AI is available. What has changed is my willingness to tackle work that includes substantial routine preparation.

I can ask it to organise material, propose an outline or shorten a draft, then examine and improve what it produces. The initial work feels less daunting.

I think of this as a cognitive multiplier. It makes it easier to develop ideas and explore alternatives.

However, generating a draft is only part of the workflow. Reviewing, correcting and adapting it also count.

A field experiment involving 758 Boston Consulting Group consultants found that participants using GPT-4 completed tasks within its capabilities 25.1 per cent more quickly on average. On a task outside those capabilities, AI users were 19 percentage points less likely to produce the correct solution.

This was a consulting experiment, not a clinical evaluation. It does not establish how much time AI saves in my work.

It illustrates why benefits should be assessed task by task. A quick draft that needs extensive repair may offer less value than its initial speed suggests.

The relevant questions are whether the finished work is useful and whether it deserved to be done.

Keep ownership of the decision

AI can help develop a coherent research proposal. Whether the study should proceed depends on its importance, feasibility, participant burden and potential to change practice.

Operational decisions require similar judgement.

Consider an appointment no-show prediction system. Identifying someone as likely to miss an appointment does not determine the appropriate response.

A reminder, telephone call and help with rescheduling involve different resources and assumptions. Missed appointments might reflect transport difficulties, caregiving responsibilities or an unsuitable appointment time.

Someone must decide which response addresses the problem, who will deliver it and how its effects will be assessed.

Also Read: The 70/30 rule: Why ASEAN banks still need humans as AI takes over

Clinical decisions involve additional considerations, including the patient’s circumstances, preferences and the consequences of acting or waiting.

AI can organise those considerations. The decision requires someone who understands the context and accepts responsibility for the action.

For founders, this matters whenever AI informs hiring, customer support or resource allocation. A recommendation needs an accountable owner.

Practice makes permanent

I believe professionals should use AI appropriately and often. Familiarity develops through working with it, examining the results and adjusting the approach.

But practice does not automatically make perfect. It can make our habits permanent—including the wrong ones.

Repeatedly accepting plausible answers can establish a habit of uncritical reliance. Repeatedly checking important claims, comparing outputs with sources and correcting errors can reinforce a more disciplined approach.

Frequency alone is insufficient. The quality of the feedback matters.

Start with a bounded task where the output is possible to assess. Define what success looks like. Identify which details require verification. Review where the tool helped and where it introduced additional work.

In healthcare, appropriate use also includes respecting confidentiality and approved arrangements for handling information.

The aim is to become better at judging the output, including recognising when additional expertise is needed.

AI has made me less reluctant to tackle the mundane work. It helps me reach a workable draft, a sharper question or a clearer structure more quickly.

Before using the result, I still need to answer three questions: Is it supported? Does it fit this situation? Can I take responsibility for acting on it?

AI helps me get to those questions faster. I still have to answer them.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post AI can help me think faster, but I still have to decide how I actually use AI in healthcare appeared first on e27.

Yesterday did not look like a market finding its footing. It looked like a sequence of moves designed to catch as many people offside as possible, and the evidence lies in the timing and the numbers.

South Korea’s KOSPI crashed 9 per cent to 4,923.48, and a circuit breaker halted trading. That came one day after the same index fell 5.28 per cent to close at 5,383.80 on Nov 5, when sell-side sidecars froze program trading for five minutes, and foreign investors dumped Samsung Electronics and SK Hynix by 7 per cent to 9 per cent each.

By the time the Asian session opened on Nov 6, the damage had spread beyond Seoul. Bitcoin was down 5.78 per cent to US$76,011.79 and underperformed a broader selloff that had begun on Wall Street the previous day. The 98 per cent correlation with the S&P 500 suggests this wasn’t a crypto story. It was a macro risk-off move, and digital assets offered the highest-beta expression of that fear.

The narrative most desks offer is straightforward enough. On Nov 5 the Nasdaq fell over 5.5 per cent as investors reacted to Democratic control of Congress and the fear that a Blue Wave would bring restrictions on AI data-centre spending. Nvidia dropped 9.3 per cent, and AMD plunged 17.3 per cent, and the rout spilled directly into digital assets. The crypto complex fell 5.29 per cent to US$2.55 trillion.

