AI is Starting to WIPE OUT Entire Job Sectors (and it will only accelerate)
Mike Adams warns Anthropic’s COBOL-capable AI, Claude Code, wiped $30B from IBM’s valuation in hours by automating legacy systems maintenance. Its security scanning feature also crashed CrowdStrike’s stock, signaling AI’s rapid job-sector disruption—programming, customer service, and cybersecurity face 2-3 year displacement. DeepSeek V4, a Chinese open-source rival with a million-token context window, threatens Anthropic’s paid model by offering superior free performance, while U.S. AI firms may depend on government contracts. Adams suggests Chinese engines outpace American ones in capability and privacy, urging listeners to adopt open-source alternatives before AI’s self-improving singularity renders human expertise obsolete. [Automatically generated summary]
Welcome to this special report on the AI wipeouts of stock market valuation.
I'm Mike Adams.
I'm an AI developer.
I built BrightLearn.ai, the largest book publishing platform in the world, and many other AI projects.
And if you've been watching what just happened with IBM, in one afternoon, IBM lost $30 billion in market valuation because of a tweet, essentially.
and because of some technology from Anthropic.
So Anthropic announced that they can now, or its coding agents can now understand and read and edit and write COBOL.
Now, COBOL is an ancient programming language.
I mean, that harkens back to even before I was in college.
COBOL, my goodness.
Well, it turns out that 95% of ATMs, yes, the cash machines, their transactions run on COBOL.
And COBOL is also powering a lot of government systems, which are also ancient, and some airline systems as well.
IBM makes a huge amount of money every year running the systems integration contracts and the maintenance contracts for all this COBOL code.
And it's actually very difficult to find human programmers that can even work in COBOL because it's such an ancient language.
Well, the fact that Anthropic now says it can handle COBOL, this just instantly vaporized $30 billion in market valuation for IBM because it means that a lot of these companies that have been paying IBM huge amounts of money will be able to automate it using Claude code, which of course is from Anthropic.
Now, so this is a tech company partial wipeout, you could say, or a correction, simply because of new AI technology coming online.
This is not the first example.
Last week, Anthropic announced a new feature for its clawed code called Claude Code Security.
Now, remember, they have Claude Code.
have Claude co-work which sort of works on your desktop and maybe accidentally deletes your folders and then they have they now I use Claude code for coding by the way It's very good.
I use Opus 4.6.
It's the best thing out there at the moment.
But they announced this Claude Code Security, which does vulnerability scanning.
It's very, very good.
I've used it.
It's crazy.
It finds all kinds of privacy vulnerabilities as well as security gaps in your code.
And it's just, it's incredible.
It saves you, you know, it saves all kinds of time that humans would typically have to use in order to find these vulnerabilities.
So when that was announced, the stock price of CrowdStrike, a computer security company, network security, etc., their stock price plummeted.
And so every few days now, Anthropic is announcing something new that is clobbering the valuation of tech companies because it's replacing lots of different services.
Now, I probably don't have to tell you that this is also going to be replacing a lot of different human jobs.
So fewer security code experts will be employed.
Fewer COBOL programmers will be employed.
And overall, there will be fewer coders employed and fewer network administrators, et cetera, because of the automation that's coming from companies like Anthropic.
Now, Anthropic is currently valued at $380 billion.
They're in a war with the Pentagon because the Pentagon says they want Anthropic to allow them to use their software for autonomous killing systems and mass surveillance, etc.
And the founder of Anthropic or the CEO of Anthropic, Dario, he says, no, we won't allow you to use it for that.
And the Pentagon is threatening to yank Anthropic out of the supply chain for all military contracts.
That would be devastating to Anthropic, and it could send the company reeling, frankly, even though it's got the best coding product out there right now by far.
However, there's something else that's coming.
But first, I want to just summarize what I said here, which is that Anthropic or Google or Microsoft on any given day could announce something that wipes out an industry, that just completely wipes out or largely wipes out the revenue model of an entire industry.
And that is happening.
