SEA of Startups

Decoding the Pulse of Founders, Capital & Conviction in Southeast Asia.
  • 5.0 (1)
  • Entrepreneurship
  • Updated Weekly

Real, raw, relatable takes on Southeast Asian startups. One investor, the week's news, no script. seaofstartups.substack.com

  1. 5d ago

    Seed is booming. Not here.

    Let me describe a company to you. Real revenue. Customers who pay every month, on time. A product that works and a team that ships. By every standard that mattered three years ago, this is a fundable company. Three years ago it raises a Series A without breaking a sweat. Today it cannot get a second meeting. Nothing about the company changed. The bar moved underneath it while the founder was busy building. That is happening everywhere, in San Francisco, London and Bangalore. But if you are building in Southeast Asia, here is the part almost nobody says out loud. The bar did not move once here. It moved twice. And the second move is one no venture podcast in San Francisco will ever explain to you, because none of them has had to raise a round across six countries at once. So this week I want to walk all the way through it. Why your next round is harder than your last one had any right to be. What it costs in months and in cash. And what I would do on Monday morning if I were sitting where you are. One. The seed mirage The headline going around is that seed funding in Southeast Asia is up this year. On paper, true. DealStreetAsia’s half-year numbers have seed dollars up 128% year on year to $328 million. Read only the top line and you would think the taps are open and the hard years are behind us. Now look at who got the money. In the first half of 2026, 63 companies in the entire region received their first institutional cheque. Sixty-three. Across six countries, 680 million people and six months. That is not a wave of new founders getting funded. That is a number you could seat on one decent coworking floor in Singapore. Here is the part that should make you suspicious. While the seed total went up, the number of deals across the region fell to its lowest level since at least 2018, and the typical seed cheque got bigger. Hold those two facts together. Bigger cheques, fewer of them than in nearly a decade. That is not money spreading out to more people. That is money piling up on fewer of them. Who gets the fat cheques? Overwhelmingly AI and deep tech. A normal software seed in this region is still one to three million dollars. The AI rounds run at roughly double that and climbing, and they are the ones dragging the average up and making the chart look healthy. The chart is not lying exactly. It is letting you believe something that is not true. When you hear “seed is back,” the honest follow-up is “back for whom.” The honest answer is: for a small number of companies that probably do not look like yours. One more number, same story at a bigger scale. The region raised $7.25 billion in the first half, the strongest half since 2022. Almost all of it was one data centre raise, DayOne’s $4.5 billion Series C. Take that single deal out and the region raised about $2.75 billion, down on the half before and below the same period a year earlier. One deal, one building full of servers, is doing the work of making an entire half-year look like a recovery. I made the same point about exits a week ago. The discipline never changes: how many deals? So here is where we stand. The money that exists is real. It is also concentrated, going to a narrow slice of companies, and the breadth a healthy early stage is supposed to have is not there. The costume says boom. The body underneath says something much tighter. Two. The bar moved, and the Americans admit it Now the thing that actually decides your future: the bar for your next round. Listen to the American shows, 20VC, SaaStr, the ones every founder here has on in the background, and the same confession comes up every few weeks. A partner gets comfortable in the chair and says the quiet part out loud. The bar to raise a Series A today would have been a Series B three years ago. They say it the way you mention the weather, and then they move on. Here are the numbers behind the anecdote. The revenue a company needs to clear a US Series A has roughly tripled. A few years ago, about $1 million of annual recurring revenue got you a serious conversation. In early 2026 the figure doing the rounds is closer to $3.5 million. Not double. More than triple. Graduation fell off a cliff too. Carta, which sees a huge slice of American cap tables, tracks the share of seed-stage companies that go on to raise a Series A within two years. For companies that raised seed in early 2018, that was about 31%. For the early-2022 cohort it was about 15%. One in three became one in six. And it takes longer. The median walk from seed to Series A is now past two years, against roughly a year and a half in the late 2010s. Put those together and you understand what people mean when they say a stage of the market got deleted. Seed became the new Series A. Series A became the new Series B. The old Series A, the round you raised on promise and early traction, quietly stopped existing. The companies standing on it got no warning. They raised their seed, hit the milestones that used to be enough, went out for the A, and found the floor had moved while they were heads down. Pre-seed founders are discovering the same thing one step earlier: the seed bar is far higher now unless you are spinning out of a frontier lab. That is the American story. It is brutal on its own, and it is the story your foreign investors carry in their heads whether they say so or not, because the serious players read the market globally. Hold onto it. Now the Southeast Asian version, which is worse. Three. One company, one raise Numbers in the abstract do not land, so let me walk you through a raise. Invent a company. Call the founder Mei. She is in Kuala Lumpur, where I am based, and she has built B2B software for mid-sized logistics operators. It is good. Two years in, she is at about $1.5 million of annual recurring revenue, almost all of it in Malaysia. Her customers renew. Her churn is low. Her team is lean. By the standard of 2021, Mei is a textbook Series A. She did everything the playbook told her to do. So she goes out to raise. Meeting by meeting, here is what happens. The local funds take the meeting, because they know her, they are polite, and they like the business. But the ones who would have led her A two years ago are either not raising a new fund right now or have quietly moved upmarket, chasing bigger, later, safer deals. They tell her she is a little early. Early, at $1.5 million of revenue and two years of history. That word is the sound of the bar moving. Then the regional funds out of Singapore. The first serious question is not about her revenue. It is about her map. How many markets are you in? When she says one, Malaysia with a little spillover, she watches the temperature in the room drop. What they are really asking is whether she is a Malaysian company or a Southeast Asian one. A Malaysian company at $1.5 million is not a regional Series A. It is a nice local business they no longer know how to underwrite, or no longer want to. Then, if she is lucky, a foreign fund. Now she is being measured, silently, against the American number. $3.5 million, growing fast, in a market the investor understands. She is at $1.5 million in a market they would need a fourteen-hour flight to visit. She does not lose that meeting because her company is bad. She loses it because she is graded on a curve drawn in another part of the world. Mei, who did everything right, walks away from a month of meetings with no term sheet, a lot of “keep us posted,” and a dawning understanding that the round she was told to raise does not exist for her at her size. That is not a failure of the company. That is three versions of the bar stacked on top of each other, all moving, and she walked into all three at once. Four. The border tax The second move of the bar, the one San Francisco never has to explain, is what I call the border tax. A software company in San Francisco proves one market. The United States is one language for business, one currency, mostly one set of rules, one enormous wallet. A founder there can reach $3.5 million of revenue without leaving home. They could reach $100 million without leaving home. A company here gets asked to prove three markets. That is the real test gating a regional Series A: not just revenue, but evidence you can cross borders. The whole thesis of a Southeast Asian company is that it is regional, and an investor writing a Series A cheque is buying the regional story, not the single-country one. So the question is never only how much revenue. It is how much revenue, in how many markets, and can you do it again in the next one. Here is what nobody puts on a slide: what market number two actually costs. • A local entity, which means lawyers and a registration process measured in months, not weeks. • Local payment rails, because money moves differently in Indonesia, the Philippines and Thailand, and your billing has to speak all of them, e-invoicing mandates included. • Localisation that goes well past translating the buttons: tax rules, invoice formats and compliance quirks that differ in every single country. • A first hire on the ground who actually knows the market, because selling into Jakarta from Kuala Lumpur over video calls does not work, and everyone who has tried it knows it. Add it up and entering one new market properly is a 12 to 18 month project that swallows a serious chunk of your round and a serious chunk of your attention. The Series A test asks you to have done it two or three times before anyone writes the cheque. Look at what that does to Mei. The American founder gets to $3.5 million by going deep in one market. Mei is asked to go wide across three, each one more than a year and a large pile of cash, on a seed round that was never sized for it. That is the border tax. It is invisible in the headline numbers, and it is the single biggest reason a good regional company stalls between seed and Series A. Not product. Not revenue quality. Geography, and the cost of cr

    Seed is booming. Not here.

