PatSnap IPO could lead Southeast Asian startups to listings in Hong Kong
The Singapore-based IP specialist is said to be lining up an HKEX listing that plays on its AI credentials
Welcome back to Asia Tech Review, your curated digest to keep up to date with tech news across Asia.
Startup exits continue to be a major deal in Southeast Asia, but a nearly 20-year-old company might help bridge the gap to Hong Kong with an IPO before the end of the year. We look at PatSnap’s listing ambitions and a tactical fundraising announcement from Sarvam, India’s leading sovereign AI contender.
To keep up with our issues, follow us on LinkedIn or WhatsApp. Please do share Asia Tech Review with people you know who want to get smart about Asia tech.
PatSnap, the low-profile contender to put Southeast Asia on Hong Kong’s IPO map
Can Southeast Asian companies find their much-needed IPO exits in Hong Kong? We are about to find out after Singapore’s PatSnap confidentially filed to go public in a dual listing across Singapore and Hong Kong, according to Bloomberg.
It’s well known that Southeast Asian startups have struggled to find the meaningful exits that venture capital ecosystems require. IPOs for Grab and GoTo, two regional flag-carriers, have massively underwhelmed and younger companies have struggled to find a path to public markets. Hong Kong, by contrast, has become a destination for tech IPOs, with AI a particular hit since MiniMax and Zhipu went at the start of this year with stellar returns for investors.
PatSnap is looking to raise between $300-$400 million at a valuation of over $2 billion, Bloomberg says, and it has some of the elements that might appeal to HKEX investors.
The 19-year-old company specialises in intellectual property, specifically it operates a database of over 120 million patents which helps companies manage their own patents and innovation, keep an eye on competitors, develop new opportunities and more. I will confess I didn’t fully comprehend its value when I wrote about it during my time at TechCrunch, but it appears increasingly important during these AI-fuelled times. AI technology allows PatSnap customers to run deeper analysis and research on the patent world, be it their own, their industries or their rivals, to help develop their own IP and innovation.
“Patent data let us kick down the door and earn respect, but now we’re looking at completely different products,” Ray Chohan, one of PatSnap’s co-founders, told me back in 2018. Even then, they were developing new products to position themselves as “the software stack for R&D teams.” No doubt AI has put that into overdrive.
The company claims over 15,000 customers. It includes HongShan, SoftBank and Tencent among its backers who gave it around $400 million in capital to date and valued the business at over $1 billion.
PatSnap can play up its position as an AI business. Its website has visible cues including a chatbot interface and agents that can scour its patent database and extract data and intelligence.
The listing is being tipped to come “as early as this year,” which suggests it could slip into 2027.
Given its previous valuation, Bloomberg’s $2 billion-plus target for the IPO isn’t particularly outrageous. Without knowing the full details, it looks like this is an opportunistic punt at Hong Kong given the company’s AI credentials and Hong Kong’s track record this year. Zhipu’s stock price has surged to give it a valuation of over $80 billion, up from around $6.6 billion when the bell rang. A relatively modest valuation isn’t necessarily a hurdle, and an exit is better than no exit, as the rhetoric goes.
We’ve seen a couple of Thai companies linked with Hong Kong IPOs, Bitkub and Line Man, but PatSnap would be the first to walk the path. I’m also curious to see what happens on the Singapore IPO side, particularly with data centre firm DayOne tipped to take a US-Singapore IPO route.
Singapore has much work to do to catch Hong Kong. A recent Financial Times article noted that IPOs on SGX increased from six in 2024 to 16 in 2025. HKEX, meanwhile, had 119 IPOs in 2025. This year already, it has hosted two blockbuster LLM listings and a host of AI companies across chip design, production and more.
Sarvam pops out a timely funding announcement but many questions remain
Timing is everything. A day after we wrote about the need for sovereign AI solutions in Asia, India’s Sarvam raised $234 million at a $1.5 billion valuation.
