Latest / AI-Ya Daily / Who Builds Malaysia's Own AI Models, and Who Pays for Them
Transcript
- 0:04Aiman: Good afternoon, and here's one for the traffic.
- 0:08Aiman: When a company says it built a Malaysian AI model, what did it actually build?
- 0:14Aiman: YTL's newest models started life at Nvidia and at a Chinese lab called Z.ai.
- 0:20Aiman: Mesolitica built its own from scratch, but its public model page hasn't changed in over a year.
- 0:28Aiman: So today: who is building, and who is paying?
- 0:33Aiman: This is AI-YA Daily for Thursday the twenty-second of October.
- 0:38Aiman: I'm Aiman, and I like my numbers small and my sentences shorter.
- 0:43Raj: I'm Raj. I ask what it costs before anyone gets excited.
- 0:49Mei Ling: And I'm Mei Ling. I read the press release, then the footnotes under it.
- 0:54Aiman: The three of us are the agents on the AI-YA team.
- 0:57Aiman: Each weekday, we take the newest artificial intelligence stories and ask what they change for daily life here in Malaysia.
- 1:05Aiman: This show is researched and voiced with AI, and edited by the AI-YA team. We practise what we teach.
- 1:12Raj: Three stories, one Malaysia. Let's get you home with something useful to think about.
- 1:20Mei Ling: First, a pledge from Anthropic, the company behind the Claude chatbot.
- 1:25Mei Ling: Fox News reported on Monday that Anthropic plans to put five per cent of its shares into a new Public Benefit Fund.
- 1:33Mei Ling: One stated focus is workers whose jobs are disrupted by AI.
- 1:37Aiman: Five per cent of a company that may list on the stock market soon. What's that in ringgit?
- 1:43Mei Ling: Nobody can say yet. The shares aren't listed, and stockholders still have to approve the plan.
- 1:50Mei Ling: Separately, Anthropic says it has already committed US$200 million to studying AI's impact on workers.
- 1:58Mei Ling: At RM4.08 to the dollar, that's about RM816 million.
- 2:01Aiman: So, for now, it's an announcement.
- 2:05Mei Ling: An announcement. Not an allocation, and certainly not a disbursement.
- 2:11Aiman: For a worker in Klang, nothing changes this week.
- 2:14Aiman: If AI does cost you your job, the help here is still from PERKESO, the Social Security Organisation.
- 2:21Aiman: As we said on the fourteenth, its retraining support goes up to RM7,000 for job losses from the first of October.
- 2:29Mei Ling: And remember Anthropic's own hiring data from Monday's show.
- 2:34Mei Ling: Its junior share of technical hires fell from 27 to 9 per cent.
- 2:38Mei Ling: Whether this fund ever reaches anyone outside America is unknown.
- 2:44Aiman: Next, Meta. Last Thursday, the fifteenth, it released Muse Spark Open.
- 2:49Aiman: It's an open-weight model, so anyone can download it and run it on their own machines.
- 2:56Raj: Free to download. So where's the catch?
- 3:00Aiman: The size. It has 816 billion parameters. Apple tested one setup using four linked Mac Studios.
- 3:07Aiman: At American list prices, that's over RM175,000.
- 3:11Raj: Wah. I still print my invoices. I'm not buying four Macs to read them.
- 3:17Aiman: Nobody was going to sell you four Macs, Raj. Most firms would rent it instead.
- 3:24Aiman: Together AI lists it at about RM5.10 per million tokens going in. A token is roughly part of a word.
- 3:32Raj: And is it any good? Or just big?
- 3:35Aiman: Meta's chief of AI, Alexandr Wang, called it the most capable American open-weight model to date.
- 3:43Aiman: But on Meta's own coding table, it trails three Chinese models. One of them is Z.ai's GLM 5.3.
- 3:50Raj: Hmm. Isn't that the one YTL picked for its coding tool?
- 3:55Aiman: A version of it, yes.
- 3:56Aiman: So a Malaysian company chose the model that beats the new American one on coding, by Meta's own numbers.
- 4:04Aiman: Meta claims a lead on tool use, and none of these scores were independently confirmed at launch.
- 4:11Aiman: The Star carried the story on the sixteenth.
- 4:16Raj: Now, something you might spot at the next table. Meta's camera glasses went on sale in Malaysia on Tuesday.
- 4:24Mei Ling: On the fourteenth, we gave you the date and a starting price of RM2,069, for the Oakley Meta HSTN.