  2. Sep 30

    Twelve doors out: Southeast Asia's exits are back, and the buyers are flying in

    There are two ways out of a company in Southeast Asia. One of them is loud. On 20 October, GCash will ring the bell at the Philippine Stock Exchange in the biggest listing that country has ever done. Cornerstone investors with names like BlackRock and T. Rowe Price. A valuation that starts with a number most Filipino founders will never see in their lifetime. The other is quiet. A few weeks earlier, on 8 September, private equity firm Navis Capital closed the sale of a chain of premium Malaysian supermarkets, Village Grocer, Ben’s Independent Grocer and a few others, to a Thai retailer. No bell. No cameras. Just a wire transfer and, reportedly, around five times the money they put in more than a decade ago. For years the complaint about this region was that there were no exits. Great companies, nowhere to sell them. That complaint just stopped being true. The door is cracking open. But before anyone pops the champagne, walk through it with me and look at who is standing on the other side. Because it is almost never a Southeast Asian. One. Six and a half times the money, four more companies Here is the number that made me put this on the docket. In the first half of this year, venture-backed companies in Southeast Asia returned $5.65 billion in exits, per DealStreetAsia. Trade sales, listings, the lot. In the first half of last year, that figure was $856 million. Call it a six and a half times jump in twelve months. Hear only that and you would think the drought is over, and the region has finally figured out how to give investors their money back. Here is the part that gets buried. That $5.65 billion came from twelve exits. Twelve. The year before, it was eight. The value went up more than six times. The number of companies actually getting out went up by four. Sit with that, because it is the whole story of this market right now. The money is real. The breadth is a mirage. Twelve companies is not a wave of liquidity. Twelve companies is a handful of very large outcomes, and everybody else is still sitting in the waiting room. We saw the same trick with funding. The region raised $7.25 billion in the first half, the strongest since 2022. Sounds fantastic, until you notice that deal volume fell to its lowest level since at least 2018, and a single data centre deal carried the headline. Strip that one out and regional funding actually fell. The median Series A cheque shrank. The average founder is not living in the record. The average founder is living in the drought. When you see “exits are back” or “funding hits a record,” your first question is always the same. How many deals? Two. The loud door, and who already has a key The public market. The listing. The dream. And in the Philippines, that dream is about to come true in a very big way. GCash, the wallet on almost every Filipino phone, is going public through its parent, Mynt. The offer could raise around 80 billion pesos, roughly $1.4 billion, and closer to $1.6 billion with the overallotment. At the top of the range, Mynt is worth 669 billion pesos, about $11.5 billion. It is the largest listing in the history of the Philippine Stock Exchange, and it beats the old record, Monde Nissin, comfortably. More than 20 cornerstone investors are in, BlackRock and T. Rowe Price among them. The offer period opens on 6 October. Trading starts on the 20th. I want to be genuinely positive here, because this matters and I am on record as bullish on the Philippines. For a decade, the knock on Manila was that it could not host a listing like this. Any serious technology company would have to go to Singapore or New York to find money deep enough. GCash is proving that wrong on home soil. That is a real milestone, and Filipino founders should take a moment with it. Then I put the investor hat back on and read the cap table out loud, because Real. Raw. Relatable. means I do not just clap at big numbers. GCash is not a scrappy startup three friends built in a Makati garage. Mynt is a joint venture between Globe Telecom and the Ayala group, one of the most powerful families in the country, and Ant Group, the fintech affiliate of China’s Alibaba. The anchor buyers flying in are the giant American asset managers. So the biggest technology listing in Philippine history is, honestly described, a telco and a Chinese fintech giant selling a slice of their joint venture to Wall Street. That is not a criticism of the deal. It is a great deal, and GCash is a great company. It is a question about the pattern. If the only way to a landmark listing is to already be owned by a conglomerate and a foreign tech giant before you start, what exactly does that prove for the founder with no conglomerate behind them? Across the region in the first half, there were 47 listings raising about $3 billion. Healthy on paper. But reliable public liquidity for a venture-backed technology company still exists on a very short list of exchanges. Malaysia will point out that Bursa has been busy, and SkyeChip showed a local chip designer can list at home. Fair. It is still the exception. For most of the region, the loud door is not actually on offer. The public door is open. Just notice how narrow it is, and who already has a key. Three. The quiet door is a grocery aisle Most exits here do not happen with a bell and the start of public trading. They happen with a strategic buyer writing a cheque and quietly absorbing your company. Trade sales are the workhorse. And the clearest picture of this entire market I have seen is a story about groceries. Back to Navis. More than a decade ago, they started putting together a group of premium Malaysian supermarkets under a company called The Food Purveyor: Village Grocer, Ben’s Independent Grocer, BSC Fine Foods and a couple of others. This month they closed the sale for about RM1.66 billion. Good outcome, clean exit, the kind of return that keeps a fund alive. The buyer? CP Axtra, the Thai company that runs Lotus’s. And this is where one deal stops being one deal and starts being a pattern, because this is not the first time CP has gone shopping in Malaysia. Rewind to 2020. CP Group bought Tesco’s entire business in Thailand and Malaysia for $10.6 billion, one of the biggest deals in Asia that year. The Malaysian piece alone was valued at around $3 billion. Seventy-four Tesco stores were rebranded as Lotus’s. Follow the timeline. In 2020, CP buys the value end of the market, the big hypermarkets. In 2026, CP buys the premium end, the shops where the affluent city crowd does its weekly run. Put them together and one Thai conglomerate now has more than 120 outlets across Malaysia, covering everyone from the most price-sensitive shopper to the most comfortable. Top to bottom. The whole aisle. And it is not only CP. The other big premium chain in Malaysia is Jaya Grocer. Grab bought a majority of it at the end of 2021, in a deal reported at RM1.5 to 1.8 billion, roughly $400 million. Grab is a Singapore company. Some Malaysians will argue that point with me, and I am going to keep calling it Singaporean anyway. So let me say the thing plainly. Malaysia’s two leading premium supermarket chains are now both foreign-owned. One Thai, one Singaporean. Malaysian operators and Malaysian investors built these businesses into category leaders, and when it was time to sell, the buyer flew in from Bangkok or Singapore. Each time. That is the exit market in one shopping trip. You build it here. You sell it there. Four. The pattern holds in tech Widen out from groceries and the data backs it up. In the first half, technology M&A in the region fell to 41 deals, down 37% on the year before, according to DealStreetAsia. So even while exit value went up, the number of tech deals went down. Fewer transactions, bigger cheques. The same concentration we started with. And who did the deals that did happen? Foreign buyers completed 30 of the 41. Nearly three in four. Domestic acquirers managed eleven. Japanese corporates, Chinese groups, regional conglomerates like CP: they are the ones clearing the market while local buyers sit on their hands. One more piece, the least glamorous and the most telling. A growing share of what gets counted as an exit here is not really an exit at all. It is one private equity fund selling to another. By some estimates, secondaries are now around 30% of exit value in the region. Funds are getting older. Their LPs want some money back. There is no strategic buyer and no open listing window. So the asset goes to the fund down the street, and everyone calls it liquidity. That is a perfectly respectable strategy. Secondaries are a well-worn path in more mature markets. But put the whole board together. Exit value is up, yes. But the deals are few, they skew large, they are increasingly bought by companies headquartered outside the region, and a big chunk of what is left is just capital rotating between funds. Five. The constraint was never capital The old complaint was that Southeast Asia had no exits. That was never quite right. The truer version, the one I want you to take away, is this: the constraint here was never capital. There has always been money. The constraint is liquidity, the ability to actually turn a company into cash. And the person most able to give you that liquidity today does not live here. Six. What I would actually do with this I do not like diagnosing and walking away. So if you are a founder or an early investor in this region, here is what I would do. 1. Build for a specific strategic buyer, and be honest with yourself about who they are. If you are building a consumer brand in Malaysia or Indonesia, your most likely exit is a regional conglomerate or a foreign strategic, not a listing. Know that on day one. Know their names. Build the thing they will eventually need to own, because buying you is faster than building you. 2. Stop worshipping the listing. The GCash IPO is a beautiful, rare event. It is not your path. F