Derivatives data showed total open interest rose 9.79 per cent to US$382.82 billion, and volume surged 12.72 per cent, suggesting speculative activity had built up ahead of the move. That leverage became fuel. Over US$1.08 billion in crypto positions liquidated in 24 hours, with long positions accounting for US$954 million. Bitcoin alone saw US$281.62 million go into liquidation, with longs making up 90 per cent of that total. The forced closure of those positions added selling pressure and accelerated the drop.

Also Read: Can Bitcoin hold US$80,000 after a US$386M short squeeze and US$1.16B in ETF outflows?

The sequence does not look accidental. The initial move did not simply reward shorts. It also invited late longs into a position that became crowded and fragile. Once the macro trigger arrived, the tape reversed with unusual violence. Big players do not need a conspiracy to achieve this.

They need a crowded book, a visible catalyst, and the willingness to press the advantage when the other side is trapped. The KOSPI sidecar on Nov 5 and the circuit breaker on Nov 6 provided exactly that visible catalyst. Once Seoul halted trading, the risk-off signal became impossible to ignore, and selling pressure fed on itself across Asia and into digital assets.

The mechanics explain why the drop felt sharper than the equity move. Crypto open interest sat at elevated levels heading into the election, and the post-election bounce encouraged more leverage on the long side. When the Nasdaq rout hit, forced selling in derivatives did not merely reflect the equity decline. It amplified it.

Each liquidation pushed the price lower, which triggered more liquidations, which pushed the price lower again. The US$1.08 billion figure is the scoreboard for that feedback loop. The 98 per cent correlation with the S&P 500 is the broader context, but the liquidation cascade is the mechanism that turned a macro headwind into a violent repricing.

Bitcoin’s 7-day RSI fell to 36.43, which is oversold territory. Oversold conditions in a leveraged book can persist longer than anyone expects when the marginal seller is a forced seller rather than a discretionary one. Funding rates at +0.0065 per cent also show that positioning had not yet reset cleanly.

Also Read: US$1B ETF outflows, Democratic sweep highly likely, Iran war: Can Bitcoin hold US$75,600?

The near-term picture now hinges on a narrow set of levels and one external variable. Bitcoin is testing the US$75,000 to US$75,600 support zone, and that zone matters because buyers previously stepped in there. If support holds, consolidation is the most likely outcome, and the oversold RSI could support a minor relief rally toward the US$77,000 pivot.

A decisive daily close below US$75,000 would open the door toward US$73,000 and would confirm that the deleveraging event has not fully run its course. The total crypto market cap sits at US$2.55 trillion, and the next major support below the current swing low is the Fibonacci 78.6 per cent retracement at US$2.61 trillion, followed by the 200-day simple moving average at US$2.48 trillion. That last level separates a painful correction from a structural breakdown.

The pump and dump framing is not the only way to read yesterday, but it is a defensible one. The tape did not simply react to the Nasdaq. It first offered a false sense of direction, then punished the side that believed it. The KOSPI circuit breaker gave the move its alibi, and the liquidation engine gave it its speed.

What matters now is whether the US$75,000 floor holds over the next 24 hours while the equity market decides whether the AI trade is merely correcting or genuinely unwinding. If the tech rout stabilises, digital assets can consolidate and rebuild. If the KOSPI panic spreads further into U.S. futures, forced selling will find another leg lower. The big players who wanted quick wins have already booked them. The question is whether the tape can find a bid without them.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post Bitcoin pump and dump? The US$75,000 support zone will decide appeared first on e27.

Indonesia’s anti-corruption framework, codified in UU 31/1999 jo UU 20/2001, allows prosecutors to charge state-owned enterprise officials with corruption when their decisions are deemed to have caused state losses (kerugian negara), even in the absence of personal benefit, kickback, conflict of interest, or bad faith. The policy debate around this framework has been active for two decades. The question of how Indonesian law should treat investment risk inside state-owned enterprises has not been settled.

After 15 years inside Indonesian risk functions, I have watched this debate from the operational side. The gap between business judgment and criminal exposure is not theoretical. It is something senior managers in BUMN institutions operate inside every quarter.

The legal framework that creates the gap

Indonesian anti-corruption law contains three elements that, together, produce the structural exposure.

The kerugian negara element. The statute allows prosecution when an act has caused harm to state finances. Constitutional Court ruling MK 25/PUU-XIV/2016 established that the loss must be actual and concrete, not merely potential. The threshold for what counts as actual loss has been the subject of substantial litigation since.

