Jensen Huang: How to Spend $60 Billion? A Complete Record of Buying from Groq to Intel

NVIDIA’s market capitalization approaches 5 trillion USD, with Jensen Huang holding 60.6 billion USD in cash to aggressively expand the AI ecosystem. In December, they reached a 20 billion USD agreement to acquire the Groq team, committed to investing 100 billion USD in OpenAI, invested 5 billion USD in Intel to turn rivals into allies, invested 2 billion USD in Synopsys, and 1 billion USD in Nokia, building a technological empire.

NVIDIA’s 60.6 Billion USD Cash as a Strategic Weapon

By the end of October 2025, NVIDIA’s cash and short-term investments reserve up to 60.6 billion USD, 4.5 times the 1.33 billion USD at the beginning of 2023. Analysts predict that just in 2025, its free cash flow will reach 96.85 billion USD, and over the next three years, it could total more than 576 billion USD. This “excess cash” dilemma, Huang Renxun provides a clear answer: large-scale strategic investments are the best capital allocation.

Although the company spent 37 billion USD on stock buybacks in the first three quarters of this year, Huang Renxun explicitly prioritizes strategic investments higher. He believes that a strong balance sheet gives confidence to customers and suppliers, and that investing in the ecosystem is “very important,” directly driving additional consumption of AI and NVIDIA chips. This massive cash reserve is being transformed into an insurmountable competitive barrier.

NVIDIA’s capital strategy is no longer just financial investment but uses “prepayments” or “investment in exchange for commitments” to lock in core customers, bind key partners, and preemptively “recruit” potential competitors. In today’s fierce AI competition, Huang Renxun’s “cash dilemma” is precisely his most powerful weapon. This strategic move to build a “shield of energy” for the technological empire through capital is reshaping the entire AI industry landscape.

Eliminating Threats: $20 Billion “Drain” Acquisition of Groq

On December 24, NVIDIA reached a roughly 20 billion USD technology licensing agreement with Groq. On the surface, Groq maintains independent operations at the corporate level, but in reality, Groq’s co-founder, CEO, and core technical executives will be integrated into NVIDIA, and their key technological capabilities will be incorporated into NVIDIA’s system. This is more like a direct invitation from Huang Renxun: “Your technology is quite good. With 20 billion USD, bring your team and architecture, and come work for NVIDIA.”

This is not an ordinary acquisition but a precise strategic defense and capability complement. Groq’s killer feature is its LPU (Language Processing Unit) architecture, which stores model weights in SRAM instead of traditional HBM, achieving extreme inference speeds, sometimes even 10 times faster than GPUs. This directly threatens NVIDIA’s “latency control advantage” in AI inference markets. Bestselling author Mark Beckmann believes inference is the key to the next decade’s scale development.

Through this deal, Huang Renxun not only eliminates a challenger with a “truly alternative architecture” but also turns the opponent’s disruptive innovation into fuel for his own acceleration. So, where will the “drained” Groq go? The deal shows that its cloud service business, GroqCloud, has been spun off. Skeptical netizens predict that without the core team and chip roadmap support, GroqCloud, like a “lamb waiting to be slaughtered,” may face low-priced acquisition or gradual marginalization.

This “drain” or “talent acquisition” model, while avoiding strict antitrust scrutiny, achieves locking key technologies and talent. This licensing agreement is similar to deals between Meta and Scale AI, Google and Windsurf. These cases point to a clear trend: when a capability proves irreplaceable, cooperation is no longer the end; internalization becomes the ultimate choice.

Building the Ecosystem Great Wall: From Chip Design to 6G Networks

Huang Renxun’s investment logic is clear: convert excess capital into “flexible control” over every critical node in the AI value chain. Investing 2 billion USD in Synopsys, a leader in semiconductor design software, directly embeds NVIDIA’s accelerated computing power into the design tools of all future chips. This means that from smartphones to autonomous vehicles, the design cycles of various chips will be shortened due to NVIDIA’s technology.

More cleverly, turning enemies into allies. Huang Renxun extended a 5 billion USD olive branch to Intel, a traditional rival in processors. This investment not only yields financial returns but also deep strategic and technological alliances. Intel will develop customized x86 CPUs for NVIDIA data centers, and Intel will integrate NVIDIA’s GPU cores into next-generation PC chips, opening a vast consumer market channel for NVIDIA.

Huang Renxun’s 2025 Investment Map Key Deals

Model Layer Binding: Commit to investing 100 billion USD in OpenAI, invest 10 billion USD in Anthropic in exchange for a 30 billion USD procurement commitment

Inference Layer Threat Elimination: Acquire Groq’s core team and LPU technology for 20 billion USD, internalize inference architecture innovation

Infrastructure Layer Control: Invest 2 billion USD in Synopsys, 5 billion USD in Intel, 1 billion USD in Nokia to lay out 5G/6G networks

Application Layer Pre-Deployment: Invest in autonomous driving Wayve, humanoid robot Figure AI, nuclear fusion Commonwealth Fusion

Huang Renxun’s vision is not limited to data centers. As AI demands for low latency and high bandwidth networks explode, communication infrastructure becomes a new battleground. Investing 1 billion USD in Nokia, jointly targeting AI-native 5G and future 6G networks. These deals echo each other and collectively aim for one goal: to make NVIDIA’s technology the ubiquitous underlying pulse driving chip design, personal computers, communication networks, and ultimate artificial intelligence.