Sea Ltd Becomes First in Southeast Asia to Deploy NVIDIA Vera Rubin: An AI Turning Point

Have you ever wondered what is really happening behind the "smart" product recommendations on Shopee, the smooth matches in Garena, or the surprisingly fast loan approvals from Monee? The answer lies in something few people see: AI computing infrastructure.

And recently, Sea Limited—the parent company of Shopee, Garena, and Monee—made an announcement that caught the attention of the entire Southeast Asian region: they will become the first business in Southeast Asia to deploy NVIDIA Vera Rubin, the newest generation AI computing platform succeeding the legendary Blackwell series.

This is not just a hardware procurement deal. It is a strategic declaration: Sea Limited is stating that they do not just want to use AI, but want to become a true "AI-native" company—where AI is the backbone and DNA of every business operation.

What is NVIDIA Vera Rubin and why is it so important?

To understand the significance of this deal, we need to understand what Vera Rubin is.

This is NVIDIA's next-generation AI platform, specifically designed for the era of agentic AI—where models not only answer questions but also autonomously execute complex tasks. Vera Rubin NVL72, the most powerful configuration, integrates 72 Rubin GPUs, 36 Vera CPUs, along with the new generation NVLink 6 interconnect system, creating unprecedented computing power in a single rack.

The numbers are truly staggering:

• NVFP4 inference performance up to 3,600 PFLOPS—a figure almost unimaginable just a few years ago.
• 20.7 TB of HBM4 memory across the system, with massive bandwidth.
• 10x energy efficiency compared to Blackwell per unit of energy when processing agentic tasks.
• Token cost reduced to 1/10th compared to the previous generation, while the number of GPUs needed to train MoE models is only 1/4th.
• 10x agent throughput compared to the previous Grace Blackwell platform.

What does this mean for Sea Limited? They will be able to train and deploy more complex AI models, at a lower cost, on a larger scale. That is a competitive advantage that money cannot buy—because not everyone can access the scarce Vera Rubin supply in the early stages.

How has Sea Limited been using AI?

Sea Limited is no rookie in the field of AI. In fact, they have quietly built their AI capabilities over the past many years, and what they are doing is genuinely impressive.

Garena: 1,300+ in-game AI chatbots and agents

At Garena, Sea Limited's gaming unit, teams have built over 1,300 AI chatbots, workflows, and agents, along with "in-game companions." These are not simple chatbots that answer questions. These are AI systems capable of interacting, supporting players, and even participating in the gaming experience in unprecedented ways.

Shopee: AI writes product descriptions, boosting conversion rates by 14%

At Shopee, AI tools are helping sellers write product descriptions and reach buyers more effectively. The results are striking: shopping conversion rates increased by 14% year-over-year in the first quarter.

This means AI not only saves time but also directly generates revenue. Every percentage point increase in conversion rate translates to millions of dollars in additional revenue on Southeast Asia's largest e-commerce platform.

Monee: AI fraud detection, increasing loan approval rates by 10%

At Monee—Sea Limited's financial services arm—AI is used for fraud detection and credit scoring. The latest credit risk models, built on transformer systems and trained on cross-platform activity data, have helped increase loan approval rates by about 10% while maintaining the same risk level.

Think about that: AI is helping Monee serve more customers, expanding credit access to those who were previously turned down, without increasing risk. This is a perfect example of how AI can expand financial opportunities for the community.

Compass Max v3.5: Sea Limited's proprietary AI "Brain"

But there is a detail even more important than Sea Limited buying NVIDIA hardware: they have built their own large language model (LLM).

That model is named Compass Max v3.5—a system with 245 billion parameters, specifically designed for Southeast Asian languages and e-commerce applications. This is a crucial point: Compass Max v3.5 is not a general-purpose model "repurposed" from global LLMs. It was built from scratch for the Southeast Asian context, understanding regional linguistic nuances, cultures, and user behaviors.

And how well is it performing? According to Sea Limited, Compass Max v3.5 and its variants have been used directly in numerous AI features on Shopee, operating with higher responsiveness and at a fraction of the cost of other leading commercial LLMs. Compass's monthly processing volume grew from 3 billion to 340 billion tokens in just 8 months, and currently accounts for the majority of Shopee's AI traffic. The system also helps detect fraud 50 times more effectively than manual checks while cutting processing costs by 90%.

