What December Quietly Confirmed: Why the Biggest Players Are Pulling Further Ahead

Key deals of the week (December 22 - December 26 2025)

Nvidia and Groq: Buying the stack without buying the company

Nvidia made its position clear in the next phase of the AI companies race. Instead of buying Groq outright, it struck a non-exclusive licensing deal for Groq’s inference chip technology and hired away its founder and key executives, including Jonathan Ross, who previously helped build Google’s AI chip efforts. Groq will remain an independent company, keep its cloud business running, and install a new CEO, but its core technical leadership and a slice of its IP now inside Nvidia. The move targets inference, the part of the AI stack where models actually respond to users and where Nvidia faces its most credible competition from AMD and specialized startups like Groq and Cerebras. Financial terms weren’t disclosed, despite reports floating a $20 billion price, and the structure mirrors a growing Big Tech pattern: absorb talent and critical technology without triggering a full acquisition, regulatory fight, or balance-sheet hit, while tightening control over the most contested layer of the AI infrastructure stack.

PayPal and SpaceX: When extreme operators start exporting companies

PayPal created operators who went on to define consumer internet and fintech; SpaceX is doing the same for hard tech, defense and deep infrastructure. When a company is valued near $800 billion, runs at extreme technical velocity, and trains people inside systems where failure is existential and iteration is brutal, the alumni don’t leave empty-handed. They leave with credibility, networks, and a playbook that VCs recognize instantly. That’s why former SpaceX engineers can raise real money early, why their startups cluster around capital-intensive, non-obvious problems, and why the “SpaceX Mafia” already looks less like a narrative and more like a pipeline.

Trian x General Catalyst and Janus Henderson: Taking Asset Management Private 

Trian and General Catalyst are taking Janus Henderson private at $7.4B, paying $49 a share and closing the loop on an October approach that started at $46. Structure here matters more: an activist fund pairing with a Silicon Valley VC best known for betting on AI-driven transformation, backed by sovereign and Asian capital. Janus keeps its leadership, London – Denver footprint and its $483.8B AUM intact. Public markets have become a constraint for asset managers, and private ownership offers room to retool in technology, talent, and operating model. 

Lightspeed: When fund structure matters as much as fund size

Lightspeed dropped a $9B fundraise, and it lands like a punctuation mark on a year defined by capital concentration. While mid-tier firms grind through slower closes and smaller checks, a handful of platforms are pulling capital at almost sovereign scale. Per PitchBook: spread across six vehicles, Lightspeed’s raise alone and alongside Dragoneer’s $4.3 billion crossover fund, accounts for nearly a third of all VC dollars raised this year. The structure matters as much as the size: bespoke co-invest and single-LP vehicles designed to give large allocators deeper exposure to concentrated AI bets. Managing partners Bejul Somaia and Ravi Mhatre, alongside Michael Romano travelled Australia, Japan, Korea, Scandinavia, Mexico and tapped LPs who were looking for proximity to the few AI outcomes that could actually move the needle. In today’s market, capital isn’t just consolidating, but it’s becoming more opinionated.

In 2025 just 1,117 venture funds have closed globally, almost half of last year’s count, while the top ten alone absorbed 43% of all capital raised. LP’s are clustering around a narrow band of outcomes, driven largely by AI, where conviction has collapsed into consensus. The belief is simple: a small set of hyper-scale AI companies are the only shots left at fund-defining returns, even if they’re burning capital faster than anything venture has seen before. 

Allocate CEO Samir Kaji put it: “Now, the focus from LPs has been getting into these ‘white truffle’ assets.” “That has now increased the attention toward these megafunds, who have access to these ‘consensus’ hyper-scale AI companies.” 

Across chips, space, asset management, and venture itself, the pattern is the same: control is consolidating upstream. Talent, capital, and conviction are compressing into fewer hands, fewer firms, and fewer bets. The market is no longer optimizing for breadth. It’s optimizing for inevitability.