This round has been rumoured for a few months (I even mentioned yesterday it still hasn’t been confirmed) and it looks like this announcement is fairly opportunistic since the target total raise is $300 million. Clearly now is a good time to be a bit more public about fundraising for a homegrown AI model for India, and of course trumpet that it is now a unicorn business. (Does that still have meaning these days…?)
That pitch convinced IT services giant HCLTech, which led the round with a $150 million cheque and is joined by Bessemer Venture Partners, Khosla Ventures and Peak XV Partners. Rather interestingly, Bloomberg noted in previous reporting that Sarvam wasn’t able to convince major investors SoftBank, Prosus, a16z, General Atlantic and Accel to join the party.
Sarvam disclosed some numbers including that it handles 2 million conversations and 10 million API calls per day, with 500,000 hours of audio transcribed per month. It says a leading fintech uses its model for its 350,000-person sales force and claims it was adopted by a leading insurance provider.
Everyone wants to fund a local alternative to Claude or ChatGPT, but the dynamics at play here are insanely tough. The company hasn’t defined its user focus yet. That’s important as we’ve seen from Anthropic doubling down on enterprise in its battle with ChatGPT.
Sarvam has a consumer AI chatbot, covering the basics, but will it focus on covering all Indian businesses? Is this a play for enterprise because it is more lucrative? Can it get the government to mandate adoption? Probably not.
Then there’s the issue of quality. Sarvam covers more than 22 Indian languages, but Claude and OpenAI are better products than anything else on the market. That includes Chinese models that have taken billions of dollars to develop. It feels unlikely Sarvam can compete on merit in the near future.
Then that returns us to the sovereign and patriotic argument, which has only really played out in China. And still we read weekly of how Tencent, Alibaba, ByteDance and other big tech Chinese firms continue to buy Nvidia and use US models overseas.
Asia needs more sovereign AI solutions, at the risk of repeating yesterday’s newsletter. But it’s not clear why investors should be a part of that, too.
Deals
Tencent has reportedly invested $20 million in the new AI lab founded by Junyang Lin, the former lead researcher behind Alibaba’s Qwen models [The Information]
Singapore’s Pints AI, which helps banks and insurers automate processes whilst remaining audited and compliant, raised $5.6 million in a pre-Series A round led by Tin Men Capital. [TechNode]
Kazakhstan wants to turn cheap and available power into AI infrastructure exports, and plans to spend up to $10 billion on computing projects, including $1 billion from a state-run telco, to develop data centres with Nvidia tech support [Bloomberg]
Markets
Zhipu shares soared further after it revealed plans to make GLM-5.2, its latest and most powerful large language model, open source [South China Morning Post]
Xiaohongshu, the parent of RedNote, reportedly plans to file for a Hong Kong IPO this month. Its last valuation was $17 billion in a 2024 funding round, but shares are said to have exchanged hands in secondary markets for as much as $31 billion. [Bloomberg]
Enflame Technology, an AI chipmaker that Tencent owns 20% of, will go public on Shanghai’s STAR board with plans to raise 6 billion yuan ($888 million) at a valuation of around $9 billion-plus. [Bloomberg]
In other news:
ByteDance is said to be in talks to expand its Chinese AI inference chip partnerships with deals from Iluvatar CoreX and potentially Baidu’s Kunlunxin on the table. It works with Huawei and Cambricon. [Reuters]
Tiezhen Wang, a former Hugging Face executive, believes China’s open-source AI push is mounting a serious challenge to OpenAI’s dominance, with freely available models increasingly matching proprietary ones on performance. [Rest of World]
Google and Black Lotus Labs worked with the FBI to dismantle a massive Chinese phishing-as-a-service that used thousands of phishing websites to steal credit card data and passwords [Bleeping Computer]
Google also revealed that a Chinese-linked hacking group spent over two years quietly stealing data from US and Canadian academic, medical, and military research institutions between September 2023 and November 2025. [Reuters]
Investors and analysts say China’s humanoid robotics industry risks repeating the electric vehicle sector’s brutal price wars [Caixing Global]