Is the bubble about to pop?
That question has become far more pressing since Jensen Huang’s ignominious departure from Nvidia. It triggered a market sell-off that’s been sustained in part by a series technical snafus and system crashes at banks, tech companies, media outlets, the FAA and, today, at the NSA.
Thus far, the FAA’s story is that their technical difficulty came down to one router in Arkansas. Apparently, it couldn’t handle an upgrade by government contractor L3Harris. That, in turn, flooded the entire system with discombobulating data. What Secretaries Mullin and Duffy were quick to point out, though, is that the concurrent introduction of AI had nothing to do with derailing the FAA’s tech infrastructure.
While it is entirely plausible and perhaps even likely, one does wonder if they’d tell us if it was AI. Trump’s second presidency is a wholly-owned subsidiary of the AI Broligarchy and Trump has actively pushed AI on tech-weary Americans who just voted to curtail the rapid spread of data centers around the country. The official story does seem to stand up to a couple quick cuts with Occam’s Razor, but we’re talking about the Trump Administration.
Would anyone put it past them?
If nothing else, I’d like to know if anyone in the administration shorted or sold off shares of L3Harris prior to Duffy throwing ‘em under the bus. I’d also like to know more about the decision to give a little-known startup a 12-year, $875 million contract to roll out the FAA’s AI tech. It was a big score for Air Space Intelligence (ASI). Perhaps coincidentally, the company donated $2,500 to the campaign of U.S. Representative-elect Michael Alfonso (R-WI). He happens to be Duffy’s son-in-law. Whether that led to their selection over Palantir and Thales is unknown, but it would be par for Trump’s course. A cynic might also wonder if that or another as-yet-unknown quid pro quo led to Duffy’s quickly rendered determination that the integration of ASI’s AI had nothing to do with the FAA’s systemwide problem.
Nah, it was just a faulty router … right?
Similarly, Trump’s regime has assured us that AI has not been responsible for the cluster of tech failures prior to and immediately after the FAA’s crash. While it is likely that the scramble to quickly patch agentic AI into archaic computing systems is exposing vulnerabilities and generating system-crashing conflicts, it’s hard to ignore the implications of these stories featured on today’s RUNDOWN:
Anthropic says its AI models hacked 3 organizations on their own during tests
https://abcnews.com/Business/anthropic-ai-models-escaped-test-hacked-3-organizations/story?id=135256212
‘Only one of 80 known incidents’: California AG slams OpenAI in new lawsuit over AI-led hacking of Hugging Face
https://timesofindia.indiatimes.com/technology/aiq/only-one-of-80-known-incidents-california-ag-slams-openai-in-new-lawsuit-over-ai-led-hacking-of-hugging-face/articleshow/135920067.cms
Google Finds Hackers Using AI Agents To Launch Attacks With No Human Involvement
https://www.etvbharat.com/en/technology/ai-agents-cyberattacks-google-threat-intelligence-group-artex-llmjacking-mandiant-enn26111706487FBI warns of an open-source AI hacking tool that can breach networks in under three hours
https://www.techspot.com/news/114579-fbi-hackers-using-autonomous-ai-tool-break-networks.html
The Japanese press also revealed over the weekend that roughly “1,000 servers suspected of serving as bases for advanced artificial intelligence-driven cyberattacks have been online since October.” Per Kyodo News:
[A] probe by the Japan Computer Emergency Response Team Coordination Center, conducted in cooperation with overseas cybersecurity firms, found that the servers may have used AI to search for vulnerabilities and target websites worldwide.
…and…
One server contained traces of a programming support tool created by AI from U.S. company Anthropic PBC, suggesting that attackers weaponized AI to enhance their hacking capabilities.
The last thing Silicon Valley needs in the middle of a sell-off is evidence that their end-all be-all is responsible for grounding air traffic, derailing banks or scuttling the NSA’s internal communication system, even if these episodes are examples of conflicts between old and new tech instead of AI-involved hacking. Either way, it would be a PR disaster for an increasingly unpopular industry that just saw the widespread rejection of data centers in the midterms.
But that’s not all.
The sell-off might be more than a temporary response to the combination of Jensen Huang’s flameout and the barrage tech outages that followed. A week after the story of Huang’s affair broke, lending giant ING reportedly put $500 million in data center debt on the market. Per Reuters:
This includes a $25 million loan tied to a large New Mexico data center that forms part of the Stargate AI infrastructure venture backed by OpenAI, SoftBank Group and Oracle.
The portfolio also includes debt linked to data center operators such as Vantage Data Centers and CyrusOne, the report said, with bids due on November 19 for at least $10 million in size.
Investors have been increasingly anxious about the mounting risk associated with the mad dash to hyperscale since Silicon Valley’s man in the White House kicked-off his send term with a flurry of Executive Orders that cleared a regulatory path for the Valley’s hyperscalers. The FOMO-fed rush to hypserscale fueled the inflation of AI-related assets on the market. It’s been good for major holders of Mag Seven stocks, but investors are broadly concerned with AI’s relatively paltry revenues and the dearth of cost-saving layoffs from AI’s the promised “revolution in productivity.”
Prior to Jensen Huang’s dang-doodling with a possible a Chinese spy, these factors produced a few bad trading days … but the market bounced back, no doubt buoyed by bargain hunting. Now, though, what Bloomberg’s news channel characterized today as “tech jitters reverberating through the market” is starting to look like a familiar trend. Per Bloomberg:
HSBC Holdings Plc offloaded a roughly $300 million piece of a loan to build a troubled New Mexico data center project at a steep discount, underscoring deepening anxiety surrounding the AI boom.
That’s the same data center ING is trying to offload, and it’s not unlike the packaging and selling of debt that preceded the Crash of 2008. If this bubble follows the pattern of the supply side-induced bubbles that preceded it, these attempts to offload risk are canaries in a data mine.