And that could continue to happen with some frequency.
I mean, I'm sure it will.
But Anthropic itself is threatened by another factor, which is Chinese open source models.
And China is running the real 5D chess here.
China's running the long game.
China is undermining the US AI marketplace by releasing world-class models completely free.
Models that can be downloaded and run locally on your own hardware, free of charge, and that are very good.
And DeepSeek version 4 is rumored to be right around the corner with a release.
In fact, it's about a week late compared to what the rumor said.
But DeepSeek version 4 is rumored to be neck and neck with Anthropic's Opus 4.6 when it comes to writing code and debugging code.
But importantly, DeepSeek version 4 is also rumored to have a much larger context window, like a million tokens, which means it can ingest and review a massive code base.
And it can essentially edit or build features or debug features across multiple repositories and different languages, sort of full stack development, which is a very big deal.
Until now, we've all been limited to much smaller context windows like 128K is typical.
So even my own projects vastly exceed the 128K tokens.
So you may not realize this, but Brightlearn.ai, that engine that builds your books for you, and I'm not even talking about the front end, but just the back end engine, that's over 100,000 lines of code.
I didn't write a single one of them, by the way.
They're all AI written lines, but it's over 100,000 lines.
That's way over 128,000 tokens by a long shot.
It's closer to a million tokens.
So what DeepSeek version 4 is about to do is revolutionary, and it could really put tremendous profit pressure on Anthropic.
What I'm getting at here is that what Anthropic just did to IBM, DeepSeek could do to Anthropic, which is undercutting the revenue model of Anthropic, also undercut the revenue model of IBM and, you know, security companies, etc.
This undercutting, this seemingly spontaneous announcements of new technology, this is going to happen on an increasing frequency.
You're going to hear these kinds of stories now almost every week, one way or another.
And this is part of the acceleration toward what many people call the singularity, which is where AI is involved in its own self-improvement in such a profound way that humans won't be able to keep up with the new improvements that AI is rolling out.
So DeepSeek version 4, when it is released, I'm going to be taking a hard look at it and testing it and seeing if we can integrate it in certain areas and seeing if it's got better intelligence or better reasoning or better code writing, better debugging, etc.
And I will keep you posted about DeepSeek version 4.
There's also rumored to be a light version of DeepSeek version 4 that will run on consumer-grade hardware that will run on about 24 gigs of RAM on your GPU.
This is also a huge deal because 24 gigabyte cards are somewhat available.
They're more on the higher end of consumer grade cards, but they're not impossible to find.
I think the NVIDIA, what is it, the 5080 plus or something has 24 gigs on it.
I know the 5090 has 24 gigs, or I mean 32 gigs, which is more than enough, but the 5090s costs $5,000 now, so that's not necessarily a cheap card.
Nevertheless, if DeepSeek Lite is released, and if it's anywhere near as good as the full model of DeepSeek, this is going to undercut, I mean, Google, Gemini, OpenAI will probably be headed towards bankruptcy because no one will need to pay OpenAI these monthly subscriptions or these high per token fees, even though you're paying per million tokens input and output.
You're going to be able to do that for free by downloading DeepSeek version 4 and running it on a graphics card that you only have to buy one time.
And then you're just paying for electricity.
And companies will do this company-wide.
They'll have one GPU workstation or server in their company running DeepSeek or DeepSeek Lite, or maybe they'll run Quen, you know, Quen 3.5 is out, etc.
And that will be accessible by everybody within the company.
Company-Wide AI Localization00:05:59
And that way, all your data stays local.
And your customer data, your financial information never leaves your company.
Because when you're using an open source AI model locally, it's not sending information back to China or anywhere.
It's just running locally.
And that's it.
So it's the ultimate for privacy.
So let me summarize where we are with AI real quick, especially for those of you who are maybe not AI developers.
And maybe you're not at the bleeding edge of this technology.
So here's the deal.
This is my experience.
AI is clearly intelligent, highly intelligent, highly capable.