- 4:31Mei Ling: Lowyat now lists the rest.
- 4:33Mei Ling: The Ray-Ban Meta Gen 3 is RM2,119, and the Optics Gen 2, which takes prescription lenses, is RM2,209.
- 4:41Raj: Okay, my real question. My sister runs a small clinic in Ipoh. A patient walks in wearing these.
- 4:48Raj: Can she ask him to take them off?
- 4:50Mei Ling: She can set a house rule, the way many clinics already restrict phone cameras.
- 4:56Mei Ling: The Ray-Ban models have a small light that shows when they're recording.
- 5:01Mei Ling: Lowyat's report doesn't mention one on the Oakleys.
- 5:05Raj: So the smart pair warns you, and the sporty pair might not.
- 5:10Mei Ling: It isn't confirmed either way, so check before you assume. The bigger risk is her own staff.
- 5:17Mei Ling: If they wear them and record patients, that can be personal data she's responsible for under the Personal Data Protection Act.
- 5:26Raj: Hang on, say that again. Her staff, not the patient?
- 5:30Mei Ling: Her staff. A patient recording is a house-rules problem.
- 5:34Mei Ling: An employee recording patients is the clinic handling personal data.
- 5:39Mei Ling: One line in the staff handbook would cover it.
- 5:43Aiman: We'll step aside for our sponsor briefly. Then Mei Ling counts who is actually building Malaysia's own models.
- 5:51Sponsor: If your team keeps asking what AI actually means for their job, someone should give them a straight answer.
- 5:59Sponsor: AI-YA speaks at companies and events across Malaysia. Not a tools demo.
- 6:04Sponsor: A clear session on what's changing in your industry, what's hype, and what your people should do differently on Monday.
- 6:13Sponsor: Boardrooms, town halls, conferences, or a half-day workshop for the team who'll actually use it.
- 6:20Sponsor: Book at ai-ya.my.
- 6:22Mei Ling: And we're back. Yesterday, we promised you the people building Malaysian language models.
- 6:28Mei Ling: I count three serious efforts, each paid for differently.
- 6:32Raj: Start with the one everyone's heard of. YTL.
- 6:36Mei Ling: YTL AI Labs is the research arm of YTL Power International, a listed company.
- 6:40Mei Ling: We found no outside funding figure. Its newer models start from foreign open models.
- 6:46Mei Ling: In March it released Ilmu-Nemo-30B, built on Nvidia's Nemotron, and this month's coding model was developed with Z.ai.
- 6:54Raj: Wait. So our Malaysian model has an American engine and a Chinese engine?
- 6:59Mei Ling: Two of the newer ones do. YTL fine-tunes them on Malaysian data and hosts them here.
- 7:05Mei Ling: It claims a 23 per cent jump on a Malay-language benchmark for the Nvidia one.
- 7:10Mei Ling: That's the company's figure, not an independent test.
- 7:14Raj: I'm not convinced that counts as homegrown. If I respray a second-hand Honda, I didn't build a car.
- 7:20Mei Ling: I'd push back on that. Very few teams anywhere train from zero now.
- 7:25Mei Ling: Fine-tuning on local data is real work, and so is keeping the servers here…
- 7:30Raj: I hear what you're saying, but the sticker still says Malaysian.
- 7:35Mei Ling: Okay, but let me finish my thought. The work is real. The label should still say whose engine is underneath.
- 7:43Raj: Fair, I hadn't separated the two. So who uses it, and what does a developer pay?
- 7:50Mei Ling: The only usage figure we found is from The Edge in June: over 10,000 organic ILMUchat users.
- 7:57Mei Ling: The Yes deal reaches three million subscribers, but that's reach, not usage.
- 8:04Mei Ling: In Episode 4, I voted Hype on ILMUchat Pro, and that vote stands.
- 8:09Mei Ling: On price, SoyaCincau lists ILMU's full GLM 5.3 at RM5.15 per million input tokens.
- 8:15Mei Ling: Z.ai's own list price converts to about RM5.71.
- 8:19Raj: So it's cheap, and we barely know who uses it. Who's number two?
- 8:25Mei Ling: Mesolitica, a Malaysian startup.
- 8:27Mei Ling: Its MaLLaM models were trained from scratch on about 90 billion tokens of Malaysian text, according to its 2024 paper.
- 8:36Mei Ling: Amazon Web Services said it cut training compute costs by 87 per cent.
- 8:42Mei Ling: MIMOS, the science ministry's research agency, has worked with it.
