/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

The Groq deal secures key talent for Nvidia, including CEO Jonathan Ross, creator of the TPU, and keeps them from companies that may try to make their own chips

Twas the night before Christmas and all through the house, not a creature was stirring, not even a... wait.  What's that?

Spyglass M.G. Siegler

Context & Ripple Effects

Nvidia’s arrangement with Groq was previously characterized as a non-exclusive technology license under which GroqCloud would continue operating while Ross and other executives moved to Nvidia. That licensing-and-executive transition makes this more than a conventional acquisition narrative: Nvidia gains people and inference know-how without fully absorbing the company.

The strategic value is concentrated in Ross’s chip-design experience and in denying that expertise to would-be custom-chip builders. It also leaves Groq as an independent operating presence, creating a split between control of key talent and continuity for its existing platform.

First-order effects

  • Nvidia adds Jonathan Ross and other Groq leaders, strengthening its internal capability around inference-oriented chip design and software architecture.
  • Groq loses senior operating talent while retaining its independent business and cloud service under the stated licensing structure.

Second-order effects

  • Companies considering in-house AI silicon lose a potential source of experienced leadership, raising the value of the remaining specialized hardware talent pool.
  • Nvidia can evaluate Groq’s inference technology inside its own product planning while Groq’s continued operation preserves a separate route to market for that technology.

Third-order effects

  • If similar arrangements proliferate, AI-chip competition may be shaped as much by targeted talent-and-IP deals as by outright acquisitions, allowing incumbents to neutralize emerging capabilities without consolidating every operating asset.
  • The pattern favors firms able to pair proprietary compute with software and elite design teams; independent chip startups may increasingly need to separate their commercial platforms from the mobility of their founders and technical leaders.

The trend: AI infrastructure leaders are using selective licensing and talent moves to deepen their integrated stacks while constraining the pool of teams capable of building alternative accelerators.

Discussion

  • @anjneymidha Anjney Midha on x
    for the groq deal to make sense, you must understand how frontier model workloads are going to look 12 months from now hint: omni, long horizon rl
  • @jukan05 Jukan on x
    BofA's Vivek envisions a setup where NVIDIA GPUs and Groq's LPUs are interconnected via NVLink and used together within a single rack. [image]
  • @zephyr_z9 @zephyr_z9 on x
    Anyone who thinks that the Nvidia-Groq deal was about solving CoWoS, energy, or HBM constraints is plainly wrong and doesn't understand the current paradigm of inference Groq deal creates another edge for Nvidia (it's not a magical game changer) Feynman and beyond may have [image…
  • @dylan522p Dylan Patel on x
    Damn even Nvidia is tax loss harvesting Tis the season I guess
  • @levelsio @levelsio on x
    Why @nvidia acquihiring @Groq for $20B is so interesting: It looks like a direct (and fast!) reaction by Nvidia to Google successfully using its own TPU chips for training AI models and inference (generating AI content) Because until recently, they used mostly Nvidia's GPUs...the…
  • @0xdevshah Dev Shah on x
    Nvidia paid 3X Groq's September valuation to acquire it. This is strategically nuclear. Every AI lab was GPU dependent, creating massive concentration risk. Google broke free with TPUs for internal use, proving the “Nvidia or nothing” narrative was false. This didn't just [image]
  • @draecomino James Wang on x
    So many bad takes on Groq as if its LPU is some magical new architecture or a TPU for hire. Groq's micro architecture does not matter. The *only* reason Groq has any traction is because it bet on SRAM. Without SRAM, there's no speed advantage, no PMF, no demand, and no
  • @mgsiegler.com M.G. Siegler on bluesky
    Maybe Groq's chips are legit, maybe they not, or maybe they're not *yet*, but even $20B is a relatively - for NVIDIA - small price to pay to effectively lock this team and tech up.  Some last-minute Christmas shopping for Jensen Huang...
  • @jerrycap @jerrycap on x
    “Increases my confidence that all ASICs except TPU, AI5 and Trainium will eventually be canceled.”
  • @raghuraghuram Raghu Raghuram on x
    @GavinSBaker Yes..workload/segment specific infra. Groq could be to nvidia what instagram was to Facebook at that time. Diff workload segments in one case different demographic segments in the other.
  • @raghuraghuram Raghu Raghuram on x
    @GavinSBaker The nuance is that long context decoding activity is likely not helped by this architecture.
  • @chamath Chamath Palihapitiya on x
    @GavinSBaker @JonathanRoss321 Agreed. He's a star of stars. Can't fade nvidia. Would be very foolish.
  • @gavinsbaker Gavin Baker on x
    @AccBalanced @weka Yes. Best way to deal with the flash shortage is to get more out of the flash in every Blackwell rack.
  • @gavinsbaker Gavin Baker on x
    @RaghuRaghuram Absolutely - Rubin CPX plus Rubin for those workloads. Mix and match.
  • @gavinsbaker Gavin Baker on x
    @chamath Thanks Chamath - interesting thoughts. Time will tell as ever. Should have also said that Nvidia is getting an extremely talented team led by the brilliant @JonathanRoss321 who spent a long time in the wilderness and chewed a lot of metaphorical glass to get Groq to this
  • @trungtphan Trung Phan on x
    Groq CEO Jonathan Ross explains the importance of speed in delivering a service and relevance for AI. For Google, a 100 millisecond speed-up leads to 8% higher conversion rate. In consumer products, there's correlation with time to dopamine and higher margin (eg. cigs > soda). [v…
  • @rohanpaul_ai Rohan Paul on x
    Very insightful post by Gavin below on Nvidia's $20B Groq licensing deal. AI inference has 2 steps, Prefill & Decode. Prefill means the model reads your whole prompt and context. Decode means it writes the reply one small chunk of text at a time. These 2 steps like different [ima…
  • @chamath Chamath Palihapitiya on x
    This is directionally right. The HBM vs SRAM tradeoff in architecture design was clear many years ago. Those that picked HBM are in a queue behind Nvidia and Google. Good luck with that. More broadly, LLM decode patterns favor SRAM. But unlike Gavin, I think this creates a
  • @patrickmoorhead Patrick Moorhead on x
    Mostly aligned with Gavin on this. Whenever I was asked, “how does NVIDIA compete with ASICs” my response has always been for the past year: 1/ the AI pipeline will split into three distinct workloads 2/ CPX fills one, maybe two of the three workloads 3/ NVIDIA will have to fill
  • @gavinsbaker Gavin Baker on x
    For the sake of clarity and as some have pointed in the replies, I should note that Nvidia is not actually acquiring Grok. It is a non-exclusive licensing agreement with some Grok engineers joining Nvidia. Grok will continue to operate their cloud business as an independent
  • @beffjezos @beffjezos on x
    The best take on all this Groq acquisition strategy
  • @prietschka Paul Rietschka on bluesky
    “[U]ltra-low latency agentic reasoning workloads”  —  These guys claim to be analysts doing analyst things, but at core they're just arranging words together like those magnetic mad libs for refrigerators.  [embedded post]