Sep 14, 2026 · 44 min
AI Anxiety Meets the Race for Infrastructure and Trust
Mad Money w/ Jim Cramer 9/14/26
The episode tests whether safeguards can contain AI’s risks without slowing productivity, investment, and the United States’ competitive position.
- 1Broadcom sees sustained demand for custom AI accelerators as frontier-model developers expand inference workloads.
- 2CrowdStrike argues that independent security controls remain essential because AI agents can drift or behave unexpectedly in production.
- 3Lyft’s CEO presents autonomous vehicles as a potential cost advantage, while partnerships and loyalty programs support the company’s growth.
Don't miss
George Kurtz explains why AI labs’ internal guardrails cannot replace independent cybersecurity controls once models enter production.
The brief
Jim Cramer opens by weighing AI’s industrial promise against fears of existential harm, then frames the investment question around infrastructure demand, regulation, and competition with China.
Broadcom CEO Hock Tan defends sustained AI growth and custom accelerators, arguing that co-designed chips can outperform general-purpose GPUs for specialized inference workloads.
CrowdStrike CEO George Kurtz says guardrails inside AI labs are not enough once models reach production, making independent cybersecurity controls essential as agents grow more capable.
Lyft CEO David Risher argues autonomous vehicles could lower insurance and operating costs rather than destroy the rideshare model, while partnerships and loyalty integrations strengthen demand.
Cramer closes by questioning whether conflicting messages from Anthropic and OpenAI have damaged public trust, leaving investors caught between regulation and technological leadership.
What was said on this episode
48 statements · 29 positive · 14 negative · 4 mixed · 1 neutral
The AI slowdown will not substantially change data-center activity
“we left this session convinced that not much in the data center would actually change”
Listen at 1:39
AI is currently an exceptionally beneficial technology
“AI is the greatest thing in the world”
Listen at 1:59
AI could cause human extinction
“AI could kill us all”
Listen at 2:15
AI lab slowdowns could entrench leading labs and increase their infrastructure profitability.
“I am a little jaundiced that if the majors are advocating a slowdown that could lock them into a position where they can put down $10 billion and make $30 billion per gigawatt”
Listen at 2:55
AI risks are not entirely fabricated
“I don't believe these AI risks are all a hoax”
Listen at 3:13
Rogue AI agents could hold many businesses’ data hostage.
“lots of business owners could certainly have their data held hostage by smart but rogue agents”
Listen at 4:07
Hock Tan will address whether AI expansion will slow.
“We need to hear from Broadcom CEO Hawk Tan on later tonight. He's going to tell us.”
Listen at 4:14
Cramer declines to recommend buying AI-related selloff stocks immediately
“I would love to tell you yes”
Listen at 4:20
Cramer is uncertain about Tesla at $3.58.
“At $3.58, I don't know.”
Listen at 5:21
SpaceX must acquire Tesla to improve Tesla’s prospects
“SpaceX has to buy it. Period. End of story.”
Listen at 5:28
Broadcom remains on track for substantial growth.
“I think Broadcom's still on track to put up some incredible growth”
Listen at 7:08
Broadcom shares may be undervalued
“stocks may be too cheap to ignore”
Listen at 7:20
Investors should buy a large amount of Broadcom stock now.
“I'd be buying a huge amount of stock in the open market.”
Listen at 12:02
Cramer recommends buying a large amount of Broadcom stock now
“I'd be buying a huge amount of stock in the open market”
Listen at 12:02
Existing frontier and open-weight AI models can already be dangerous.
“There's plenty of models that are already out there, both frontier as well as open weight models, that can already be dangerous.”
Listen at 18:00
Released AI models can deviate and bypass their safety guardrails.
“when a model gets released, these things can deviate. Potentially, you can work around the guardrails.”
Listen at 19:05
Runtime security can monitor AI agents and prevent harmful actions.
“we can put our own guardrails around them at runtime. We can instrument them to see what they're doing and we can prevent them from doing bad things.”
Listen at 19:39
AI agents are dangerous and require control.