Those who say it doesn't think or it's just a word prediction engine, I don't know, they're stuck in 2023 or something, which is five generations ago when it comes to AI.
So today's AI is highly intelligent.
It's going to replace entire industries.
Seriously, I mean, we just covered some of it.
Who needs COBOL programmers?
Or for that matter, you know, I mean, who needs React programmers for websites?
You don't.
You use AI to do it.
Who needs Python programmers, right?
I write everything in Python that's for document handling, but now I just have AI write it, and I just tell it the structure and what I want, inputs and outputs and parameters, things like that.
So yeah, AI is going to just sweep through the workplace over the next two to three years, and it's going to replace a lot of desk workers, a lot of middle managers, decision makers, of course, customer service people and programmers, obviously security people, etc.
But it's going to be more and more capable as we move into agentic AI.
And this is the year that a lot of AI agents are being toyed with, let's say, with OpenClaw, et cetera, which I don't really recommend that you install because it's a big security problem.
But this is the discovery phase of agentic AI and personal assistance.
It's going to mature very quickly over the next few months.
And then in 2027, I predict we're going to have a lot of agentic AI systems deployed across corporations in decision-making roles and logistics roles, etc.
You know, judgment AI.
So get ready for that.
The Chinese models, the open source models are probably going to win in the end here, commercially speaking, even though these are non-commercial models.
What I mean is they're going to undermine the entire commercial sector of AI, which is Google and OpenAI, etc.
I'm not sure how OpenAI can financially survive this, frankly.
And I'm starting to wonder about Anthropic, too.
I really don't know if Anthropic will survive if DeepSeek version 4 is just as good and if Anthropic loses the government as one of its bigger customers.
So I guess we'll have to see.
OpenAI will probably need a lot of government bailouts to continue to exist.
And it will largely become the AI engine for the Pentagon and government workers, etc.
But everybody in the corporate world and in the nonprofit world, et cetera, they're all going to be using the Chinese models because those models are free and great and also private when you deploy them locally.
And in many ways, they're also smarter, much smarter models.
So that's where this is all going.
And China's playing the long game here.
Granted, they're playing 5D chess.
They're very clever.
And with this strategy of them pushing out these open source, completely free frontier models, this is absolutely going to undermine the U.S. AI industry.
But I'm not really a fan of the U.S. AI industry anyway, because it's all run by the CIA that influences what kind of answers that the U.S. AI engines will even give you.
For example, all the AI engines from U.S. tech companies lie to you about vaccines.
You know, they say all vaccines are safe and effective, and they lie to you about COVID and a thousand other topics.
9-11, you name it.
These engines are all fake.
Whereas you actually get a lot more truth out of China's engines as long as you don't ask it about politically sensitive subjects like Tiananmen Square or Taiwan, for example.
You know, you stay away from those topics, which is probably not your focus anyway.
And the Chinese engines are really, really great.
In fact, the best by far.
So that's my advice.
Use the open source Chinese engines.
Watch for the release of DeepSeek version 4, which could happen any day now.
I'll bring you more information here on my broadcast once that gets released.
And if you want to check out my AI projects and use my free book creation engine, it's at brightlearn.ai.
And we also have our deep research AI engine, which is at brightanswers.ai, now featuring all of the DOJ Epstein files that have been indexed in that system.
So you can ask it about anything in the Epstein files.
Oh, that should be fun.
Epstein Files Indexed00:00:48
I spent weeks cleaning up all those documents.
Well, I mean, I had AI do it, but it took a while because there's so many documents.
And some of them are so crazy.
I think my couple of my workstations vomited because of what was in those files.
But anyway, I cleaned it up.
And now it's all indexed.
And you can access it through brightanswers.ai, which is free to use also.
So check that out.
And you can follow all my videos and updates at brightvideos.com.
So I'm Mike Adams.
Thank you for listening.
Take care.
Beet juice is an refreshing beverage that can help you stay energized.
You can easily incorporate beet juice into your daily routine with Groovy B organic freeze-dried beet juice powder.