When Jim Cramer tells his massive audience of loyal followers that it is “only a matter of time” before OpenAI admits it has fallen behind Google’s Gemini 3, you know the market narrative is changing. In his recent comments, Cramer warned that a rush of “tens of millions of users” could move from OpenAI to Gemini 3 as Google’s model pulls ahead.
For anyone who has been closely reading Jason Criddle’s latest article on DOMINAIT.ai, Ryker, and what he calls “durable intelligence,” that shift feels less like a surprise and more like confirmation.
Criddle has been saying for years that OpenAI and its peers are not building the future of artificial intelligence at all. In his words, the giants are “building toys” while companies like DOMINAIT.ai are quietly building the next generation of real AI infrastructure.
Putting it as simply as I can bear, the cracks Cramer sees in OpenAI’s dominance are the same cracks Criddle has been describing from the inside of the AI industry.
OpenAI’s LLM Problem
OpenAI’s flagship product, ChatGPT, is still wildly popular, but popularity is not the same as durability. Cramer’s comments highlighted exactly that. As he pointed out, Gemini 3 is not just another chatbot release. It represents a serious threat, backed by Google’s reach and its custom Tensor Processing Units that directly challenge NVIDIA’s grip on the AI hardware stack.
The deeper issue is architectural. It is now going further than simple software as other players catch up. OpenAI’s current strategy remains locked into large language model thinking. Bigger models, more compute, more parameters. That worked when LLMs defined the landscape, but the frontier has now moved.
Google is integrating Gemini into search, productivity tools, and Android at a platform level. Chinese contender DeepSeek is proving you can build competitive AI systems at a fraction of the cost, accelerating the debate over an AI bubble and exposing how fragile many Western AI valuations really are… which is what Criddle has been proving all along.
On top of that, open and community driven ecosystems like Hugging Face are giving developers powerful tools without forcing them into the walled garden of any single company. In other words, while OpenAI focuses on centralized products and closed APIs, the rest of the world is racing toward distributed, interoperable, and cost aware AI.
This is exactly the landscape Criddle described in his latest DOMINAIT.ai piece when he wrote that “compute first AI is unsustainable” and that trillion parameter LLMs “do not equal intelligence.”
Toys vs Infrastructure
The key criticism that runs through Criddle’s writing is simple. Current AI leaders are building expensive demonstrations, not enduring infrastructure. The toys are getting shinier and more viral, but they are not solving the core challenges of resilience, safety, and real world autonomy.
Criddle argues that most of the AI boom has been a race to ship chatbots and image generators while ignoring structural weaknesses such as:
- dependence on a single vendor for GPUs
- centralization of models in a few hyperscale data centers
- fragile safety systems retrofitted on top of giant models
- runaway cloud costs that investors are finally starting to question
Image Courtesy of DOMINAIT.ai
In his latest article he states bluntly that “AI demand remains strong, but the industry’s financial structure is unsustainable,” and that investors are waking up to the question, “Is this durable” when they look at AI companies.
Jim Cramer is effectively asking the same question from a market perspective.
His recent prediction that OpenAI will have to admit it has fallen behind Gemini 3 is less about one model beating another and more about a strategic pivot of anything. If OpenAI continues to build incremental upgrades to a centralized chatbot, while competitors roll out smarter, more integrated, and more efficient systems, the market will not wait politely.
As Criddle likes to say, “#sorrynotsorry, Sam Altman.”
DeepSeek and Hugging Face have already shown that you do not need OpenAI scale to matter. Google’s Gemini 3 is now proving that with enough infrastructure and product reach, you can make OpenAI look reactive.
That is where DOMINAIT.ai and Ryker enter the story. And their launch is right around the corner, coming in early 2026.
DOMINAIT.ai and RCI, Ryker Class Intelligence
The most striking thing about Criddle’s DOMINAIT.ai work is that it has never been about chasing OpenAI. He is not trying to build a slightly better ChatGPT… it’s never been about competition. He is trying to build an entirely different class of intelligence that he calls RCI, or Ryker Class Intelligence.
In his newest article and many others, he describes Ryker as a reasoning engine, not a language model. “RCI isn’t an LLM. It is a new brain. A different level of intelligence entirely,” he writes.
Ryker does not learn by memorizing and predicting text (although he can),but by understanding why humans solve problems the way they do. The goal is not to answer more prompts. The goal is to orchestrate actions across devices, business systems, and distributed nodes.