    Twelve doors out: Southeast Asia's exits are back, and the buyers are flying in

  3. Sep 23

    Count who gets paid: the least glamorous AI money in Southeast Asia

    Somebody is going to stand on a stage in Singapore this month and tell you two things. Southeast Asia now has more than 60 profitable tech startups. And the region is in the middle of a $4 billion AI boom. Both numbers are true. Both are also lies. Not made up. Lies the way your uncle lies at dinner, where every word checks out and the whole thing is still nonsense. So I am going to tell you up front where this lands. The most profitable AI business in Southeast Asia is not building the model. It is not even the thing people are calling the AI boom. It is teaching people how to actually use it. That is a services business. It gets paid on day one, or net 30. And it is the one lane where being a follower is not a weakness. It is the strategy. I am writing this from KL with the haze back, the Twin Towers invisible from three streets away and a throat that sounds like I gargled a satay grill. We cannot see through the sky this week. We can still see through the press releases. One. The 40 people who never logged back in This did not start with a report. It started with watching the same movie play out across my portfolio and my network. A company I am close to spent good money last year buying every AI tool you can name. Seats for the whole team. I did the same thing myself. I had so many subscriptions I lost count, and I justified every one as my R&D budget. Fast forward to today and ask how it is going. Past the slides, the honest answer is this. A handful of people went fully red-pilled and got the lift. The rest logged in once, maybe twice, felt a bit confused or a bit silly, and quietly went back to doing the job the way they already knew how. Nothing was wrong with the tools. The tools are incredible. What was missing was the boring part. Nobody sat down with 40 people and taught them, in plain language, for their actual job, how to make the thing useful before nine o’clock on a Monday. Once you see it, you see it everywhere. Startups, SMEs, corporates. The gap is not the model. The gap is the person in the chair who was never taught, and was too embarrassed to say so. Multiply that person across 680 million people and you do not have a technology problem. You have the biggest teaching job this region has ever faced. And where there is a job that big that nobody has done yet, there is money. Two. Take a wrench to the headline numbers First, the 60-plus profitable startups. It comes from a single list, quoted everywhere, and nobody says who is on it or how profitable is defined. I sanity-checked it against my own portfolio and I would apparently be a meaningful chunk of the region’s total. I am a good solo GP. I am not that good. Here is what I will concede. Directionally, it is right. I sit across from early-stage founders every week, and genuinely profitable venture-backed companies are less rare than they used to be. The capital drought forced it. Treat the number as a direction, not a fact. The second one is worse. Native AI startups in the region raised about $4.1 billion so far this year, roughly double all of last year. Pop the champagne. Now pull the thread. Kling AI, the video generation unit of Beijing’s Kuaishou, raised a $2.8 billion Series D. That one round is about 68 percent of the entire “boom.” Strip it out and you are looking at a fraction of the headline. The number of rounds actually fell, from 41 last year to 23 this year. That is not a boom. That is one cheque and a thinner deal count. Of what is left, almost all of it lands in Singapore. With the usual caveat I will keep repeating until it stops being true: half the region is incorporated in a Singapore holdco, so the geography in these tables is partly an accounting artefact. So put both numbers down. Neither tells a founder in Manila or Bangkok where to make money in AI this year. Here is the rule I want you to carry for the rest of this piece: Do not count logos on a slide. Do not count dollars in one giant round. Count who sells a thing and collects cash. Three. AI projects do not fail. They stall on people. We have all seen the MIT-style reports on enterprise AI projects that never deliver a return. I do not think most of them fail, exactly. They stall. And not because the model is not smart enough, or the tool is badly designed. They stall because nobody knows how to use it, nobody trusts it, and the pilot dies quietly in a slide deck. If you have ever rolled out a CRM at a growing company, you know this movie. The software was never the hard part. Changing the daily habit was. Research on the roughly $400 billion corporate learning market lands on the same point: workforce readiness, not technology, is what blocks most enterprise AI. One survey has three in four companies saying they cannot keep up with their own demand for new skills. They have the tools. They have the budget. They do not have the drivers. That flips the story. If the model is not the moat and the model is not the blocker, the bottleneck in the entire AI economy is human. It is a teaching problem. A change management problem. A get-accounts-payable-to-trust-the-tool problem. That is not a San Francisco problem. That is an on-the-ground, local, human problem. And being close to the ground is one of the few things we are genuinely good at. Four. Sell shovels, not a prayer The AI training and enablement slice of corporate learning is somewhere in the six to seven billion dollar range today, heading toward something like 18 by the end of the decade. Estimates vary and I am not going to pretend to precision. Every serious forecast points the same way: up and to the right, fast. Here is why I care. Teaching people is a services business, and services get paid differently. No eight-year burn. No praying you hit an inflection before the money runs out. A company signs a purchase order, you deliver, you invoice, you get paid. Assuming a good payer. We have all met the other kind. This is the picks and shovels trade of the AI era. You can run at the hills and maybe strike gold. Or you can sell shovels to everybody running, and get paid whether they find gold or not. You do not need a frontier lab or a $200 million round. You need expertise, a curriculum, a customer and a purchase order. That is a business. Not a bet. Five. Three kinds of buyer, all moving at once Demand for this is not spread evenly around the world. It is concentrated here. Surveys have roughly 96 percent of Southeast Asian employers prioritising upskilling, against around 85 percent globally. Young workforce, fast adoption, hundreds of millions of people who came online mobile-first and now have AI in their pocket from day one. The instinct that let this region skip the desktop and skip the credit card is now pointed at AI. And this is not a prediction. Two of these moves landed this month while most people were not looking. Individuals. On 16 September, Singapore rolled out more than 200 AI courses under SkillsFuture, and citizens who take one get six months of a premium AI tool for free, so they can actually practise. A government putting shovels directly in people’s hands. Educators. ASEAN education officials endorsed AI Ready ASEAN, backed by Google.org and the ADB, moving from awareness into applied learning with a target of more than 1.7 million students and teachers by the end of 2028. That is not a webinar. That is a regional workforce being built on purpose. Governments. Here at home, Malaysia’s national AI skilling push with Microsoft, Microsoft Elevate, has already reached 80,000 learners from its pilot. The part I like is who it includes: educators, small business owners, civil servants, even retired servicemen and women. That is a whole-of-country bet that the bottleneck is people and the fix is teaching. A word to my friends in the Philippines, because this one is yours to win. The youngest population, the strongest English, and a workforce that already trains and supports the rest of the world for a living. You have done the BPO. You have done the data annotation. If any country here is built to become the place that teaches the region to use AI, it is you. Do not let anyone tell you that is a small ambition. Demand is not the question. Demand is the one thing we are not short of. The only question is who supplies it, and whether that supplier is one of us. Six. What it looks like as a company Take Gnowbe out of Singapore. AI-powered microlearning for enterprise. The client list is not hopeful startups. It is General Mills, Shangri-La, MetLife. Blue chips paying real money. The clever bit: they are SkillsFuture-accredited, so corporate clients can get up to 90 percent of the training cost co-funded by the government. Malaysia has its own version in HRD Corp. Employers pay into the levy whether they like it or not, and a lot of them genuinely struggle to spend it. There is cash sitting there. Smart operators build straight on top of it. What I respect is that this is a capital-light business. Nobody raised nine figures and set it on fire. It is a book of blue-chip customers, government co-funding pulling demand through the door, quietly compounding. That is what a healthy company looks like when the hype leaves the room. And it is not one company. A business in our portfolio has started running AI training in India and has landed some very large contracts. Someone in my network doing the same in the Middle East tells me they have more demand than they can possibly supply. Corporate bootcamps are running week-long intensives for executive teams. None of it shows up in that $4 billion number, because nobody is counting it. Which is exactly why there is still room. Seven. The plumber is now worth more than the engine builder There is a job title that barely existed three years ago and is now the hottest role in AI: the forward-deployed engineer. The idea is simple. Instead of selling a company a model and wishing them luck, you send an engineer to sit inside the customer an

    Count who gets paid: the least glamorous AI money in Southeast Asia

  4. Sep 16

    Exposing the AI panic: who actually profits?