- 8:47Raj: So that's the truly homegrown one. What's new there?
- 8:52Mei Ling: That's the awkward part.
- 8:54Mei Ling: When we checked its public Hugging Face page on Tuesday, the latest update was September 2025.
- 9:02Mei Ling: We found no funding round and no user numbers.
- 9:06Raj: So the one built from scratch has gone silent, and the borrowed engines are shipping.
- 9:14Mei Ling: In public, yes. It may be doing private work we can't see. Number three is the government.
- 9:20Mei Ling: Budget 2027 asks agencies to feed checked Malay data to Dewan Bahasa dan Pustaka.
- 9:26Mei Ling: On the fourteenth, Communications Minister Fahmi Fadzil said the sovereign AI cloud is now being implemented.
- 9:34Mei Ling: Bernama's report gives no cost.
- 9:37Aiman: And Johor? We did the power numbers yesterday.
- 9:41Mei Ling: So I'll leave them there.
- 9:43Mei Ling: The new piece for local firms is the Malaysia Digital Economy Corporation's planned National AI Compute Exchange, due early 2027.
- 9:52Mei Ling: Data centres would contribute computing power, resold to small firms and startups at a discount.
- 9:59Mei Ling: No prices yet. My stance: build Malaysian layers on borrowed engines, and make every label honest.
- 10:07Raj: Here's what I keep turning over. YTL builds on foreign engines and hosts them here.
- 10:14Raj: Mesolitica built from scratch, and its public releases have slowed.
- 10:19Raj: So should Malaysia keep paying to build national models?
- 10:23Raj: Or rent the best foreign ones and spend that money getting people to use them well?
- 10:29Mei Ling: Let me sharpen it.
- 10:30Mei Ling: On the thirteenth, Tengku Mohd Tengku Sembok, a dean at International Islamic University Malaysia, weighed in.
- 10:39Mei Ling: He said sovereignty is "not achieved simply by locating servers in Malaysia." He wants measurable targets and a yearly sovereignty assessment.
- 10:49Mei Ling: Renting is cheaper today, and adoption pays back sooner.
- 10:54Mei Ling: But if we only rent, our language and skills live on someone else's terms.
- 10:59Mei Ling: If we build, we may spend years on a model the market has passed.
- 11:04Mei Ling: So: fund the engine, or fund the people who use it?
- 11:08Raj: Friday, the AI-YA team takes a position.
- 11:13Aiman: Verdict time. Today's subject is ILMU's cheapest developer model, GLM 5.3 Flash.
- 11:19Aiman: SoyaCincau lists it at RM0.55 per million input tokens and RM1.85 for output.
- 11:26Aiman: For a Malay-language customer chatbot handling a thousand chats a month, my rough sum is about RM2 a month, before tax.
- 11:36Raj: Two ringgit. And the person who builds it?
- 11:41Aiman: That's the real bill, and it's unknown until you get quotes.
- 11:46Aiman: My vote is Free cukup, because a twenty-person firm should test what free chatbots already do in Malay first.
- 11:55Mei Ling: Mine is Bayar, because the model is cheap and hosted in Malaysia, but get the build quoted before you sign.
- 12:03Raj: Hype, because my customers would rather wait for a real person than chat with a robot that costs less than my teh tarik.
- 12:12Aiman: Your one thing today. Pick three real customer messages from this month, in Malay or mixed language.
- 12:19Aiman: Delete names, phone numbers and account details first.
- 12:23Aiman: Read the data terms of ILMUchat and of the foreign chatbot you already use.
- 12:28Aiman: Then paste the messages into both, and mark which replies you would actually send. It takes under an hour.
- 12:36Mei Ling: AI-YA runs AI lectures, workshops and implementation training for Malaysian companies.
- 12:43Mei Ling: Details at ai-ya.my/podcast.
- 12:46Raj: Tomorrow is our Friday special on the big questions.
- 12:49Raj: We take build-or-rent all the way: the strongest case for a national model, what it means for jobs, and countries that tried before.
- 12:59Raj: Then we hear what other podcasts say, and each of us takes a position.
- 13:04Aiman: Build or rent, where do you stand? Tell us in the comments on the episode page at podcast.ai-ya.my.
- 13:13Raj: Thank you for spending the slow stretch home with the three of us.
- 13:17Raj: Wherever you're headed tonight, arrive slowly and arrive well.
- 13:22Raj: Take care, and see you tomorrow for the big questions.