The Outsider Perspective Issue 428 | The Beltway Outsiders

If you’d like to read this issue on my website, click here! If you’d like to sign-up, and receive this in your inbox each week, click here! Read past issues here.

Good Friday Morning! Except to Punxsutawney Phil, who will be asked about his shadow this weekend. NOAA published a study determining which groundhog was the most accurate at predicting early or late spring. Phil didn’t make the list. The most accurate groundhog is “Staten Island Chuck, who accurately predicted the arrival of spring 85% of the time.”

After Chuck, “The rest of the top five were Georgia’s General Beauregard Lee, at 80%; a Wyoming Prairie Dog statue named Lander Lil, with 75%; a mythical West Virginia groundhog named Concord Charlie, at 65%; and Illinois’ Gertie the Groundhog, with 65 percent.” Sadly, Punxsutawney Phil is only accurate 35% of the time, hardly useful at all. Fortunately, Bill Murray’s classic movie “Groundhog Day” is hilarious 100% of the time.

Thank you to everyone who signed up for the American Almanac. The link I sent last week ended up going dead on us. Please sign up here, if you’ve not received the Alamanc in your inbox in the past two weeks: https://americanalmanac.beehiiv.com/subscribe

In just two active weeks, we’re already up to 2,500 subscribers and growing fast. We appreciate you signing up, and look forward to making this one of the premier newsletters out there. If you still have issues signing up, email me directly at daniel.v@conservativeinstitute.org – I’ll have our tech team take a look at things.

This week witnessed an artificial intelligence bombshell. China, long thought to be behind on the technology, launched its own challenger to ChatGPT called DeepSeek. That has shaken Wall Street and the federal government. I’ll get into that and more below—links to follow.

Quick Hits:

The devastating plane crash in Washington D.C. is awful. It’s the worst plane disaster in the U.S. since 2009 and the most deadly since 2001. Charles C. W. Cooke had a good post arguing that part of what makes this shocking is that U.S. plane disasters are exceedingly rare. There’s been a rush to blame one political side or the other, but that is too early. A commercial pilot who runs the Captain Steeeve YouTube channel did a walkthrough explaining the Air Traffic Control recordings for the layman. From his vantage point, it’s a mistake of the Black Hawk helicopter crew. We’ll see after the investigation.

Where you can find me this week

Please subscribe, rate, and review The Horse Race on YouTube — the reviews help listeners, and readers like you find me. Make sure to sign up for the Conservative Institute’s daily newsletter.

It Turns Out Biden Released Murderers, Too – Conservative Institute

Senate Democrats Rush To Protect Antisemitic International Court – Conservative Institute

Trump Begins His Fight Against Inflation – Conservative Institute

The United States Hits An Allegeged ‘Sputnik’ Moment In AI

Early in the week, U.S. markets were shaken by Chinese AI startup DeepSeek’s arrival in the app stores and skyrocketing to the top of every list. The sell-off hit hard because DeepSeek claims it can do everything OpenAI’s ChatGPT can do (and better) but at a fraction of the cost.

Who does cheaper AI impact? Companies like NVIDIA have skyrocketed to the top of the list of most valuable companies because their chips are vital in AI development. It also hits the broader tech sector as a whole because if AI is actually cheaper, that makes you question the valuation of the expense side of the ledger and whether everyone is focusing on this right now.

There are many assumptions in what I’ve said so far, not all of which I believe. But it gives you an idea of what’s happening. In effect, China told the United States that it could do what the United States is doing, only a lot cheaper. Where the United States has had to shell out billions to develop these large language models like ChatGPT, DeepSeek claims it only took them between $5.5 – $6 million.

Or, as Marc Andreesen put it, “Deepseek R1 is AI’s Sputnik moment.”

For those who don’t know that reference, Sputnik was the USSR’s first satellite into orbit. It stunned the United States, which got caught with its pants down and kicked off the space race. The Soviets also leaned into this in their propaganda to claim they were better than the United States.

If you want to get into the nitty-gritty of DeepSeek’s situation, I’ve been recommending a CNBC documentary on it that was published a week ago, before the market sell-off. It’s called How China’s New AI Model DeepSeek Is Threatening U.S. Dominance. I shared this on X, and Jim Geraghty used it in his write-up on the topic in the Morning Jolt.

The Chinese narrative that they magically created a better AI model than anyone else has a few problems.