I also heard a rumor of a new Cybersecurity tool they will be launching soon after Alpha with a team member from SentinelOne and Arms Cyber, but I still need to get the scoop.
Where OpenAI is bounded by a centralized service model, DOMINAIT.ai runs Ryker across The Grid, a distributed network of user owned nodes. Instead of pouring billions into more data centers, DOMINAIT.ai scales by connecting more participants and using their existing compute.
Criddle puts it plainly when he states: hyperscalers “cannot outspend physics,” while DOMINAIT.ai sidesteps the compute bottleneck entirely by decentralizing the workload.
Image Courtesy of DOMINAIT.ai
This approach directly targets the weaknesses that people like Cramer and others see forming around OpenAI. If the industry is heading into a period where investors punish heavy capital expenditure without clear paths to profitability, an AI network that grows without hoarding GPUs and cloud spend starts to look very attractive to portfolios.
From LLM Bubble to Durable Intelligence
One of the more interesting subtexts in all of this is the emerging idea that we are not in an AI bubble in general, but in an LLM bubble specifically. Many hundreds of billions have been poured into systems that are fantastic at autocomplete and content generation, but limited when it comes to grounded reasoning, safety, and autonomy. Toys…
Criddle’s writing leans into this with lines such as “The bubble is the belief that LLMs are the future of intelligence” and “The giants are building toys. DOMINAIT is building intelligence.” These are not throwaway jabs. They are an architectural stance.
And well, maybe a few jabs..
Cramer’s recent remarks sit on the other side of that same argument. By highlighting how rapidly Gemini 3 can attract users at OpenAI’s expense, he is pointing out how fragile LLM moats really are. If a single new model from a better positioned competitor can trigger what he calls a rush of users to a different platform, then OpenAI’s dominance is contingent, not structural.
I say this as I literally get this notification from my email:
Dominance built on toys is temporary.
Dominance built on infrastructure is durable.
DOMINAIT, which is built on intelligence, is the next stage of AI.
DeepSeek is already pressuring Western AI firms by proving that cheaper, leaner models can still be competitive. Hugging Face has become the default backbone of the open AI ecosystem. Google is now using Gemini 3 to anchor its entire product universe. Each in its own way is a sign that the post ChatGPT era is arriving.
Meanwhile, DOMINAIT.ai is positioning Ryker to be the intelligence that runs inside that era rather than a toy that entertains it.
Why DOMINAIT.ai and Ryker Could Be the Ones That Truly Change the Environment
If OpenAI’s problem is that it has been building toys, then the solution is an AI ecosystem that treats intelligence, not as a consumer novelty, but as critical infrastructure. That is the gap DOMINAIT.ai is trying to fill. Well, not just trying, but doing.
Ryker is being trained to operate business tools like QuickBooks, Salesforce, Calendly, publishing platforms, and custom enterprise systems. He is not just summarizing emails or drafting copy. He is orchestrating workflows across companies, reconciling financials, scheduling, writing and publishing, and coordinating teams in real time. In Criddle’s words, “Ryker does not just execute tasks. He learns why you do them.”
That kind of reasoning is exactly what LLM bound systems struggle with. It also happens to be the kind of intelligence that corporations, investors, and governments will pay for long after the novelty of chatbots wears off.
OpenAI Proved the Market for AI. DOMINAIT.ai Wants to Redefine It.
Jim Cramer’s prediction that OpenAI will have to admit it has fallen behind Google Gemini is not necessarily the end of OpenAI, but it is a sign. A sign confirmed by Sam Altman’s recent RED ALERT.
It signals that the era of automatic dominance is over. The field now belongs to whoever can deliver durable, distributed, reasoning driven intelligence.
Google is making its play with Gemini 3. DeepSeek is proving cost efficient models can still win. Hugging Face is empowering the open ecosystem.
And DOMINAIT.ai, with Ryker and RCI, is quietly building the kind of AI infrastructure that can operate across all of it.
If the giants have been building toys, then the next wave belongs to the builders of tools, grids, and thinking systems. In that race, OpenAI may have started first, but it is far from clear that it will finish ahead of Google, DeepSeek, Hugging Face, or DOMINAIT.ai and Ryker.
For investor inquiries, send an email to info@dominait.ai
