PLUS: OpenAI's math blunder, Nvidia's China exit, and the AI energy squeeze
Good morning
Wall Street is raising alarms about a "circular financing" trend emerging in the AI industry. Tech giants are funneling billions into startups, which then use that same capital to purchase products and services from their big-tech investors.
This feedback loop rapidly accelerates development, but it also raises a critical question: is the current AI boom driven by genuine market demand, or is it an investor-fueled bubble on the verge of popping?
In today’s Next in AI:
AI's circular financing feedback loop
OpenAI's major math misstep
Nvidia's China market share vanishes
The AI boom's growing energy costs
AI's Circular Money Machine
Next in AI: Wall Street is raising alarms about a "circular financing" loop in AI. Tech giants are investing billions into their startup customers, who then use that capital to buy back their investors' products, creating questions about the sustainability of the demand.
Decoded:
The most prominent example is Nvidia's extensive funding, with plans to invest up to $100B in OpenAI, alongside $6.3B into data center operator CoreWeave and a reported $2B into xAI.
This is not an isolated strategy, as OpenAI itself is building a network of deals with companies like Oracle, CoreWeave, and AMD to secure its own infrastructure.
The practice has investors on edge, with Blackstone president Jonathan Gray calling it a top concern and comparing the frenzy to the dot-com bubble.
Why It Matters: This feedback loop accelerates AI development by ensuring startups have the capital and hardware they need. However, it blurs the line between genuine market demand and investor-funded growth, posing a significant risk if the capital flow slows.
OpenAI's Math Blunder
Next in AI: OpenAI researchers faced public embarrassment after prematurely announcing GPT-5 solved 10 unsolved math problems, only to be corrected that the model simply found existing solutions in scientific literature.
Decoded:
Mathematician Thomas Bloom, who maintains the website listing the problems, clarified that the announcement was a “dramatic misrepresentation” because the problems were only listed as unsolved from his personal awareness.
The misstep drew sharp ridicule from rivals, with Meta's Chief AI Scientist Yann LeCun quipping that OpenAI was “hoisted by their own GPTards” and Google DeepMind's CEO calling the situation “embarrassing.”
An OpenAI researcher later acknowledged that only solutions in the literature were found, but argued that the model’s ability to effectively search scientific literature is still a significant technical accomplishment.
Why It Matters: The incident highlights the immense pressure on AI labs to announce major breakthroughs in a highly competitive field. While not a display of novel reasoning, GPT-5's ability to surface obscure knowledge showcases its powerful potential as an advanced research assistant.
Nvidia's China Shutdown
Next in AI: Nvidia CEO Jensen Huang revealed in an interview with Citadel Securities that U.S. export controls have wiped out the company's presence in China. The chip giant's market share in the country has plummeted from 95% to virtually zero.
Decoded:
The financial impact is stark, with Nvidia now completely excluding China from its financial forecasts. Any future sales in what Huang calls "the second largest computer market in the world" are now viewed simply as a potential bonus.
This situation stems from U.S. policies restricting the export of advanced AI chips, which prompted Nvidia to develop special export-compliant processors that Chinese regulators have reportedly discouraged local firms from purchasing.
Huang argues the policy is a strategic error, as it prevents roughly half the world's AI researchers in China from building on American technology, which could undermine U.S. leadership in the long run.
Why It Matters: These export controls create a significant vacuum in one of the world's largest tech markets, opening the door for domestic Chinese competitors to accelerate their own chip development. The long-term effects could challenge the global dominance of U.S. semiconductor technology.
The AI Energy Squeeze
Next in AI: The insatiable energy demand from AI data centers is pushing Georgia's power grid to its limits, prompting a nearly $16 billion expansion request that raises alarms about rising utility bills and environmental impacts for residents.
Decoded:
The surge is no surprise, as the Atlanta metro area led the nation in data center construction last year, making it a key hub for AI infrastructure growth.
To meet this demand, the state is turning back to fossil fuels, with plans that include developing three new natural gas turbines to power the new facilities.
Lawmakers are now pushing back with legislation aimed at making data centers shoulder more of the cost to protect residents from footing the bill for AI's energy appetite.
Why It Matters: The AI boom's physical footprint is becoming impossible to ignore, forcing a confrontation between technological progress and its real-world infrastructure costs. This situation in Georgia serves as a critical test case for how communities will balance the promise of AI with the essential needs of their residents.
AI Pulse
OpenAI announced that it will begin allowing erotic content on ChatGPT for verified adults in December, with Sam Altman citing a “treat adults like adults” principle for the policy change.
President Trump posted an AI-generated video to Truth Social depicting him in a fighter jet labeled "King Trump" dropping a substance on protesters following nationwide "No Kings" rallies.
AI-generated content surpassed human-written articles online for the first time, with one analysis finding that 52% of web articles are now machine-written, up from under 10% before 2022.