    This week a guy who, as far as anyone can tell, had never posted anything in his life put up a thread about artificial intelligence and got 170 million eyeballs on it in five days. A former Anthropic employee, or so the byline says. Within 48 hours you had Senator Bernie Sanders talking about banning superintelligence, market people calling for a bloodbath on Monday morning, and half my group chats forwarding the same screenshot to each other. Then, almost on cue, the chief executive of one of the biggest AI labs on earth published an essay saying, and I am paraphrasing, we need to slow this thing down before it gets away from us. And then China told him to get lost. So here is the question I want to sit with. Not “is AI going to kill us all.” That is the wrong question, and frankly it is the question everybody wants you asking, because it sells. The question is this. When the most powerful people in the most powerful technology of our lifetime suddenly all agree you should be afraid, or suddenly all agree you should not be, who actually benefits from that? And the second one, the question that matters if you are reading this from KL, Manila, Bangkok, Jakarta or Singapore. While the big boys fight about the brakes, are we building this future, or are we just being told what the rules are? One. Where I stand, so you know Let me put my cards on the table before I say another word, because I am about to be skeptical of a lot of people in this piece, and you deserve to know where I am coming from. I am an optimist on AI. Full stop. I invest in it. I use it every single day, in my fund, in my other businesses, in how I read the market. I think it is one of the best things to ever happen to a small team trying to compete against a large incumbent. If you are a founder in this region with four people and a good idea, this is the closest thing you have ever had to a fair fight. So I am not here to scare you. But optimism is not the same thing as being naive, and this is the part people get wrong. Every powerful technology in human history arrived with bad actors attached to it. Every single one. Fire, gunpowder, nuclear fission, the internet. Even the ATM got used for fraud. Cheques got used for fraud. The tool does not decide. People do. There will always be rogue states. There will always be criminals. There will always be somebody who wants to take the sharp end of a good thing and hurt someone with it. That is not an AI problem. That is a human problem, and it is as old as we are. So hold both of these at once, because I am going to keep coming back to it. I am bullish. I am also clear-eyed. Those are not opposites. Anybody telling you that you have to pick one is selling you something. Two. Panic has sponsors now Back to our mystery man and his 170 million impressions. Here is what should bother you about that thread, and it has nothing to do with what he actually said. Multiple outlets that dug into it came away thinking the virality itself was manufactured. An account with basically no history, suddenly spiking on numbers that do not behave the way organic posts behave. In plain language: somebody may have paid to make you panic. I do not know that for certain, and neither does anyone else, and that is exactly the point. We now live in a world where the fear itself is a product. Somebody can build it, ship it and sell it. Whether the goal is attention, influence, or moving a stock on a Monday morning, panic has sponsors. So when a piece of content seems engineered to make your stomach drop, the correct response is not to feel the fear and forward it to the family group chat. The correct response is to ask a boring question. Who wanted me to feel this way, and what do they get if I do? That is not cynicism. That is media literacy in 2026. And it applies just as hard to the calm, reassuring voices as it does to the scary ones. Keep it in your pocket, we need it in about two minutes. Three. Sincere and self-serving are not opposites On 12 September, Dario Amodei, chief executive of Anthropic, published an essay called “We Must Pace the Frontier.” Just under 4,000 words, which by the standards of this genre counts as restraint. The core argument is reasonable on its face. He is not saying stop, and he is careful about that. He is saying slow the rate at which these models get more capable, so that the safety work and the outside checking can keep up. His worry is recursive self-improvement, which is a fancy way of saying an AI that gets good enough at building the next AI that the whole thing starts accelerating on its own, faster than any human can supervise it. His ask is that each generation gets aligned and checked by independent evaluators before anyone sprints to the next one. Honestly, as reasoning, that is not crazy. If you were building a bridge and you had a machine that could design the next, taller bridge by itself, you would want to inspect a few of them before you drove your family across. But here is where you put the glasses back on. When the leader of one of the two or three companies out in front says we should all slow down, bring in third-party evaluators and regulate this properly, what world is he also describing? He is describing a world where the companies already ahead get to lock in their lead. Where the rules get written around the way the current leaders already do things. Where it becomes a lot harder and a lot more expensive for a challenger to catch up. Regulation is a wall, and walls protect whoever is already inside the castle. I am not accusing him of dishonesty. This is the nuance I actually want you to take away. He can be completely sincere about the safety risk, and it can also happen to be extremely good for his business. Both of those can be true at the same time. Sincere and self-serving are not mutually exclusive. Grown-ups can hold that. So I do not clap and I do not boo. I note it. A safety argument from a company that benefits from a safety regime is still worth listening to. It is just not scripture. And by the way, some of the reporting says other lab bosses nodded along with him. Be careful there. “Signalled agreement” is doing a lot of work in those headlines, and I have not seen anything I would put my name to on exactly who endorsed what. So I am going to leave it alone rather than build a story on a soft source. You should demand the same from anyone reporting this to you, including me. Four. The paragraph the Western coverage walked past Here is the part of that essay most of the Western coverage glided over, and it is the part that matters most for us out here. The same essay that says let us all slow down together for safety also says, and I am being fair to the text, that a Chinese lead in AI would be a grave danger, and that the United States should keep restricting the sale of the most advanced chips and chipmaking equipment to China. Read those two things back to back. Slow down, everybody, in the name of global cooperation and safety. And also, we should keep choking off one specific competitor’s access to the hardware. China noticed. Of course China noticed. Their foreign ministry called it fearmongering and said, more or less, that confrontation and vicious competition will wreck any chance of real global cooperation on AI. Global Times called the essay a Cold War playbook dressed up as safety. The commerce ministry said the real game is the US trying to hold a monopoly on the industry. Now, same rules apply. I am not going to sit here and tell you Beijing is the innocent party and the noble truth-teller. A government framing a safety debate as an anti-China containment plot is also incredibly convenient for that government. It rallies their own industry. It plays well across a developing world that already carries some anti-American sentiment. And it lets them skip straight past the actual substance of whether frontier AI is dangerous. A rebuttal can be fair and strategic at the same time. Again, hold both. And while we are handing out skepticism evenly, let us not forget Washington. The Trump administration has doubled down the other way, leaning hard into acceleration and winning the race. That is also a position, and it also has a book behind it. Chipmakers, defence contractors, and the simple political value of looking strong. Nobody in this fight is a neutral referee. Everybody has a jersey on. Five. The dancing robot and the soldier robot are the same robot All of this slow down, speed up, contain China material can feel abstract. Percentages, essays, ministry statements. We have been living inside a geopolitical bubble long enough that it all starts to sound the same, and it becomes very easy to tune out. And then last month the abstract got legs. Literally. In Beijing, from 22 to 26 August, they held what people are calling the robot Olympics. The official name is the World Humanoid Robot Games. More than 2,000 humanoid robots. They ran, they boxed, they danced, they played football. Three years ago these machines could barely get down a corridor without a spotter. My first reaction was not fear. It was wonder. I am a bit of a tech nerd, and this is a genuine feat of engineering. A huge chunk of it is coming out of Chinese companies who, by some counts, now ship around 95 percent of the world’s humanoid robots. That is real capability. That is not a PowerPoint and it is not an essay. But here is where it gets heavy, and here is why it fed straight into the panic cycle. Within days of the games ending, the PLA Daily, the official newspaper of the Chinese military, said the quiet part out loud. It called for accelerating the move of this technology from the laboratory to the military training ground. There is a growing pile of reporting on Chinese defence research into humanoids for urban combat, for reconnaissance, for the jobs you do not want to send a human into. There is a state-backed platform, the Fuxi, described as capab

    Exposing the AI panic: who actually profits?