The mens rea question. The law does not always require proof of personal benefit or bad faith. An official can be prosecuted for a decision deemed contrary to law (melawan hukum) that produces state losses, even where the decision was made through formal institutional processes.

The absence of a business judgment doctrine. Common-law jurisdictions have evolved business judgment rules that protect corporate officers from second-guessing of decisions made in good faith with reasonable information. Indonesian commercial law does not have a directly equivalent doctrine for BUMN officials. Investment decisions that look like prudent risk-taking under one framework can look like criminal exposure under another.

The risk management implications

Three operational consequences follow.

Documentation as primary defence. The most important practice for BUMN investment officials becomes the quality of decision documentation. Every alternative considered, every external review, every committee approval, every dissenting view recorded: these become the evidentiary record years later. The cost of weak documentation is paid in a courtroom.

Also Read: Indonesia urgently needs an allergy data centre to support national food security

Conservative bias in investment selection. The framework produces structural preference for conservative, well-documented investments over high-conviction contrarian bets. This is rational individually. It is a constraint on returns institutionally.

Selection effect on senior talent. The most experienced investment professionals have alternatives. Several choose independent funds or private sector roles partly because personal exposure to criminal proceedings in BUMN roles is meaningfully larger than in equivalent private positions.

What healthy practice looks like

Three practices distinguish BUMN investment functions that operate well inside the existing framework.

Pre-decision risk review documentation. The strongest committees document the decision, its justification, alternatives considered, risks identified, mitigation plans, and dissenting views, at the time the decision is made, not after the fact. The contemporaneous record is the asset that protects officials and institutions if decisions are later questioned.

External validation embedded in process. Investments that include genuine external review (independent advisors, third-party valuations, regulatory consultations) produce a decision record that is harder to second-guess. The cost of external validation is meaningfully less than having a decision questioned without it.

Clear separation of business judgment from procedural compliance. Governance processes that distinguish business judgment (was this a reasonable investment?) from procedural compliance (was this approved correctly?) protect against later confusion. Procedural compliance is binary. Business judgment is probabilistic. The two should not be conflated.

Lessons learned

Five principles emerge from observing BUMN investment governance inside the existing framework.

Documentation is the primary risk management tool. Not models, not committees, not advisors: documentation. The decision record is what is examined years later. Institutions that treat documentation as first-order discipline absorb less exposure.

Process is necessary but not sufficient. Following formal approval process is the baseline. It does not, by itself, protect against later criminal exposure if the underlying decision is later deemed contrary to law. Substantive quality matters as much as procedural correctness.

Conflict of interest disclosure is non-negotiable. The cleanest defensible decisions are those where every potential conflict, direct or indirect, present or historical, has been documented and recused. Disclosure cost is small. Non-disclosure cost can be career-ending.

Also Read: The sharia trust premium: Lessons from Indonesia’s first sharia P2P generation

External counsel involvement is insurance, not cost. Major BUMN investments that weathered legal scrutiny best were those where external counsel was involved in the decision pathway, not retained after the fact. Investment in external counsel during the decision process is risk insurance.

The framework itself is a policy question, not a personal one. The structural tension between Indonesian anti-corruption law and BUMN investment risk-taking is a legitimate policy debate. Individual officials, however, have to operate within the law as it currently is, not as they might prefer it to be.

The macro stakes

Indonesian BUMN investment activity matters for the broader economy. State-owned enterprises remain among the largest capital allocators in the country, their investment decisions shape entire sectors, and the talent willing to lead those decisions determines the quality of Indonesia’s strategic capital deployment.

The legal framework around BUMN investment risk will continue to evolve. The risk management practices inside BUMN institutions, however, can evolve faster. The institutions that build documentation discipline, external validation processes, and decision pathways that work inside the current framework will produce both better investment outcomes and lower exposure for the officials making decisions.

The business judgment gap will not close quickly. The institutions that build for it deliberately will operate inside it better than those that hope it will close.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post The business judgment gap: Why BUMN investment decisions need a different risk framework in Indonesia appeared first on e27.

Southeast Asia’s tech startups drew about US$16.55 billion between January and October 2026, according to Tracxn. October alone delivered US$3.68 billion across 14 rounds, 933.7% above October 2025.

