Introducing MongoDB $MDB Atlas Revenue Forecast.
MongoDB Atlas Revenue Forecast tracks the signals that shape MDB's growth trajectory in real time, not after the fact.
Atlas Revenue Forecast
Atlas is the growth engine. The Play forecasts Atlas revenue ahead of official
I agree with your core point: from the perspective of the active decode path for a single sequence, Kimi's hybrid linear attention significantly reduces KV fetch overhead. Roughly 75% of the layers maintain only a fixed-size recurrent state, shifting the critical bottleneck
The argument that deepseek cuts KV down more is moot, because every layer has KV and it grows with context.
Additionally searching through and fetching KV is the bottleneck of the operation. For Kimi it’s fetching the previous state layer which is a few megabytes at most and
Preview| $NOW 2Q26: New Pricing Pull-Forward Tailwind Drives the Beat, AI Still Lukewarm
We surveyed 3 top ServiceNow channel experts this quarter. Feedback points to a broad-based sequential acceleration in 2Q26, driven by 1Q’s delayed orders landing in 2Q, recovering US
Deep|LLM: Kimi K3's KV Cache Is Smaller - Why That May Actually Be Positive for DRAM/NAND
K3 is still a large-memory model: Kimi K3 has 2.8T total parameters, activates 16 of 896 experts per token, and Moonshot recommends deployment on high-bandwidth supernodes with at least 64
Research|LLM: Kimi K3 - Scaling Still Works; An Expensive Model Competing at Front Tier
Kimi K3’s overall performance now sits in the global top tier. It is a large 2.8-trillion-parameter MoE model with a 1-million-token context window, and it stands out in tests spanning
Weekly|Kimi K3 "DeepSeek Moment" Jitters, TSM and $ASML Beat-and-Raise, $ATI & the Starship Supply Chain, $INTC , $AXTI , $NBIS
Semis got smoked this week, with the indexes closing out weekly losses on fears of another “DeepSeek moment” after Kimi K3’s release. We think that
I don't think the gross margin has reached that high. The actual numbers should be much lower than this. DeepSeek is unique in its aggressive attempt at NAND-Based KV Cache, which KIMI has not yet achieved.
More relevant as a reference is GLM's gross margin of approximately
Yes
Take Kimi K3 for example, the total parameters are 2.8T (served at fp4, so the total size will be around 1.4T)
Kimi K3 is also the sparsest frontier model on the market at just 1.7%
They have at least 75%-85% GM on inference
GOOD: Morgan Stanley's sales desk called me a supply chain smartie in an email to their hedge fund and mutual fund clients.
BAD: My name is misspelled on second mention. :/
GREAT: Stock market is much smarter now versus DeepSeek last year because good smart takes FROM SUBSTACKS
The deadliest mountain in the world is K2 in Pakistan.
But how deadly will the new K3 model by Moonshot be?
In my view, this is not another "DeepSeek" moment. Kimi K3 is net-positive for the AI trade and compute demand, while being a net-negative for Western fronteir labs.
Research|LLM: Kimi K3 - Scaling Still Works; An Expensive Model Competing at Front Tier
Kimi K3’s overall performance now sits in the global top tier. It is a large 2.8-trillion-parameter MoE model with a 1-million-token context window, and it stands out in tests spanning
FUNDA Search is now live on our research platform.
Enter a company, ticker, theme, technology, or question, and get a summarized answer drawn from reports, meeting transcripts, earnings calls, and more in seconds.
Each result links directly to its source. Primary evidence and
Aerospace Catalyst Tracker has expanded its research coverage with 5 new modules now live on the platform.
Supply Chain Network
Companies across the commercial-space value chain are organized into layers, from raw materials and propulsion through to government and regulatory
(1/2)Please tell me the full background and context of why the lawsuit was filed, and how it may affect Samsung as well as the broader DRAM industry: The U.S. International Trade Commission voted to institute a Section 337 investigation into certain dynamic random-access memory
(2/2)Please tell me the full background and context of why the lawsuit was filed, and how it may affect Samsung as well as the broader DRAM industry: The U.S. International Trade Commission voted to institute a Section 337 investigation into certain dynamic random-access memory
Deep| $ATI : The Materials Chokepoint of the Starship Era; Superalloy Content Per Vehicle Is 25-30x a Falcon 9
ATI is framed as the public-market chokepoint for Starship-era superalloy demand: ~39 Raptors per stack at 800kg-1t each imply 30-40t of superalloy, or 25-30x a Falcon