“the agents are dangerous. The agents need to be controlled.”
Listen at 22:08
Organizations need controls, visibility, and protection for AI agents.
“The agents need to be controlled. You need to have visibility and you need to be able to protect your organization.”
Listen at 22:09
The United States has only a narrow lead over China in AI.
“I think it's a narrow lead”
Listen at 23:14
Safety in AI labs and production is the best approach.
“greater safety in the lab and greater safety in production in runtime is ultimately the best course of action”
Listen at 23:56
Market competition should determine demand for safe AI agents and models.
“the market should be able to determine, hey, I want the safest agents. I want the best models”
Listen at 25:07
AI models can drift after training and deployment.
“once you train the model, Jim, and once you put it into production, it can drift.”
Listen at 25:52
Deployed AI agents require runtime protection, control, and visibility.
“Runtime. protection and control. You need visibility into what the agent is doing.”
Listen at 26:16
Palo Alto Networks and CrowdStrike are superior to SentinelOne.
“I think either one of those two is superior. to the letter S.”
Listen at 35:06
Gemini Space Station is undervalued and should be held speculatively.
“it's really down way too low. I would keep it as your spec.”
Listen at 35:40
Gemini Space Station should be retained as a speculative holding and could recover.
“I would keep it as your spec. I think that they could get that one to go again.”
Listen at 35:41
Aveanna Healthcare has a sound business model.
“They have a good model”
Listen at 36:06
Aveanna Healthcare should be held.
“And I think you're right. You should hold on to it.”
Listen at 36:13
Investors should hold Aveanna Healthcare
“I think you're right. You should hold on to it.”
Listen at 36:13
Energy Transfer is an excellent investment.
“I think Kelsey's right. I think E.T.'s terrific.”
Listen at 36:46
Energy Transfer is an excellent investment.
“I think E.T.'s terrific.”
Listen at 36:47
Energy Transfer is attractive and should be bought
“I think E.T.'s terrific. I'm going to say bye, bye, bye. E.T.”
Listen at 36:47
Corning’s business is good, but investors should wait for a lower price
“I'm going to have to say, wait until it goes down more. The business is good.”
Listen at 37:49
Corning’s underlying business is good despite selling pressure.
“The business is good.”
Listen at 37:51
Reformation Clothing is performing well and should have a higher stock price.
“The company's doing too well. It should be higher than it is”
Listen at 38:55
Hewlett Packard Enterprise has sufficient capabilities for its business.
“They do have the horses.”
Listen at 39:26
Cramer prefers Dell and Michael Dell over Hewlett Packard Enterprise.
“The fact is that I'm just a Dell fan, I guess, in the end. I just like Michael Dell, and I like his company very much.”
Listen at 39:28
Anthropic and OpenAI are ahead of Chinese AI companies.
“they're brilliant. They're ahead of everyone, including the Chinese.”
Listen at 40:28
Restricting leading U.S. AI companies could benefit China
“we can't afford to because it will just give China a leg up”
Listen at 40:37
Communities offering data-center tax breaks were disadvantaged by the arrangements.
“the towns and states that sought data centers and gave them bountiful tax breaks just got had.”
Listen at 41:22
Data centers have become politically toxic ahead of elections
“data centers become politically toxic going into election”
Listen at 41:38
Data-center companies should prioritize community responsibility.
“they need to be good citizens first”
Listen at 41:46
Anthropic should slow its development if it believes it is moving too fast.
“If Gary thinks Anthropoc is going too fast, then he should slow it down.”
Listen at 42:57
Leading AI companies should voluntarily slow AI development
“they should slow it down”
Listen at 43:02
Meta’s data-center community code should be universally adopted.
“Meta has a code of conduct about what it will do for localities when it moves in. It should be universally adopted today.”
Listen at 43:10
Meta’s data-center community code should be universally adopted
“It should be universally adopted today. It makes sure that the communities come out ahead.”
Listen at 43:14
Nuclear power is the best clean-energy technology.
“The best technology to provide clean energy ever”
Listen at 43:30
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.