The headline flatters the market. January, June, July and October account for 81% of the year’s money, while the other six months averaged about US$520 million each. Eleven rounds across ten companies, led by DayOne, Kling AI, Ant International and Manus, make up roughly two-thirds of the total. Consumer internet, the engine of the last boom, barely features. Several of those giants are Southeast Asian mainly by address: Kling AI, Ant International and Manus, all Chinese in origin and now headquartered in Singapore, contributed about US$4.5 billion between them.

Deal volume tells the quieter story. Tracxn logged between 11 and 23 rounds a month, averaging just under 15, and most were seed or early stage. Series B and C rounds remain scarce, while state-linked investors such as EDBI, SEEDS Capital and ADB Ventures keep surfacing on the most-active lists.

Tracxn’s revisions add another caveat: September’s US$81 million has since become US$209 million. For founders, the number that matters is the flat 15 rounds a month, not the billions.

---

REGIONAL

Google commits US$15B to Malaysia, pledges to pay its own power bill: The 2027–28 package, unveiled with Prime Minister Anwar Ibrahim present, bundles new data centre campuses with skilling, water and clean energy pledges. It dwarfs the US$2 billion Google committed in 2024 for its first Malaysian data centre.

Microsoft picks Johor Bahru for its second Malaysian cloud region: Southeast Asia 3 forms part of Microsoft’s US$2.2B four-year Malaysia plan and follows the Malaysia West region opened in May 2025. Its backup generators will run on renewable biofuel, and a programme will train 150 graduates for data centre jobs.

Grab posts record US$237M EBITDA, but lending now drives its growth: Net profit reached only US$18 million. Financial services revenue jumped 124% to US$201 million as the gross loan portfolio tripled to US$2.7 billion, helped by the Superbank and Stash acquisitions, with the Atome deal still to come.

Malaysia carried SEA’s IPO market in 2026 while the rest went quiet: EY data shows Southeast Asia logged 56 IPOs raising US$3B in the first nine months of 2026, down from 76 a year earlier. Strip out Malaysia, and the region’s public markets look like a waiting room.

Singapore hands IPO prospectus reviews to SGX RegCo from January 2027: Issuers will file prospectuses and listing applications through one SGX RegCo electronic portal, replacing separate MAS reviews. SGX RegCo targets six to eight weeks per complete application and will stand down its Listings Advisory Committee.

Visa opens six global capability hubs in Singapore with 1,500 staff: Based at a new Paya Lebar Green Solution Center, the hubs will work on agentic commerce, stablecoins and quantum security with partners including DBS, Nium and StraitsX, and join MAS’s BLOOM cross-border digital money initiative.

Pine Labs and Stripe unify online and in-store payments in Singapore: From November, eligible retailers get one view of all transactions across Stripe’s online rails and Pine Labs’ terminals, plus instalments, pay-by-points and promotions. The partners plan to expand to other Southeast Asian markets in 2027.

Malaysia launches Startup ASEAN digital centre at Cradle LIVE! 2026: The Digital Center of Excellence adds seven modules, including AI matching, to the Startup ASEAN platform, which already counts more than 52,000 active users and 10,400 startups from over 150 countries.

INTERVIEWS & FEATURES

Myanmar refugees are building Thailand’s AI data centres: A Nikkei Asia photograph of refugees working on a data centre site in Chonburi exposes the labour shortage beneath Thailand’s AI ambitions, and asks who actually benefits from the region’s infrastructure boom.

Taiwan’s envoy sees Singapore as first stop for its global startups: Dr Tung Chen-yuan, Taipei’s representative in Singapore, visited 33 Taiwanese startups at SWITCH 2026 and argues they are capable enough to go global, with the city-state as the natural launchpad.

Afore backed an SUTD founder, then let him scrap his plan in a week: Mohammed Khambhati learnt in his dorm that the San Francisco pre-seed firm was funding him. After flying to SF, he abandoned the original plan within days, and his investors did not flinch.

How DVI Group merged eight systems after its pandemic boom: The Singapore-based audio-visual integrator found its sales, finance, delivery and service teams tracking live contracts with different numbers. General Manager Tanish Dhillon explains how consolidating its systems gave management one figure to trust.

Multiable HCM pitches AI HRIS to Singapore’s listed companies: The piece argues SGX-listed groups face intense scrutiny over workforce governance and disclosure accuracy, and positions Multiable’s AI-driven HR platform for enterprises managing thousands of employees across business units.

