Nvidia just spent $20 billion to NOT acquire a competitor.

On Christmas Eve 2025, Nvidia executed its largest transaction to date: a $20 billion deal with AI chip challenger Groq.

On paper, Groq says it will "continue as an independent company." In reality, Nvidia just absorbed the founder, the president, and 90% of the engineering team.

The market is celebrating "innovation-friendly partnerships." I’m calling it something else: The Death of the Merger.

Nvidia didn’t buy Groq; they executed a "License-and-Lift."

By structuring a full acquisition as a "non-exclusive licensing agreement," Nvidia achieved total market neutralization while bypassing the 18-month antitrust gauntlet.

This is the 2025 playbook in action - A multi-billion dollar sleight of hand designed to keep "the fiction of competition" alive for regulators.

Read on for: The 5 signals proving this is systematic, why it works, what regulators are (finally) doing about it, and where this strategy breaks down.

Signal #1: The Nvidia-Groq Deal Structure

Nvidia paid $20 billion for Groq's assets.
Groq raised $750 million at a $6.9 billion valuation three months ago. That's a 3x premium.
Groq CEO Jonathan Ross and President Sunny Madra joined Nvidia. Most senior leaders followed. Groq's GroqCloud continues "as an independent company" led by the CFO.

What this means: Every indicator of an acquisition is present except the legal classification.

Signal #2: The Pattern Is Systematic - $40B+ Since 2023

Big Tech deployed over $40 billion in "license-and-lift" deals from 2023-2025:

What this means: This is a repeatable playbook, not an isolated incident.

Signal #3: It Bypasses Hart-Scott-Rodino Review

The Hart-Scott-Rodino (HSR) Act requires pre-merger notification for deals above $133 million. Filing triggers:

Licensing deals with separate talent hires would mean no HSR filing required.
No merger classification, no antitrust review.

What this means: The structure is designed explicitly to avoid regulatory scrutiny.

Signal #4: Analysts Are Calling It What It Is

Bernstein analyst Stacy Rasgon wrote: "Structuring the deal as a non-exclusive license may keep the fiction of competition alive."

Hedgeye Risk Management analysts: "This transaction is essentially an acquisition of Groq without being labeled one (to avoid the regulators' scrutiny)."

What this means: Market observers recognize this as regulatory arbitrage, not legitimate alternative deal structure.

Signal #5: Regulators Are Starting To Catch On

The DOJ is investigating Nvidia's Groq deal as a "regulatory bypass." The inquiry could result in "behavioral remedies" preventing Nvidia from prioritizing investment partners for chip access.

An FTC 6(b) report in early 2025 concluded that massive investments by Microsoft, Amazon, and Google into OpenAI and Anthropic created "lock-in" effects forcing startups to spend capital back into investor cloud infrastructure.

What this means: The regulatory window is closing, but slowly enough that Big Tech is executing dozens of these deals before enforcement arrives.

Despite the "independence" narrative, look at the asset migration:

Big Tech has discovered that time is more valuable than equity.

If Nvidia had attempted a traditional merger, the FTC would have blocked it to "save" the inference market.
By the time the DOJ finishes its current investigation in 2027, the Groq team will have already integrated their tech into Nvidia’s "AI Factory" racks.

Why this matters for you:
The exit landscape for AI startups has fundamentally shifted.
The "Goldilocks" exit is no longer an IPO or a clean M&A. It’s an Elimination Premium.

Nvidia isn't just buying chips; they are buying immunity from the next decade of competition.

If you're a founder:

1. Understand "partnership offers" from Big Tech are often elimination strategies

If you're building technology that threatens an incumbent and they approach with a "strategic partnership" that includes:

What to do: Recognize this is acquisition-without-acquisition. Your company will become a shell. Your team will be absorbed. Your technology will be controlled. If that's your desired exit, negotiate accordingly. If you want to remain independent, decline. There's no middle ground despite what the term sheet says.

2. If you're Series A-B, build with acquisition defense in mind

The license-and-lift structure works because startups at Series A-B have concentrated talent (small team means easy to absorb) and early-stage IP (easy to license).

What to do: Build distributed technical capabilities across larger teams. Make your company harder to "license-and-lift" by ensuring no single person or small group holds all critical knowledge. Document IP ownership clearly to make licensing require actual asset transfer (harder to structure as "not an acquisition").

3. Consider regulatory exposure as a competitive moat

If your technology is threatening enough that acquiring you would trigger regulatory review, that's actually a moat. Big Tech can't acquire you traditionally, and license-and-lift becomes riskier as regulatory scrutiny increases.