  5. Sep 9

    The tourists went home: who is actually left to fund Southeast Asia

    Last week one of the firms that taught this region what seed capital even looks like quietly admitted that it will not raise another Southeast Asia fund. 500 Global showed up in KL, in Jakarta, in Bangkok more than a decade ago, when honestly almost nobody in the Valley could find these cities on a map, let alone wire money to them. They ran the accelerators. They wrote the small early checks that nobody else was writing at the time. For a lot of founders here, a 500 check was the first time a real Silicon Valley name said yes. DealStreetAsia reported on 31 August that they are not raising another dedicated Southeast Asia fund. Future bets here come out of the global pool. A smaller local team, mostly managing and winding down what they already own. To be fair, and I am going to try to be fair the whole way through this, Khailee Ng pushed back a few days later. Still committed to the region, still managing the existing funds, still looking at new mandates. I believe him. He is tied to this place far more than I am, because I was not born here. And both things can be true at once. You can be genuinely committed to a region and still decide you will not raise a dedicated fund for it. Hold that thought, because the gap between what people say and what they will actually write a check for is the entire point of this piece. And this is not a 500 story. That is just the way in. One. Two canaries, one coal mine Rewind and look at Y Combinator, the most famous accelerator on the planet. In 2021 and 2022 their batches swelled to something like 400 companies. Founders from everywhere, checks everywhere, the whole world invited. Then under Garry Tan they cut the batch back to roughly 100. They stepped back from late stage. And they pulled back from the rest of the world. The Winter 2023 batch was about 21% international. The batch right before it was 42%. Cut in half. Today something like 80% of the batch is an AI company, which surprises nobody. Now look at what those two decisions have in common. The biggest early stage machine in the United States, and one of the most important funds this region ever had, a decade apart, looked at the whole wide world and reached the same conclusion. Smaller. Closer to home. Pointed straight at AI. That is the tide going out. Two canaries in the same coal mine, singing the same note. This piece is about what the tide leaves behind here. Two. The boom is four companies wearing a trench coat Start with the number everybody is celebrating, because that number is part of the tell. Global venture capital just had a record six months. Around $510 billion in the first half of 2026. That is more than all of last year in half the time. If you only read that one headline, you would think we are living through the single greatest funding boom in the history of the industry. Put your skeptical hat on with me, because here is the number that actually matters. Artificial intelligence took roughly 77% of all global venture dollars in that half year. Two companies you already know, OpenAI and Anthropic, raised somewhere around $217 billion between them. Say that a different way so it lands. Two companies, essentially in California, took about 43% of all startup funding on planet Earth. The US on its own was about $412 billion, and 86% of that went to AI. Now the caveat, because I promised I would and because the whole brand of this show is that I do not read you a number without telling you what is wrong with it. These figures move depending on who is counting and what they choose to count. Different reports land on slightly different numbers, so treat them as directionally correct rather than precise. And the definition of an AI company right now is extremely generous. Anyone with an API wrapper and a login screen qualifies. Out here we are throwing data centres and chip companies into the same bucket, and it is a fair question whether hardware belongs in an AI funding total at all. Every version of the number tells the same story anyway. The money is concentrating into the US. Inside the US it is concentrating into AI. Inside AI it is concentrating into about four companies, mostly the two I just named. So when a fund manager gets on a stage and tells you venture is back, ask the simple question. Back where? Strip out those few mega rounds in San Francisco and the rest of the planet is not in a boom. Some markets are recovering. That is not the same thing. The oxygen got pulled into one room, and we are not in that room. Three. This time they have something better to do at home Here is the part that should sting, because it is a change from the old story. This is not the usual complaint where American money ignores us because it does not understand us. This time the US money has something better to do at home. So ask yourself honestly: why would capital that has that choice get on a 14-hour flight to underwrite a logistics startup in Kuala Lumpur, when it can walk three blocks from the office and put a check into a frontier lab that might be worth a trillion dollars? It would not. It is not. That is not a slight against Southeast Asia. That is gravity. The returns moved, so the money moved. Capital is not loyal and it never was. Even the money that does show up in our numbers is not always what it looks like. You will hear that Singapore now captures something like 78% of all Southeast Asian startup funding. Seventy-eight percent, one city. Be careful with that figure too, because a chunk of it is not activity, it is address. Companies born elsewhere domicile a holding company in Singapore for the tax and the safety, and the raise gets counted as Singapore. Strip out the holding company effect and the infrastructure deals, and the amount of genuine early stage, founder-led money landing across the rest of this region is thin. Thinner than the conference panels will tell you. Four. Nobody is funding the funds Which brings me to the question the rest of this is really about. If the tourists have gone home to chase AI, and the money that stayed is huddled in one city, who is actually left to fund the next founder in Manila, in Ho Chi Minh, in Kuala Lumpur, in Bangkok? So I went looking at who is raising money to invest in this region. Not investing. Raising. That is the leading indicator, and almost nobody talks about it, because a lot of it is non-public by regulation and quite a lot of it is non-public by design. A fund that cannot raise cannot write checks next year, no matter how confident it sounds or how good the website looks. Here is what you find, and it should stop you. In the first half of 2026, I could not find a single dedicated Southeast Asia fund that reached a final close. Not one. Only one Southeast Asia focused venture fund closed at all in that window. And as of August, no Southeast Asia headquartered venture manager had reported a final close for the year. My data set may well be incomplete. Some of the paid sources I do not subscribe to will show something I cannot see. But the word on the street matches the numbers I can get to. It is tight. Let me translate that out of finance language. The machine that raises the money, that then funds the founders, has basically stopped for the year. Think about how a fund actually works. A general partner goes to family offices, pension funds, high net worth individuals, sometimes corporates, and at the larger end the sovereigns. That is the raise, and it takes a year, two, sometimes three. Then they spend the following years deploying it into startups. So if nobody is closing a fund in 2026, that is not a bad year. That is a hole in the checks founders will feel for the next several years. Regional deal value is already down, and the tank that is supposed to refill it is running close to empty. Keep that one in your pocket for the next time somebody at an event talks glowingly about this region’s bright future. Bright futures are funded in advance, and right now almost nobody is funding the funds. And before you tell me this is just the small tourist players washing out and good riddance, look at the biggest name of all. Peak XV, formerly part of Sequoia, generally well regarded, announced $1.3 billion in February. Genuinely good news. But it is an India fund and an Asia Pacific fund. Not a Southeast Asia fund. The biggest name that ever carried this brand in this part of the world raised money and pointed almost none of it specifically here. India gets its own dedicated vehicles. Asia Pacific, the big blurry bucket, gets one too. Southeast Asia as its own thing, with its own fund and its own partners? Apparently not. That is what I mean by the herd thinning. It is not always a dramatic exit and a farewell blog post. Sometimes it is just a fund quietly taking the words Southeast Asia off the label and hoping nobody notices. I noticed. And I am guilty of a version of it myself. We added Japan, and to make that work we swapped Southeast Asia for APAC. Five. The three letters that decide who keeps their job So why did the money that was so in love with this region a few years ago go quiet? I am not going to let us off the hook here, because a big part of this we did to ourselves. The answer is DPI. Distributions to paid-in capital. In plain English: of the money the investors gave the fund, how much did the fund actually give back? Not on paper. Not in a markup. In the bank account, in real dollars, returned to their own investors. The honest answer in Southeast Asia is not much, and not lately. Exit value fell about 32% last year, to something like $4 billion for the entire region. Almost every good exit this region ever produced is still clustered back in 2021 and 2022. Since then it has not been slow. It has been quiet. Put yourself in the shoes of a limited partner. You are a pension fund, a family office, a sovereign wealth fund. You put money into Southeast Asia venture in 2019. In 2021 you were told with a straight face and a beaut

    The tourists went home: who is actually left to fund Southeast Asia

  6. Sep 2

    Southeast Asia's Superapps Finally Made Money. Do Not Copy How.