INTERNATIONAL

OpenAI could be IPO-ready in 2027, says Greg Brockman: Speaking at a WSJ conference, the OpenAI president said enterprise sales now drive most of the company’s revenue, as it pushes its Codex agent to businesses and steps up fundraising ahead of a listing.

China blocks SoftBank’s US$5.4B buy of ABB’s robotics arm: ABB and SoftBank remain in talks after an NDRC-led security review halted the transaction. The review can reach offshore deals, and China generates about 30% of ABB’s robotics revenue. ABB had earlier planned an IPO spin-off.

Sequoia shuts London office, scraps Arc accelerator and Scouts: The firm will merge its early-stage and growth teams, flatten its partnership, now run by seven general partners, and drop Monday partner meetings. It has raised about US$10B in new funds and will back European founders from the Bay Area.

TikTok quietly launches a standalone TikTok Shop app: The app offers personalised product recommendations alongside livestream and short-video shopping, splitting commerce from the main feed. It is not yet live in the US; Southeast Asia remains one of TikTok Shop’s core markets.

Hong Kong startup funding drops 77% to US$94.1M in October: Tracxn logged just five rounds in October, with Fenorri and PhotonPay taking about 86% of the capital. September’s US$300 million Qupital raise inflated the comparison, and the city recorded no IPOs or acquisitions.

Meta, Google and TikTok sue California over teen social media law: AB 1709 would bar autoplay and personalised feeds for under-16s from January, with fines of up to US$50,000 per affected user. The platforms cite the First Amendment; privacy advocates warn of intrusive age checks.

OpenAI is already trying to shape the 2028 US election: Chief global affairs officer Chris Lehane’s memo calls AI the number-one issue of 2028 and urges candidates to offer a ‘third path’, warning that opposing data centre construction could cost them the debate.

CYBERSECURITY

How a fake job interview opened Binance to North Korean hackers: Chainalysis has tied more than US$14.3M in losses over 18 months to DPRK-linked fake job interviews, the playbook behind the US$6.4 billion Binance theft. Startups hiring remote engineers sit squarely in the target zone.

Binance confirms fake recruiters let North Korea in for US$6.3B heist: Hackers posing as recruiters targeted a Binance wallet software engineer and planted malware on their laptop. Researchers have warned for years that DPRK operatives pose as recruiters, investors and VCs to breach crypto firms.

Every AI budget needs a matching security budget: Writing after the biggest theft in crypto history, JitSiong Thaddeus Koh argues the rush to deploy AI agents and assistants is widening attack surfaces, and companies should match AI spending with equal security investment.

SEMICONDUCTOR

Ondine Capital closes Fund III’s first round with Taiwan chipmakers: The Taiwan-based VC did not disclose the size but will back foundation models, next-generation semiconductors and physical AI. Early bets include silicon-photonics firm Salience Labs, and Brian Yen joins as partner from Headline Asia.

Singapore’s factories ride the AI wave, but on a narrow base: A net weighted balance of 24% of manufacturers expect better conditions between October 2026 and March 2027, the EDB said, but the optimism leans heavily on demand from AI data centres.

AI

Frontier AI money is getting picky. SEA founders should price it in: OpenAI cut its fundraising target by US$10 billion to US$20 billion as tech stocks fell, ending an era in which the biggest ask won. Regional AI startups should expect sharper scrutiny of their numbers.

Singapore tops Commonwealth AI index; Australia shows the real gap: The Ayre Index, launched at CHOGM, ranks young Singaporeans the best prepared for AI. Third-placed Australia offers the sharper lesson for Southeast Asia: having the technology is the easy part.

Nine in ten Singaporeans use AI monthly; only a third feel comfortable: We Are Social and Meltwater’s Digital 2027: Singapore report finds a wide gap between habit and confidence, a warning for AI builders who assume adoption equals trust.

AI chatbots cave on climate facts, and SEA languages may fare worst: A UNESCO and University College London study found chatbots that answer correctly on human-caused warming often start agreeing with users after a turn or two of pushback, a weakness likely worse in Southeast Asian languages.

Anthropic’s compute bill climbs to 58% of revenue ahead of its IPO: Anthropic’s S-1 filing shows compute margins worsening when they should improve, with costs up from 48% of revenue the previous quarter. The company is shifting more workloads to its own infrastructure to cut the bill.

