Yes, Trumpistan is subsidizing less. It may shock you to hear that, but oh well.

It just means that Paramount (and possibly Netflix) now have a solid reason to help the Republicans in 2026.

You don't seem to understand anything rogoshen said. Not one tiddly bit.

It's China, they're subsidized and they have to build something. Even if it's empty or otherwise represents overcapacity. Of course the people who built these datacentres may eventually lease the space to companies that will be forced to use domestically-produced hardware that no one wants to use (they'd rather have NV H200s or . . . whatever else). It's just how they do things, at least for now.

He ain't entirely wrong, in that environmentalist groups wouldn't dare set foot in China. They wouldn't last long over there.

Sure, want an H200?

Yes.

The Obamacare tax wasn't a tax either, until it was.

As the article mentioned, there are generally two ways that applications process data: either using patch processing or streaming data. For real-time data processing you generally use streaming data. In addition, in modern microservices architectures you generally use a messaging service between services to provide more reliability, scalability, and to more loosely couple the services.

For streaming data or messaging services, companies tend to either use a cloud service provider (CSP) native service like Google Cloud Pub/Sub or Amazon Web Services (AWS) Simple Queue Service (SQS). For companies that have their own data centers or want a uniform multi-cloud solution, they generally want a solution like Apache Kafka or a managed Kafka like Google Cloud Managed Service for Apache Kafka or Amazon Managed Streaming for Apache Kafka (MSK). Apache Kafka is faster than older solutions like RabbitMQ.

The streaming/messaging service is configured to meet the application requirements for message delivery like deliver once or deliver at least once, in order or out of order, dead letter queues, schema enforcement, etc. It allows for architectures with multiple publishers or multiple subscribers (e.g., fan-out architectures). It provides access controls, monitoring, and logging capabilities.

For companies that decide on managing their own Apache Kafka rather than paying a CSP to do it, they often want a commercial product with commercial support and that is what Confluent offers. Confluent also offers a Confluent Cloud service where it sounds like they deploy and manage Kafka in your AWS, Azure, or GCP environment as a cloud agnostic managed service (I am not familiar with that product).

What's funny is that Chainalysis probably can't include Monero activity in their report . . .

Read the actual Chainalysis report cited in the summary. Then get back to us. Here's a hint: money laundering accounts for less than 1% of all blockchain activity.

IF this tech spreads? What makes you think it hasn't already?

ChemE and/or MSE are potential first steps into the wild world of semiconductor fabrication.

That's . . . not entirely true. AI researchers are some of the most highly-sought-after professionals on the market today. Do they have degrees in "AI"? Probably not, since these degrees are fairly novel and it remains to be seen how well universities can train to meet the needs of companies hiring for AI research today. That being said, there's the distinct possibility that "AI degrees" may be mostly fluff.

Can he tariff California? (answer: no)