Account Info
Log Out
English
Back
Log in to access Online Inquiry
Back to the Top
AI chip giants gather at Computex 2024: Will it bring new opportunities?
Views 2.9M Contents 257

$SNOW keynote at customer event was pretty thin.

It seems like either 1) $Snowflake(SNOW.US)$ is very early / behind in developing GenAI solutions with customers, or 2) GenAI adoption at customers is just too early, or both.

First, CEO’s pitch was that their AI team has good people but there was no AI product showcase / demo. So it seems like bet is that they can catch-up over time. Also implies they are quite far behind their competitors who have more fully-fledged products / demos live today.

Second, the customer panel ought to have been an opportunity for large customers to highlight how they plan to roll out AI w/ SNOW but was really just a high-level discussion and it wasn’t clear at all that GenAI projects will be based on SNOW (as opposed to Databricks, etc.).

Highlights
-CEO seems like a good technologist but not the most inspiring speaker. Has to compete against visionary personalities at $Microsoft(MSFT.US)$ / $Databricks.

- $NVIDIA(NVDA.US)$ CEO joined virtually from Taipei where “all AI infrastructure is made [stick head in geopolitical sand]. Announced $Snowflake(SNOW.US)$ will use $NVIDIA(NVDA.US)$ micro-services to do more AI function [seems like $NVIDIA(NVDA.US)$ doing more of the value-add in this relationship]. Jensen’s answer for how NVDA itself is using AI was pretty thin [mostly coding enhancements]

-Customer panel was pretty high-level. Seems like most focused on meat / potatoes of data management. GenAI projects were more futuristic and no one specifically said using $Snowflake(SNOW.US)$ for this.

Notes
AI strategy
-Unstructured is foundation of AI
-Basically we know what people need for AI, have a good team
-Cortex allows you to use a bunch of models – GOOG, META, Mistral, Reka, Snowflake [No Open.AI???]
-NVDA CEO joined via webinar from Taipei and said “all AI infrastructure made here” [virtual joining? Must not have bought enough GPUs!] [Also
-Arctic developed for only a few million dollars
-SNOW will now use NVDA NEMO retriever – basically these are NVDA microservices that do all the AI work on top of your data. [So bear case is more advanced functionality being developed by $NVIDIA(NVDA.US)$ and available to all – also available to data wherever it sits]

NVDA
-Using AI to help w/ design of chips
-Engineers can try experiments that previously couldn’t do
-Use AI to invent new businesses [what? then mentioned autonomous AV / robotics]
-Every group in company looking for ways to apply to GenAI

Customer Panel
-JPM – been using ML/AI for a number of years. Been using for risk avoidance / cost avoidance, and now focused on revenue-generating side like personal offers for customers, sales coaches, contact centers, etc. Have been investing in GenAI in engineering, cust service and operations.
-NYC Health / hospitals – more about traditional data warehouse / sharing use cases
-Ericsson -
-BKNG – heavily use ML/AI for structured data (advertising / product suggestion / etc.). GenAI will be another tool [but didn’t highlight whether using $Snowflake(SNOW.US)$ for this]. Excited for Iceberg (taking storage out of $Snowflake(SNOW.US)$ ).
Disclaimer: Community is offered by Moomoo Technologies Inc. and is for educational purposes only. Read more
4
+0
Translate
Report
41K Views
Comment
Sign in to post a comment