Nvidia (NVDA.US) in Talks to Acquire South Korea’s “K-Nvidia Project Main Player” Rebellions, Jensen Huang’s AI Chip Empire Expands Further

Zhitong Finance APP reports that, in the battle to consolidate supremacy in AI chips, Jensen Huang is turning his attention to South Korea. According to reports, Nvidia (NVDA.US) has entered preliminary talks with South Korean AI chip design company Rebellions regarding potential cooperation, with options including technology partnership, investment, and even a full acquisition. This week, Nvidia CEO Jensen Huang personally met with Rebellions' co-founder and CEO Sunghyun Park at Nvidia’s Santa Clara headquarters in California, indicating the high strategic priority of this potential deal for Nvidia.

Currently, negotiations are still at an early stage and may ultimately fall through. However, if an agreement is reached, it would mark another major move by Nvidia in the South Korean AI ecosystem after investing $1 billion in Naver in July. For Rebellions, which is on the eve of its IPO, Jensen Huang's olive branch could completely change its narrative as a challenger to Nvidia.

Rebellions: South Korea’s “K-Nvidia” Built by the Whole Nation

Established in 2020, Rebellions is a core focus of the South Korean government’s “K-Nvidia” program. In March this year, the Korean National Growth Fund directly invested 250 billion won (about $166 million) in Rebellions through the “Advanced Strategic Industry Fund”—the first direct investment under this program.

So far, Rebellions has raised about $850 million, with investors including semiconductor giants SK hynix, Samsung Ventures, and Arm Holdings. Its latest valuation is around $2.3 billion. Samsung and SK hynix both hold shares in Rebellions, giving it unique advantages in securing critical memory chip resources.

On the product front, Rebellions focuses on AI inference chips—which, compared to training, handle much larger workloads. Its flagship product, Rebel100 NPU, is manufactured with Samsung’s 4-nanometer process, uses chiplet architecture, is equipped with 144GB HBM3E memory, and delivers around 2 PFLOPS of FP8 compute power. The company has launched the RebelCard accelerator card based on the Rebel100 and plans to IPO in South Korea in the first half of 2027, with future ambitions to list in the US.

Nvidia’s “Talent + Technology” Hunt: From Groq to Rebellions

This potential deal continues Nvidia’s recent strategic approach of acquiring AI talent through unconventional means. At the end of 2025, Nvidia reached a non-exclusive licensing agreement with AI chip startup Groq—gaining access to Groq’s inference technology while absorbing most of its engineering talent, but Groq retained its independence. This arrangement has been widely interpreted as a “backdoor acquisition” to avoid antitrust scrutiny, triggering investigations by US Senator Elizabeth Warren and others. The engagement with Rebellions is the latest sign of Nvidia’s ongoing “light acquisition, heavy integration” strategy.

Nvidia dominates the AI training chip market, and any large-scale acquisition would be closely watched by the US Justice Department for antitrust risk. The Groq model—technology licensing plus talent absorption—may serve as a template for Nvidia to acquire innovative technology under regulatory pressure.

Rebellions’ expertise lies in AI inference—handling the large-scale workloads required to run trained models. As AI moves from the training phase to large-scale application deployment, the inference market will far surpass the training market in size. While Nvidia is overwhelmingly dominant in training chips, the competitive landscape for inference chips is still forming, and securing a strategic position in promising inference chip companies aligns with Nvidia’s long-term strategy.

For Rebellions, being “partially absorbed” by Nvidia is not an impossible outcome. In the inference chip sector, Groq has already been “technically locked” by Nvidia and SambaNova acquired by Intel. For inference chip startups unable to go public, exit options are narrowing.

The South Korean government is fully advancing the “K-Nvidia” plan, with SK hynix and Samsung Electronics—two giants in memory—both shareholders of Rebellions. By forming an alliance with Rebellions in some form, Nvidia could further embed itself within South Korea’s AI supply chain.

Regulatory “High Voltage”: Facing Dual US-Korea Scrutiny

Any deal above a certain scale may face strict regulatory scrutiny. Nvidia’s dominance in the AI training chip market puts every move under antitrust radar. US senators have already launched an investigation into Nvidia’s deal with Groq. If a deal with Rebellions is characterized as a “backdoor acquisition” aimed at eliminating a potential competitor, it could face strong opposition from the US Justice Department.

On the Korean side, semiconductors are considered vital strategic assets. As the centerpiece of the “K-Nvidia” strategy, any sale of control of Rebellions to foreign interests could trigger rigorous review by the Korean government. As shareholders and pillars of the Korean memory chip industry, Samsung Electronics and SK hynix's attitudes will also significantly influence the outcome of the deal.

Strategic Significance: Jensen Huang’s “South Korean Puzzle”

If Nvidia reaches a deal with Rebellions, it will form a “tripod” layout in South Korea’s AI ecosystem: a $1 billion investment in South Korean internet giant Naver in July, making Nvidia its third-largest shareholder; deep supply chain partnerships with memory powerhouses Samsung and SK hynix; and now, aiming at the government-backed “K-Nvidia” Rebellions.

The strategic logic is clear: upgrade South Korea from a “memory chip base” to a strategic foothold for Nvidia’s AI empire—securing HBM and other critical memory resources, absorbing top-tier AI chip design talent, and embedding deeply in South Korea’s government-led AI sovereignty drive.

Rebellions’ focus is inference chips—a relatively weaker area for Nvidia. As AI moves from training to mass inference deployment, the inference chip market is exploding. By acquiring Rebellions’ NPU technology and engineering team, Nvidia could build a more comprehensive competitive moat in the inference space.

Negotiations are still early and the transaction may not happen. But Jensen Huang’s personal engagement already demonstrates Nvidia's high regard for this potential deal. As the AI chip competition enters a new stage of “buying talent, buying technology, and building ecosystems,” Rebellions is becoming yet another target on Jensen Huang’s global hunting list.