China ripped off ChatGPT. If you ask DeepSeek’s model what it is, it will tell you it is OpenAI’s ChatGPT. Microsoft and OpenAI have launched an investigation to see if DeepSeek stole proprietary information from OpenAI or is using another source. I suspect they’ll find out China stole information and data.
The U.S. Government is investigating DeepSeek to see if it got access to illegal processing chips. Under current export bans to China, we cannot ship them NVIDIA H100 chips. These are the high-powered NVIDIA chips driving AI innovation. DeepSeek claims it didn’t need those chips and innovated new ways to build AI models. Some tech founders disagree: “Elon Musk responded to a clip of Scale AI CEO Alexandr Wang, who said that DeepSeek has around 50,000 NVIDIA H100s that they cannot talk about due to U.S. export controls. Musk wrote in response: “Obviously.“
We know nothing about who DeepSeek is. Jim Geraghty makes this point in his newsletter, and I agree. All the claims of this being some plucky Chinese lab don’t make much sense. Like many Chinese national companies, the connections between them and the ruling party are… murky.

There’s one other thing that should raise everyone’s eyebrows, too. Trump was accelerating Project Stargate on the United States side, announcing $500 billion in funding for a massive AI project to keep the United States at the forefront of this technological revolution.

A few days later, we get China going full-bore into launching an open-source product, seeking to undermine every other AI platform by being cheaper than everyone else. The play here from the Chinese is like everything else they’ve done: steal American designs, copy them, and then sell them back to us at pennies on the dollar to drive American companies out of business.

We’ve seen this in rare earth minerals and many other industries. China will subsidize an industry to give its companies an unfair advantage in the global market. What it struggles to do is innovate from that, which keeps it behind the United States. With DeepSeek, it’s claiming it innovated and made something in the United States better.

The thing is, it is open source, so we can verify what DeepSeek is claiming. Some of the claims are true. You’ll likely see U.S. companies use those innovations to drive further into artificial intelligence. So even if DeepSeek ripped off ChatGPT completely, they have tinkered enough to improve it and decrease costs. Whether that matters long term is another matter because we’re trying to get AI to improve at every turn. If China’s claims are true, they’ve only innovated on a model while not building anything new.

Put another way, you can fine-tune a Honda Civic and pull tons of power out of it. But if you’re racing a McLaren, you’ll get dusted.

In the end, though, this truly kickstarts a global artificial intelligence arms race. This is the new Space Race, and there’s no going back. All the talk about ethics, morality, and more are over. Even Elon Musk, who talks a lot about the dangers of AI, understands that AI in the hands of the Chinese is a dangerous bridge. American companies have their issues, but the Chinese are actively looking to rewrite history and reality with their versions of AI.

The country that wins the AI race will dominate the 20th century. The race will be over who owns the AI platform that everyone uses. Right now, that’s an open question. But as more hits the open-source space, you’ll see developers move to that because it’s cheaper to build on top of open-source software than propriety systems.

I agree with Jordan Schachtel, who wrote that the U.S. response to China’s AI may define his second term.

I’ve written many times about how AI is impacting every industry. That is continuing. Meta advertised AI chatbots to me on Facebook this week, one of which was an AI Astrology Chatbot with more than 15 million users. The impact of those things is still growing, but the competition between the United States and China will outpace all of that.

It’s an artificial intelligence arms race.

Links of the week

U.S. Navy bans use of DeepSeek due to ‘security and ethical concerns’ – CNBC

Talk loudly and swing a yuge stick: Trump does Teddy Roosevelt one better – NYPost

Federal workers reeling over Trump’s ‘buyout’ offer: ‘This is the last lifeboat in town’ – Politico

The case of the radical ‘Zizian’ vegan trans cult and the shooting death of border patrol agent – NYPost

The Democrats’ Governance Problem: They’re bad at it and getting worse. – The Liberal Patriot

The Costs of Cultural Illiteracy: Lessons From the Israel/Hamas War – RealClearPolitics

Kash Patel says FBI needs to ditch politics, get back to fighting crime – The Washington Times

NTSB Chair Calls Out Media Hypocrisy on Crash Speculation – PJ Media

Meta to Pay $25 Million to Settle 2021 Trump Lawsuit: President had sued social-media company after his accounts were suspended following Jan. 6 riot – WSJ

Caroline Kennedy is Why I Am No Longer a Democrat: What has become of a once-great political party is now nothing more than a mouthpiece for the establishment government. – Sasha Stone