    Three of the biggest companies this region has ever produced just did something they had never done before. They made money. Sea, the Singapore company behind Shopee and Garena, posted a net profit of 487 million US dollars in a single quarter, on 7.8 billion dollars of revenue. GoTo, the Indonesian giant behind Gojek and Tokopedia, booked its second profitable quarter in a row. And Grab, maybe my most hated company in Southeast Asia but the ride hailing app none of us can avoid, is now profitable for a full year and buying back its own stock. If you are a founder, the obvious temptation is to take one lesson from that. Burn. Burn hard, burn big, buy the market, and the profit comes later. That is what the superapps did, and look at them now. I want to talk you out of it. The real story of Southeast Asian tech in 2026 is not that burning worked. It is that Southeast Asia grew up. And the strategy that got these three to profit is the single most dangerous thing a normal founder can copy. THE SHORT VERSION · The burn is not what made them profitable. Discipline in the last two or three years did. The burn is just what they survived. · Two ingredients made it work, and you have neither. A once in a generation land rush, and a backer deep enough to outspend everyone until they died. · Read past the adjusted number. Sea still carries a 6.6 billion dollar hole. GoTo’s real quarterly profit is about 15 million dollars. Grab trades at under a third of its listing price. · Payments is becoming a loss leader for lending. Which means the profit engine of regional fintech is a bet on the credit cycle. · The funding window is not shut. It got selective. That is good news, if you are building something that actually works. 1. The scoreboard nobody reads past the headline Start with the good news, because it is genuinely good. For a decade the knock on Southeast Asian tech was simple. These companies do not make money. They set investor cash on fire to buy growth, and one day the music stops and there is nothing underneath. I said versions of that myself. A lot of us did. Well, the music got quieter, and they adapted. Sea is the cleanest win. Full year 2025, just under 23 billion dollars of revenue and about 1.6 billion dollars of net income. Not adjusted, not massaged. Actual net income. The stock ran from around 102 dollars in late July to about 130 on the back of the quarter. If you held Sea through the pain, you got paid. Now read the fine print, because the fine print is the whole point. Sea The headline: 487m US dollar net profit in the quarter. 1.6bn net income for FY2025 on about 23bn of revenue. The fine print: Accumulated deficit still around 6.6bn dollars. Two great years have refilled a bit more than 2bn of the hole they dug. GoTo The headline: Second straight profitable quarter. First time crossing 1 trillion rupiah of adjusted profit. The fine print: Actual bottom line net income for the quarter was about 252bn rupiah, roughly 15m US dollars, on a company that size. Grab The headline: Just under 1bn dollars of quarterly revenue, up 22 percent. Adjusted profit up more than half. 750m dollar buyback. The fine print: Trades around 3 dollars and change against a listing price near 11. Day one shareholders are still deep underwater. The number the headlines love is adjusted, and adjusted quietly removes things like share based compensation and financing costs before it shows you the shiny figure. The burn was vindicated at Sea. It was survived at Grab. It is still being paid off at GoTo. Three outcomes from one strategy, and only one of them is the fairy tale. 2. Why the burn worked, and why you have neither ingredient Even for Sea, burning only worked because two things were true at the same time. You need both. You almost certainly have neither. One, this was a genuine land rush. A region of more than 600 million people came online, on mobile, all at once, in categories that really are winner take most. Ride hailing. E commerce. Food delivery. When the prize is a whole region and the market collapses to one or two winners, spending to be that winner can make sense. Most businesses are not like that. Most markets have room for several healthy players, and there, buying share with subsidies just trains your customers to leave the moment you stop paying them. Two, and this is the one people forget. The burn was funded by a capital pipeline almost no founder on earth can access. Sea had Tencent. Grab had SoftBank. GoTo had SoftBank, Alibaba and the biggest names in Indonesia. These companies did not out innovate their rivals. They out funded them. They raised so much, from backers so deep, that they could afford to lose money for longer than the competition could stay alive. The capital was the moat, not the app. So when a founder tells me they are going to run the Grab playbook in their category, I ask one question. Who is your SoftBank? If you do not have a backer willing to fund losses all the way to the horizon, and a market that pays exactly one winner and kills the rest, you are not running the Grab strategy. You are running the part of the Grab strategy that bankrupts everyone who is not Grab. Here is the quiet irony in the numbers. What actually turned these three profitable was not the eight years of losses. It was the last two or three years of doing the opposite. Cutting the burn. Killing the vanity lines. Focusing. For 99.9 percent of founders here, growth at all costs is not a strategy. It is a faster way to die. 3. The same pitch is back, with AI on the label I am banging on about a decade old strategy this week for a reason. It is being sold again, right now, with AI on the front. Same pitch. There is a land grab, get big fast, spend whatever it takes, profit is a later problem. Globally the frontier labs are raising tens of billions for the model race. In this region the AI funding year is basically one giant cheque. For a lab with a bottomless backer, fine. Maybe that is a real land rush, and there is going to be a lot of lost capital when the winner take most reality arrives. But if you are a founder in KL, Manila or Bangkok, building a product on top of somebody else’s models and burning your seed round on compute to grow before you charge, hear me clearly. You are copying the one part of the playbook that only ever worked for a tiny handful of companies with sovereign sized money behind them. That is land grab cosplay without the balance sheet to survive it. The lesson from Sea, Grab and GoTo is not burn to win. It is get disciplined and survive long enough to matter. Bank that one. Throw the other one away. 4. Payments finally makes money. Just not from payments. The most useful signal in the whole quarter, read properly, is where GoTo’s profit came from. Not ride hailing, which is now under pressure from pricing regulation anyway. It came from fintech. GoTo’s fintech arm overtook the on demand business to become the single biggest profit contributor in the company, with adjusted profit in that unit up more than 400 percent year on year. Law of small numbers, sure, but the direction is clear. It is not just the giants. In the Philippines, PayMongo hit breakeven early this year with gross profit up 240 percent, a company that was bleeding cash a couple of years ago and turned itself into a real business under new leadership. Easy headline: fintech is booming, payments won. Here is the trap inside it. The actual fee on a payment in this region is collapsing, by design. In Indonesia the QR standard, QRIS, charges a merchant about 0.3 percent, and the central bank waived the merchant fee entirely on transactions under 500,000 rupiah at the end of 2024. Compare that to a Visa or Mastercard swipe at one and a half to three percent. Governments across the region, and globally, are building payment rails that are near free on purpose, because they want financial inclusion, not fee income. Add stablecoins. Add cross border interoperability, my Maybank QR working in Thailand, the Philippines, Singapore. The squeeze only gets worse. So how is anyone profitable? Because the payment is not the business anymore. It is the hook. The money sits on top of it. · Lending. GoTo’s profit engine is credit. · The float. Interest earned on balances sitting in the system, especially with central bank rates where they are. · Cross selling. Credit to the merchant and to the consumer. PayMongo’s growth product is literally called Capital, and it lends merchants money against future sales. The payment gets you the customer and the data. The loan makes you the margin. And the caveat I cannot skip, because this is not all sunshine. A profit engine built on lending in emerging markets is a bet on the credit cycle. It looks fantastic while rates are high and everyone is paying their loans back. It looks very different in a downturn, when the same loan book that made you profitable turns around and bites. We have watched that movie in peer to peer lending, in fractional purchases, in every iteration of consumer credit that got ahead of itself. So when someone tells you fintech is the new profit machine of Southeast Asia, the right response is: great, now show me the loan book, and show me what happens to it when things get hard. If you are building a pure payments startup charging a fee per transaction, be worried. That is a race to zero, and the state is the one pushing it there. 5. The money did not leave. It got smart. Last one, for the founders who have been told all year that the funding window is shut. It is not shut. It got selective. And selective is good news if you are building something real. Late in August, a Vietnamese company called M Village, now rebranded Modern Village Lifestyle, raised 26 million dollars led by Japan’s Mizuho with Trip.com adding on, at a valuation north of 120 million dollars. More than double its last round. An up round, in this market. Underneath it: a proven o

  7. Aug 26

    Next, Owned by Whom?