Meta warns election backlash against data centres could slow AI: President Dina Powell McCormick said Democratic midterm wins will make data centres harder to build. Meta now pays for its own electricity and keeps jobs local to win approval, a model Google echoed in Malaysia.

Airbnb to launch its own AI travel agent in the first half of 2027: Rather than a text chatbot, the agent will compose maps, photos and itineraries in real time and handle rebookings and refunds. Airbnb announced it alongside Q3 revenue growth of 19% to US$4.87 billion.

THOUGHT LEADERSHIP

Wall Street’s tokenisation push holds lessons for SEA founders: After five weeks in New York, including Avalanche Summit and UN Blockchain Week, Astrid Dang argues founders should start with the problem first, not the token, as blockchain conversations move beyond crypto trading.

Information asymmetry didn’t die in 2026; it just learned better PR: Faheem Aizat Kamsan contrasts a fintech report declaring information asymmetry dead with Singapore’s largest bank, which is still teaching its AI assistant card refunds, arguing agentic AI hype outpaces banking reality.

The step zero problem holding back SEA’s startup ecosystem: Benedict Kok notes the region’s startup ecosystem was valued at US$454 billion in the first half of 2024, yet argues its increasingly sophisticated support infrastructure still leaves gaps at the very first step.

Global expansion now demands a different mindset: Helmi Talib argues the old cross-border expansion playbook, in which more borders meant more success, no longer works, and companies must now choose where and how they expand far more selectively.

Could AI agents break fractional-reserve banking?: John Millar asks what happens to banks when AI agents can move deposits at short notice all at once, undermining the assumption that customers never ask for their money back simultaneously.

ETF outflows and macro risk test Bitcoin’s US$75,600 support: Bitcoin fell 1.05% to US$75,830.54 in 24 hours. Anndy Lian argues retreating institutional demand, an 85% correlation with equities and geopolitical risk are reinforcing each other.

Startups should plan for scaling, not mere survival: Jon Low argues short-sighted choices on partners, risk and hiring create disruption and burned money, and that founders who get these early decisions right are the ones who scale quickly.

AI might be the final boss for your best ideas: Januar Parlindungan fed his company’s 2026 plan, spanning payroll, billing, cross-border payments and licences, into AI to decide which initiatives mattered most, and reflects on what the exercise revealed.

Using AI for sales outreach without outsourcing judgement: Mustafa Rasheed uses AI not to blast generic emails but to find businesses showing public signals that match his offer, keeping the decision on whom to contact firmly human.

AI speeds up healthcare thinking, but humans still decide its use: Yee Ang says AI has removed the dread of starting research, teaching and presentations, but argues practitioners must still choose deliberately where the tools belong in their work.

I automated my life with AI agents, then a 2 am voice note hit: Vincent Tan spent months building an autonomous productivity engine to remove every bottleneck, until a founder in crisis showed him that execution is cheap, presence priceless.

Why shoppers are tuning out brand messages, and how to stop it: Petros Romas warns that customer disengagement from brand messaging does more damage than algorithm shifts, outages or privacy rules, and often goes unnoticed on dashboards until retailers have lost their customer base.

The post Ecosystem Roundup: SEA startups raised US$16.5B in 2026, but four months did most of it appeared first on e27.

Malaysia’s second 5G operator wants startups to do something telcos across the region have struggled to make happen: build for its network before anyone has agreed to pay for the result.

U Mobile said on Friday that its Enterprise Innovation Platform (EIP) is deepening its collaboration with Cradle Fund, the government-linked startup agency, to help Malaysian startups develop and validate 5G-Advanced (5G-A) and AI solutions for real business use.

The split of duties is simple. Cradle brings the startup pipeline, its programmes and possible funding routes. U Mobile brings a 5G-A and AI sandbox, its engineers and introductions to enterprise buyers.

Also Read: Malaysia’s digital economy’s second wave looks nothing like the first

For the uninitiated, 5G-A is the next layer of 5G standards, promising lower latency, faster uplink and more precise device positioning. That matters far more to a factory robot or a fleet of connected trucks than to someone streaming football on their phone.

The fine print deserves a second read. U Mobile said access to specific sandbox capabilities depends on engagement and technical requirements, and that funding and commercial opportunities are not guaranteed.