This is Sea Limited's true competitive advantage. While many other companies rely entirely on APIs from OpenAI, Google, or Anthropic, Sea Limited has built in-house AI capabilities with lower costs and higher customizability.

AI Centre of Excellence in Singapore: Long-term commitment

The Vera Rubin deal is not an isolated action. It is part of a much more methodical strategy.

Sea Limited has established an AI Centre of Excellence in Singapore, with the goal of driving AI-native innovation and building internal capabilities. This center focuses on three main areas: developing foundational models and internal tools, translating research into production-ready solutions at scale, and nurturing AI-native talent alongside new AI-supported operating models. Over the next three years, the center is expected to generate demand for at least 100 research and innovation positions in Singapore.

Sea Limited has also partnered with some of the biggest names in the global AI industry:

• Google—partnering to develop AI applications for Garena, Shopee, and Monee, including trials of agentic shopping on Shopee.
• OpenAI—partnering to expand AI access for Shopee users and sellers, as well as developing AI-powered commerce solutions.

Partnering with both Google and OpenAI—direct competitors of each other—shows that Sea Limited is pursuing a partner diversification strategy, not relying on any single AI platform. That is the mindset of a company that truly wants to master technology, not just consume it.

Regional Context: The AI race in Southeast Asia is heating up

Sea Limited's move does not happen in a vacuum. It is part of a wave of AI investment spreading across Southeast Asia.

Malaysia has witnessed massive investments in AI infrastructure. Projects like the AI data center in Johor with investment commitments of up to $6.73 billion will deploy NVIDIA Blackwell, Vera Rubin, and Rubin Ultra across multiple phases. Another project in Indonesia plans to build an AI facility with 170,000 GPUs, utilizing Vera-Rubin, Grace-Blackwell, and Vera platform accelerators from 2027 to 2028.

In this context, Sea Limited becoming the first business in Southeast Asia to deploy Vera Rubin is not just a title. It is an ahead-of-the-competition advantage in AI computing infrastructure—an advantage that may take years to close.

Why does this story matter to you?

You might be thinking: "I am a consumer, why should I care about Sea Limited's AI infrastructure?"

Well, it is very relevant.

First, AI is changing your shopping experience. When you browse Shopee and see product recommendations that feel eerily "tailored to your taste," that is AI at work. When you receive quick answers from support chatbots, that is AI. When you check out using automatically suggested payment methods, that is also AI. Vera Rubin will make these experiences faster, smarter, and more personalized.

Second, AI is changing financial opportunities. When Monee uses AI for credit scoring, people who were previously denied loans due to lacking traditional credit history can be served more fairly. This opens up opportunities for millions of Southeast Asians previously excluded from formal financial services.

Third, AI is changing the gaming market. When Garena uses AI to create "in-game companions," your gaming experience becomes richer and more personalized. You are no longer playing alone—you are playing alongside an AI that understands you.

And most importantly: the AI race in Southeast Asia will shape the economic future of the entire region. Companies and countries investing in AI infrastructure today will be the leaders of tomorrow's digital economy. Sea Limited is placing a huge bet on that future—and they are doing it ahead of all regional competitors.

When "AI-native" is no longer just a slogan

I have witnessed many companies declare themselves "AI-first" or "AI-driven" over the past few years. Most stopped at integrating a few APIs or hiring a few experts. But Sea Limited is doing something different: they build AI capabilities from the root—from proprietary language models (Compass Max v3.5) and centers of excellence to next-generation computing infrastructure (Vera Rubin).

That is a long-term commitment, not a marketing campaign. And in a region experiencing a booming wave of AI investment like Southeast Asia, Sea Limited being one step ahead can create a sustainable competitive advantage for years to come.

Yet the bigger question remains open: Will other businesses and countries in the region keep up? And will Vietnam—with significant AI potential but infrastructure challenges—find its footing in this race?

What do you think about a Southeast Asian enterprise like Sea Limited pioneering the deployment of next-gen AI infrastructure? Is it a bold move opening a new era for the region, or an expensive gamble while AI applications still hold many unknowns? If you are a Vietnamese business looking to apply AI, what will you learn from Sea Limited's strategy—and where will you start? Share your views below, because in this AI race, those who dare not place bets will forever remain spectators!