    As you read this, thousands of people are in a convention hall in Bangkok. Techsauce Global Summit, 26 to 28 August, at the Queen Sirikit National Convention Center. Three hundred sessions, hundreds of exhibitors, and one word on every lanyard and every panel title. The word is next. The race to the next model, the next unicorn, the next Thailand. So I want to be a little contrarian this week and not talk about next. I want to talk about what Thailand has already won, because it has won real things and nobody on that main stage is going to say them plainly. And then I want to ask the only question that actually matters when the music is this loud and the capital is this cheap. Not what is next. Next, owned by whom? Because here is the number I cannot get out of my head. This year Thailand pulled in more than two billion US dollars of AI data centre money. Microsoft, Google, ByteDance, the whole parade. Thai AI startups, the actual companies with founders and cap tables, raised about four million. Same country. Same year. Two billion and four million. Say those back to back and you barely need the rest of this piece. But stick around, because the gap between them explains almost everything about Thai tech right now. The country is the host, not the guest of honour Start with the money, because a lot of it is real. Microsoft has committed over a billion dollars to Thai cloud, AI and data centres running from 2026 through 2028. Google opened a Bangkok cloud region in January and put a number on it, forty billion dollars of economic value over five years. ByteDance is building too. National IT spend this year is heading toward 1.1 trillion baht. If you are the Thai government, this looks like winning. The photos are great. Prime minister, hard hats, ribbon, a rendering of a big grey building with a lot of cooling on the roof. I want to be fair, because this is not nothing. Data centres mean construction, power upgrades, some high-skilled jobs, a reason for engineers to stay in the country. Real assets get built. That matters. But I have been doing this long enough to have one reflex when a big number lands on the table. I ask who owns the thing the number is attached to. So let us do that. Who owns the compute in those buildings? Microsoft, Google, ByteDance. Who owns the models running on it? The same names. Who owns the margin, the recurring, high-fat software margin that compounds for twenty years? The same names again. And what does Thailand own? The land. The power lines. The water. The local joint venture partner who helped with the permits. The electricity bill. There is a phrase for this and it is not AI hub. It is compute arbitrage. Thailand is renting out cheap power, cheap land and a plugged-in population of seventy million to companies headquartered eight thousand miles away. The country is the host. It is not the guest of honour. Now hold that against the four million dollars that actual Thai AI founders raised this year, per DealStreetAsia data. A country can host the entire weight of global AI on its soil and fund almost none of its own. Capex or equity: the question to ask all week Here is the part I want founders to hear, because it is where this gets useful and not just cynical. When you are at Techsauce this week, or reading about it afterwards, and somebody says Thailand’s AI investment hit record levels, stop and ask them one thing. Capex or equity? Data centre capex is offshore-owned infrastructure. Startup equity is Thai-owned companies. They are two completely different things, and this week they will be blended into one flattering headline every single time. Capex builds someone else’s asset on your land. Equity builds your asset. Full stop. The win for Thailand is not the ribbon cutting. It is the day a Thai company owns a real layer of this, something above the substation and the security fence. Until then, be honest about what you are celebrating. You are celebrating being a very good landlord. There is nothing wrong with being a landlord. Ask Singapore. But a landlord does not get wealthy the way the tenant does, and the tenants here are trillion-dollar American software companies. Just know which side of the lease you are on. The champion, and the question underwriters will circle If the hyperscalers own the infrastructure, where is the Thai champion? Where is the homegrown AI company we can actually point to? There is one. It is called Amity. And it is genuinely impressive, which is what makes this interesting. I am not here to knock it. I am here to hold it up to the light. Amity raised a hundred million dollar Series D in March, one of the largest generative AI rounds in all of Southeast Asia. Revenue over a hundred million a year, up more than tenfold since 2022, targeting two hundred million by the end of this year, and lining up an IPO in 2027. Enterprise AI, chatbots and agents for big organisations: banks, government, retail. By any measure this is a real company. Not a deck and a dream. Revenue. Now here is the detail that tells you a lot about how Thai tech actually works. Amity was founded by Korawad Chearavanont, grandson of the founder of CP Group and son of CP Group’s current chief executive. CP is the largest private company in Thailand. It owns 7-Eleven across the country. It is the majority owner of True, one of the two big telcos. It is in food, agriculture, retail and property, and basically everything you touch in a Thai day. So picture the founding conditions of this startup. You are building an enterprise AI company, and your first potential customers include the biggest convenience store operator in the country, one of the two largest telcos, and a web of banks and companies that all sit inside or next to your family’s group. CP All, the 7-Eleven operator, True and PTT all show up as Amity clients. I need to be precise here, because the lazy version of this take is wrong and I do not want to do the lazy version. The lazy version is that it is just selling to daddy’s companies. That is not accurate. Reporting is clear that only about 35 to 40 percent of Amity’s revenue is generated in Thailand at all, and the company says the majority of its revenue and customers are now outside Thailand entirely. So it is not a captive shell. It has gone and won business in other markets. Credit where it is due, and the accurate version is more interesting than the lazy one anyway. Amity may have been born captive and then grown out of it. A family conglomerate handed a young founder the one thing no other founder in Bangkok can get at any price: a guaranteed home market. Banks, state enterprises, 7-Eleven and True as your first reference points, all while you are still figuring out the product. In enterprise software, a trusted first customer who will not churn while you learn is extremely valuable. He got a whole portfolio of them at the birth of the company. And look, if I am the investor writing that Series D cheque, I do not hate this. A de-risked launch market is a feature. It is plausibly why the thing scaled from ten million to a hundred million of revenue in two years, after taking almost a decade to reach the first ten. The family base was the launch pad. But here is the question I would put on the table if I were sitting across from them, and it is the question underwriters are going to circle before that 2027 IPO. How much of that hundred million plus is arm’s length, won on the merits from customers who could have picked anyone? And how much is related party, business that flows because of the name on the building? That is not an attack. That is diligence. It is the single most important number in the whole story and it is the one that never makes the press release. If most of it is arm’s length, Amity is a genuine Southeast Asian software champion, and I would love to say that loudly. If a big chunk of it is the family buying from the family, then the valuation is telling you a story the revenue quality does not back up. I do not know the answer. Neither do you, and neither does most of that convention hall. But now you know the question. When a Thai founder tells you their startup is crushing it with enterprise logos, the first thing to gently ask is: whose enterprise? And for the ninety-nine percent of Thai founders who do not have a billionaire relative, the lesson is brutal and important. Your hardest first year is finding a customer who will trust you before you have proof. Amity basically skipped that year. You will not. Plan for it. First to adopt is not the same as first to win Here is where Thailand genuinely leads, and where the lead is thinner than it looks. Thailand has the highest AI adoption in ASEAN. Not second, first. An AWS report this year puts it plainly: 43 percent of Thai businesses now use AI consistently, up from 32 percent a year ago. Ninety percent of Thai students use generative AI. The country took to this faster than anyone else in the region. If you only read the headline, Thailand looks like it is running away with it. Then you read the next line, and the next line is the whole story in one statistic. Seventy-four percent of those adopters are stuck at what the report calls the basic stage. Off-the-shelf chatbots, ready-made tools, someone in the office with ChatGPT open in a tab. A few people paste things into it. The actual work, the workflows, the decisions, the output, has not really changed. Longtime listeners will remember Tiwa York on this show calling this exact thing the level 1.5 trap. A company where AI is bookmarked, mentioned in the strategy deck, played with by a few keen people, and yet nothing downstream is different. He called it adoption theatre. Thailand has just gone and proven you can run adoption theatre at the scale of an entire nation. I want to be careful, because first to adopt is not worthless. It means the workforce is not scared of these tools. It means a founder in Bangkok is selling in

    Next, Owned by Whom?

