Still my favorite quote taken from this Board:
You do not beat the market by thinking you are “smarter” than the market. You beat it by understanding what the market never understands as it always either overreacts to FUD or underreacts to those few extreme world changing businesses.
Since the time I began to follow Saul’s advice and the contributions from everyone here:
|2023||Month to Date||Year to Date|
|Allocations as of||2/28/23||1/31/23||12/31/22||11/30/22||10/30/22|
This portfolio is what is in our non-taxable Roth and Rollover IRAs only. We have not added any money to these accounts for many years. To buy something I’ve sold something else. I don’t trade options or use any leverage. I stay fully invested at all times and keep less than 1% in cash.
This months included only one Investing Decision:
What I did: I trimmed 13% from Datadog and added 9.5% to my Snowflake position.
Why I did it: Datadog revenue growth slowed even more than I expected and Snow share price dropped more than Datadog, despite of or perhaps because of Snowflake’s purchase of three businesses in January.
Snowflake purchased Myst.ai. I see Myst.ai as a massive future growth driver.
Per Muji at Hhhypergrowth.com-
*Myst.ai provided a platform for time-series forecasting that was most focused on the energy industry. It allowed data teams to pull in data, enrich it with a specific set of 3rd-party data, to then build, validate, and deploy ML models for time-series forecasting quickly to predict energy prices, load demand, operating conditions, or green energy generation.
From a blog post: “The Myst Platform helps users build and deploy machine learning models to forecast time series, such as energy prices. Since a machine learning model is only as good as the data it has access to, we put a lot of thought and care into making the right data available to our users. Through our integration with our partner Yes Energy, market data across all Independent System Operators (ISOs) in the United States is accessible directly in the Myst Platform.”
This could be a hint of a coming energy-focused Industry Data Cloud, but I think it is much bigger than that. Essentially, they are a forecasting engine, which allows the use of open-source Python ML libraries (PyTorch, Scikit-learn, XGBoost) for rapid experimentation and data modeling through a low-code visual interface. While they focused heavily on energy industry use cases, this is a feature that is heavily applicable to any time-series data across industries (IoT, healthcare, fin svc, retail/CPG), and the announcement post walked through a number of use cases across those other industries (healthcare outcomes, supply chain mgmt, inventory planning). I expect this to be tightly integrated into Snowpark for Python to add a new low-code platform for time-series forecasting – and really any ML modeling (and open-source ML engine) use cases from there. I expect them to immediately open this up to wider use cases like revenue & inventory forecasting.*
Me here: LeapYear appears fundamental to what Snowflake provides, on the whole of it. And frankly, it scares me that the privacy they provided to their customers needed this additional layer. Perhaps this added to the drop in Snowflake share price.
*LeapYear is a platform focused on allowing privacy-preserving ML over sensitive data, which will clearly be folded in as a capability within Snowflake’s global data clean rooms. LeapYear provides “differential privacy” to allow orgs to share confidential information with “mathematically-proven privacy protection”. LeapYear has been creating solutions around specific data types and use cases, with a focus on the use cases in capital markets, data monetization, and the Federal gov.
LeapYear is at the bleeding edge here, with a blog post highlighting why differential privacy is needed. Existing data privacy techniques like dynamic data masking or anonymization can ultimately be reverse-engineered or reconstructed by enriching the obscured data with other datasets, and techniques like redaction and aggregation might lose the details important to ML/AI algorithms. Differential privacy has been researched for 15 years now (emerged in 2006 research), and while it has been adopted within Apple, Google, and the US Census Bureau, LeapYear calls itself one of the first commercial solutions built around it.
Some examples they gave were allowing healthcare to share PHI data from 100M+ patients with 3rd parties, allowing retail to share insights across data sovereignty borders, and for capital markets to analyze holdings across clients – all without exposing the underlying PI or sensitive customer data. The challenge from here is how specific the needs are around use cases across industries.*
Refrigeration made money; but, the big money from refrigeration was made by the Coke Cola Company.
Coke and Pepsi are in the packaging and distribution of sugar water business.
AI companies will make money; but, the big money will be made in the packaging and distribution of AI. IMO, Snowflake and Cloudflare are already the Coke and Pepsi for the data needing to be packaged and distributed to and from AI models.
My understanding is that Datadog now has just a single agent that can abstract away a mountain of work needed to coordinate operations and with one click will now add up to 18 additional areas businesses with a Cloud footprint need to use to coordinate their operations fully.
In the category of Artificial Intelligence for IT Operations (https://investors.datadoghq.com/news-releases/news-release-details/datadog-named-leader-aiops-independent-research-firm), Forresternot only singled out DDOG’s data insight and visualization capabilities but said its R&D boasts the “highest number of data scientists dedicated to machine learning and artificial intelligence product advancements.”