What to do: Stay independent longer. Go public earlier (hard to "license-and-lift" a public company).
Partner with multiple incumbents (makes any single acquisition or elimination harder).
Force Big Tech to compete for your technology rather than absorb it.

If you're a VC:

1. Model exit scenarios assuming acquisitions are structured as licensing deals

Traditional M&A multiples don't apply when Big Tech uses license-and-lift.
Groq got 3x in 90 days because Nvidia eliminated a competitive threat.

What to do: Underwrite investments assuming exit structure matters as much as exit valuation.
Companies that can credibly threaten incumbents get elimination premiums via licensing deals.
Companies that don't threaten incumbents get traditional (lower) M&A multiples, if they get acquired at all.

2. Advise portfolio companies on licensing-as-exit structuring

This playbook benefits founders (premium exits) and investors (fast liquidity, no regulatory delay risk).

What to do: Educate portfolio CEOs on how to structure offers from Big Tech. Push for "all or nothing" deals - either genuine partnership OR full licensing + team transfer at premium.
The middle ground (licensing without premium) leaves you with nothing.

3. Prepare for regulatory backlash that kills late-stage valuations

If DOJ succeeds in Groq investigation, license-and-lift becomes riskier. Late-stage AI companies that positioned for Big Tech elimination exits could see valuations crash.

What to do: Diversify exits. Push portfolio companies toward IPO or international buyers (especially if US regulatory scrutiny increases). Don't underwrite late-stage rounds assuming Big Tech will pay elimination premiums if the structure becomes legally risky.

If you're building in AI infrastructure:

1. Nvidia is systematically eliminating inference competitors

Groq (inference chips): $20B.
Enfabrica (networking to connect 100K+ GPUs): $900M.
Pattern: Nvidia dominates training, now absorbing inference layer.

What to do: If you're building AI chips/infrastructure, assume Nvidia will attempt to absorb or eliminate you once you prove traction. Build regulatory defense (partner with multiple cloud providers, international distribution) or plan for licensing exit early.

2. The "independent cloud" narrative is dying

Meta absorbed Scale AI ($14.3B).
Microsoft absorbed Inflection.
Google absorbed Character.AI and Windsurf.
Every major AI infrastructure layer is being absorbed by incumbents.

What to do: If you're building middleware (APIs, data platforms, model serving), assume your exit is licensing to an incumbent, not independent scale. Price your product accordingly (premium while independent because time is limited).

3. Build for international markets if you want to stay independent

US regulatory framework enables license-and-lift. EU and UK are scrutinizing these deals more aggressively but still slow.

What to do: Build customer base outside US first. Makes you less attractive for US Big Tech absorption (they want US market control) and gives you leverage if they approach.

My prediction:

Bold summary: License-and-lift isn't a loophole. It's the new M&A standard.
By the time regulators adapt, Big Tech will have absorbed every threatening competitor.

Traditional acquisition is dead. Regulatory arbitrage is permanent.

Have you been approached with a "strategic partnership" that felt like acquisition without the label?

Reply with your take:

This week’s sources:
→ CNBC: Nvidia buying AI chip startup Groq for about $20 billion in biggest deal ever
→ CNBC: Nvidia-Groq deal is structured to keep 'fiction of competition alive'
→ CNBC: Nvidia spent over $900 million on Enfabrica CEO, AI startup technology
→ CNBC: Scale AI founder Wang announces exit for Meta, part of $14 billion deal
→ FourWeekMBA: Big Tech's $40B License-and-Lift Strategy
→ Yahoo Finance: Nvidia's Groq deal underscores how the AI chip giant uses its massive balance sheet
→ Fast Company: What is the reverse acquihire and why is it so controversial?
→ Columbia STLR: FTC Scrutiny of Licensing and Hiring Arrangements
→ Market Minute: The Quasi-Merger Crackdown—DOJ investigating Nvidia-Groq
→ FourWeekMBA: License-and-Lift Becomes the Default M&A Template

See you next week,
Pavan

P.S. Nvidia spent $20 billion to eliminate a competitor without filing a single regulatory form. If that's not proof that antitrust enforcement is broken, I don't know what is.

P.P.S. Groq was valued at $6.9B on September 4, 2025. Nvidia paid $20B on December 24, 2025. That's a 190% premium in 111 days. The cost of eliminating competition has never been higher—or easier to execute.