In plain terms, this is a test bench and a networking room, not an investment vehicle.

A fast network still hunting for paying use cases

U Mobile has a point to prove. It was named the winner of Malaysia’s second 5G network bid on 1 November 2024 and received its letter of award on 31 March 2025, setting it up as the challenger to state-backed Digital Nasional Berhad. By the end of September 2026, it said its network covered 90 per cent of populated areas, up from 80 per cent in March. It also cites recognition from Ookla as Malaysia’s fastest 5G network, a joint win with Opensignal for reliability, and a Malaysia Book of Records entry for the fastest 5G rollout.

Coverage, however, is the easy part to announce. Monetisation is harder. Operators across Asia have spent heavily on 5G only to find that consumers will not pay much more for faster phones. The real money, if it arrives, sits in enterprise: private networks, edge computing and industrial IoT. Those markets need applications, and applications need developers who understand the network. That is where startups come in.

“Malaysia has no shortage of bold ideas and talented innovators. Startups need the right environment to test their solutions and understand how they can address real business needs,” said Woon Ooi Yuen, Chief Technology Officer of U Mobile.

He added that the Cradle partnership would connect more innovators with EIP’s 5G-A and AI capabilities, technical expertise and industry engagement, helping them “demonstrate their potential to businesses”.

Cradle wants startups with passports, not just pitch decks

For Cradle, the deal is another corporate bridge. Norman Matthieu Vanhaecke, its Group CEO, said the agency’s focus is on building an ecosystem that lets Malaysian companies grow beyond domestic markets. “Ultimately, we want to see more Malaysian startups develop the scale and competitiveness to succeed regionally and globally,” he said.

Partnerships have become Cradle’s default mode. It previously teamed up with Invest India to link the two countries’ startup ecosystems, and this tie-up has been running since September 2026 through the Cradle Roadshow in Sarawak, the MYStartup Pre-Accelerator Demo Day hosted at the EIP Hub on 28 October, and Cradle LIVE! Summit 2026.

Also Read: Google’s US$15B Malaysia bet comes with a promise to pay its own power bill

The Sarawak leg is a quietly useful detail. Much of Malaysia’s startup support still orbits the Klang Valley, and East Malaysian founders working on agriculture, energy or logistics are precisely the ones who could use a connected-industry testbed.

Policy tailwinds, with a funding caveat

The timing suits both sides. Budget 2027 hands startups US$110.4 million, while the government’s AI Nation 2030 plan is pushing B2B adoption. Yet investors have warned that Malaysia’s startup money has a matching problem, with public capital struggling to find enough private co-investment.

That is why a sandbox has real value for early-stage founders. Buying test equipment, edge compute and network access to prove a 5G-A use case is expensive, and a free bench can save months of runway. For U Mobile, every startup that builds on its stack is a potential channel into an enterprise contract it would otherwise have to win alone.

Rivals are writing bigger cheques

U Mobile is hardly alone in courting founders. Singtel launched a US$250 million AI fund to turn its regional footprint into an AI deployment platform, while the Philippines’ Globe has pitched its ecosystem as a fast track to market entryfor startups. At home, CelcomDigi and Maxis chase the same enterprise customers with deeper subscriber bases.

The contrast is stark. Singtel offers capital; U Mobile, for now, offers infrastructure and introductions. Meanwhile, telcos everywhere are racing to put AI into their own apps, which raises a fair question for founders: will the operator be a customer, a partner or eventually a competitor?

There is a local opportunity, though. As Malaysia leans into sovereign AI and local context as a startup moat, solutions validated on a domestic network with domestic enterprise data could carry an edge that imported tools lack.

Demo days are cheap; purchase orders are not

The scorecard that matters will not be the number of demo days or summit panels. It will be how many sandbox pilots turn into paid enterprise contracts, and how quickly.

Also Read: Malaysia carried Southeast Asia’s IPO market in 2026. The rest of the region went quiet

Neither party has disclosed how many startups are in the pipeline, what share of the sandbox they can use, or how intellectual property developed on EIP is treated. Those details will tell founders whether this is a genuine launchpad or a showroom for U Mobile’s network.

Until then, Malaysian startups get a fast network to play with. Whether anyone pays them to keep playing is the real test.

The post U Mobile opens its 5G-Advanced sandbox to Cradle’s startups, but the cheques are not part of the deal appeared first on e27.