  8. Aug 19

    Industrial Policy in a Startup Jersey

    There is a company in Singapore that has raised close to half a billion US dollars and has never sold a single thing. Not a small amount of revenue. None. It has a chip. It has a box. It has a founder story, a serious set of engineers, and one of the bigger names in Singaporean venture writing the cheque. What it does not have is a customer you can name or a price you can look up. And here is the part that should make you sit up. That is not a scandal. In 2026, that is a business model. So this one is about a Singapore chip company called Acrab. But Acrab is really just the door. Behind it sits the question every founder and every investor in this region needs an answer to. When a company raises hundreds of millions before it has a product in the market, is that smart money seeing the future early, or is it the dumbest money in history dressed in a lab coat? And underneath that, something more practical. This whole idea of running powerful AI on a small box on your desk instead of in the cloud. Is it real, or is it a slide in a deck? The box On 6 August, Acrab closed a $130 million Series B led by Vertex Ventures SEA and India, together with Vertex Growth. Hold that name, Vertex, in your head. We are coming back to it, and it turns out to be the most interesting thing in the whole story. That round took Acrab past $480 million in total funding. The company was founded in 2024 and only came out of stealth this year. By its own description it has visible paths to industrial deployments and expects revenue within 2026. Read that last part again. Expects revenue within 2026 is a very polite, very well lawyered way of saying we have not made any money yet. So what did they build? This is where it gets genuinely interesting, and I do not want to be cynical about the engineering, because the engineering looks real. The chip is called GELIX 1. It is a system on a chip built on a 5 nanometre process, which is a serious modern manufacturing node, pairing a 20 core Arm CPU with a dedicated neural processing unit and 273 gigabytes per second of unified memory bandwidth. If none of those words mean anything to you, here is the translation. This is a real piece of high end silicon, not a science project. Around it they built a product called Agent Box. Think of it as a small computer that sits on your desk and runs AI models locally, on the device, without sending your data off to a data centre somewhere. Acrab says it is designed to support models in the 100 billion parameter class, with persistent memory, multimodal input, and agents that actually do things, all running on the box in front of you. That is the pitch. Now let me do the part I actually get paid to do, which is not believe the pitch. The headline number is a benchmark. Acrab claims its chip processed a prompt at 1,416.8 tokens per second, against 188.9 on a Mac Mini with an M4 Pro. Roughly seven and a half times faster than a high end Apple machine. Sounds incredible. It might be. But let me tell you what that number actually is, because this is where you earn the right to have an opinion. That measurement is something called prefill. Prefill is the speed at which the machine reads and digests your question before it starts answering. It is the reading phase, not the writing phase. It is genuinely useful, and it is exactly one half of the job. The other half, where the machine generates the answer token by token, is called decode. Acrab has not published its decode rate. It has not published time to first token. It has not published power consumption. It has not published thermal data. And crucially, nobody independent has verified any of it. Every one of these numbers comes from Acrab. Even the tech press covering the launch flagged this openly. The claim is prefill only. It is not proof the chip is seven and a half times faster at everything. It is proof it is fast at one specific thing, in one specific test, that the company designed. And the 100 billion parameter claim, the one doing the work in every headline, notice the wording. Designed to support. Not runs. Not we measured it running a 100 billion parameter model at this speed while drawing this many watts. Designed to support is a target, not a result. I am not saying anyone is lying, and I want to be clear about that. I am saying that as of today, GELIX 1 is in validation. It is chasing its first industrial adoption. The first market is the smart cockpit inside cars. No named customer, no public price, revenue as a forecast. So here is your honest scorecard on the product. Real chip, real team, one genuinely fast benchmark on the reading half of the task, and the entire set of numbers that would tell you whether this actually works quietly missing. Half a billion for nothing is now normal Before you laugh at half a billion dollars for a company with no revenue, look at what is happening everywhere else. Acrab is not the strange one. Acrab is the small one. San Francisco. Safe Superintelligence, founded by Ilya Sutskever, formerly chief scientist at OpenAI, raised $2 billion at a $32 billion valuation. Around 50 employees. No product. No published research. And it has said out loud, on purpose, that it does not intend to ship anything until its mission is complete. Nvidia is one of its backers. Two billion dollars, fifty people, nothing to sell, by design. Thinking Machines, founded by Mira Murati, formerly chief technology officer at OpenAI, raised roughly $2 billion at a $12 billion valuation, then another $5 billion at $50 billion. To its credit it did finally ship a model in July. But it raised the first two billion on a founder name and a thesis, not on a product. And this is not fringe. In 2025, OpenAI and Anthropic between them absorbed around 14 percent of every venture dollar invested anywhere on earth. Anthropic was valued at $965 billion by May this year. The money is not spread out. It is piling into a handful of bets on faith. That is America. Here is where I actually wanted to take this, because Asia is running the same playbook with local accents, and almost nobody here describes it that way. Tokyo. Sakana AI raised $135 million at a $2.65 billion valuation, making it the most valuable startup in Japan. And here is the elegant part. Sakana’s entire philosophy is small. Cheap, efficient models trained on small datasets, tuned to the Japanese language and Japanese business. One of its founders, Llion Jones, co-wrote the original research paper that created the modern AI boom. Rich valuation, thin revenue, sold on pedigree and a thesis that says bigger is not always better. Sit with that, because it is the same bet Acrab is making. The anti-giant bet. Do more with less. Beijing. Moonshot AI raised $2 billion at a $20 billion valuation and is reportedly chasing $30 billion. But here is the twist, and this is the one that should sting. Moonshot is not pre-revenue. Its annualised recurring revenue passed $200 million in April. Its rivals Zhipu and MiniMax already listed in Hong Kong in January. The Chinese champions are not raising on faith. They are raising, making money, and ringing the opening bell. Bangalore. India has decided AI is a sovereignty question. Sarvam raised $234 million at a $1.5 billion valuation, led by the Indian IT giant HCLTech. And Ola’s founder Bhavish Aggarwal is building Krutrim, which is not just making models but building its own chips and its own computing infrastructure. That is a direct echo of what Acrab is doing, wrapped in the flag of national self-reliance. Put the map together. Silicon Valley is doing pre-product on pure faith. Japan is pricing efficiency on pedigree. China is already making money and ringing the bell. India is treating this as national security. And Southeast Asia’s entry into the contest is a box from Singapore that runs AI models on your desk. Pre-revenue is not pre-product One distinction, and it is the difference between a lazy take on this and a correct one. Pre-revenue is not the same as pre-product. Safe Superintelligence has literally nothing to ship, deliberately. Acrab has a chip and it has a box. And a chip company is not a model company. Silicon costs a fortune before you make a single dollar, because you have to design it and manufacture it long before anyone can buy it. That is the nature of hardware, not a warning sign. So if you are going to judge Acrab, do not judge it against a software lab in California. Judge it against Sakana in Tokyo and Krutrim in Bangalore. Judge it against the other people trying to build hardware and sovereignty in Asia. That is the fairer fight. Where the engine was built So who runs it? The chief executive is Dr Ken Phua, the former co-chief executive of Arm China. He spent two decades at Arm and stepped in to co-run its China business in 2020, during a very messy boardroom fight. Let that land. The technical spine of this Singaporean deep tech contender was built at the top of Arm China. I want to be careful and fair here, because this is exactly the kind of thing this show exists to say out loud without turning it into a smear. That is a real, world class credential. It is not a red flag on its own. But it is the same thread we pulled on with Manus and with DayOne. When something is wearing a Singapore jersey, it is always worth asking where the engine was actually built. Here, the engine has an Arm China lineage. There is also a small mystery I could not fully close. A couple of outlets name someone called Tian Qi as the 2024 founder, and that name appears almost nowhere else, while Phua is the public face. I am flagging it, not asserting it. If you know who else is in there, tell me. So is edge AI real, or is it a slide? Now the question the whole thing hangs on, in the plainest language I can manage. Edge AI just means running the model on the device in front of you, instead of sending your request over the internet to a giant data centre and waiting for the answer to bounce back

    Industrial Policy in a Startup Jersey
See All (93)

Ratings & Reviews

About

Real, raw, relatable takes on Southeast Asian startups. One investor, the week's news, no script. seaofstartups.substack.com

Information

  • Creator
    Decoding the Pulse of Founders, Capital & Conviction in Southeast Asia.
  • Years Active
    2022 - 2026
  • Episodes
    93
  • Rating
    Clean
  • Copyright
    © Sea of Startups
  • Show Website
    SEA of Startups