Sep 16, 2026 · 32 min
AI autonomy collides with safety, funding and geopolitical pressure
AI Fears Mount on CEO Calls for Slowdown
The episode connects AI’s race for scale with unresolved questions about oversight, investment returns and the wider economic consequences of technological change.
- 1AI developers face pressure to slow down even as U.S. officials prioritize maintaining an advantage over China.
- 2Agentic systems could execute complex workflows autonomously, increasing the need for auditing, monitoring and accountability.
- 3Vehicle growth, robotaxis and stressed commercial property debt reveal how technology and higher capital costs are reshaping markets.
Don't miss
Mandeep Singh’s explanation of agentic AI shows why systems that execute many steps autonomously create a sharper oversight problem than familiar AI tools.
The brief
Matthew Bloxham frames the lead tension: the AI industry is debating a slower pace while the United States seeks to preserve its advantage over China, making cooperation unlikely.
OpenAI’s delayed IPO and rapid cash burn sharpen the funding question, while Mandeep Singh explains why agentic systems require oversight as they execute increasingly complex workflows.
The conversation then turns to electric vehicles and robotaxis, comparing regional markets and deployments in the United States and China while testing how close autonomy is to broad availability.
Laura Rapoport describes commercial real estate’s refinancing problem, CPACE’s role in restructuring capital stacks, and the difficult economics of converting New York offices into housing.
The episode’s through line is capital meeting uncertainty: AI must prove its returns and governability, while transport and property markets absorb technological and financial change.
What was said on this episode
47 statements · 21 positive · 20 negative · 1 mixed · 5 neutral
U.S.-China cooperation on AI is viewed as very unlikely.
“I think it's seen as a kind of very low probability outcome”
Listen at 4:17
U.S.-China cooperation on AI is viewed as a very low-probability outcome.
“I think it's seen as a kind of very low probability outcome.”
Listen at 4:17
Reaching AI-related agreement between the U.S. and China would be extremely difficult.
“finding agreement between, say, the US and China would be incredibly difficult”
Listen at 4:43
China has near-dominant control of the rare-earths industry.
“China has, you know, a stranglehold almost over the rare earths industry”
Listen at 5:02
AI companies are consuming cash at a very fast rate.
“these are companies that are burning through cash at a very fast rate”
Listen at 5:55
AI companies probably need additional funding.
“they probably do need more money”
Listen at 6:01
SoftBank may increase its exposure to delayed-IPO AI companies.
“SoftBank that may be seeing that they can perhaps increase their exposure a bit”
Listen at 6:08
AI companies delaying IPOs will need external funding to sustain investment.
“if they're delaying the IPO, they'll need money from somewhere to kind of sustain the investments that they'll be making”
Listen at 6:19
Delayed-IPO AI companies will need funding to sustain planned investments.
“they'll need money from somewhere to kind of sustain the investments that they'll be making”
Listen at 6:21
AI coding agents and customer-service applications have been highly successful.
“everyone sees the use cases of AI, the coding agents, the customer service. Those have been hugely successful”
Listen at 9:51
Agentic AI is a distinct emerging aspect of artificial intelligence.
“The second aspect of this is agentic AI.”
Listen at 10:05
Goal-seeking AI agents can perform many actions users do not understand.
“they can do so many things that people have no clue about with the pursuit of that goal”
Listen at 11:03
Top surveyed enterprises increased AI token budgets two- to fivefold in twelve months.
“enterprises, the top 100 enterprises we had in our survey have increased their token budgets by two to five times over the span of the last 12 months”
Listen at 11:30
Top surveyed enterprises increased AI token budgets two- to fivefold in twelve months.
“the top 100 enterprises we had in our survey have increased their token budgets by two to five times over the span of the last 12 months”
Listen at 11:31
AI models will eventually require regulations for evaluation, auditing, and monitoring.
“at some point there has to be regulations around you know how do you evaluate these models how do you audit and monitor these models”
Listen at 12:29
Goal-seeking AI agents can potentially break any software system.
“if the agents are goal seeking, which is what we saw in the hugging face incident, they can go to any length to break any software system”
Listen at 13:26
Goal-seeking AI agents may be able to break software systems.
“they can go to any length to break any software system”
Listen at 13:30
Agentic AI deployments require greater verifiability and auditability.
“you've got to have more verifiability and auditability”
Listen at 14:15
Agentic AI systems require greater verifiability and auditability.
“You've got to have more verifiability and auditability.”
Listen at 14:15
Regulators should demand stronger controls for agentic AI.
“the regulators have to come in and ask for more”
Listen at 14:20
SUVs are currently extremely popular among vehicle buyers.
“SUVs are incredibly popular”
Listen at 18:31
Hybrids are experiencing renewed popularity in the United States.
“hybrids having a kind of a resurgence in popularity in the United States”
Listen at 18:34
Passenger-vehicle markets in India and Southeast Asia are growing rapidly from low motorization levels.
“markets like India and Southeast Asia, which are kind of growing rapidly from very low basis of motorization”
Listen at 18:45
Robotaxis are increasingly becoming commercially viable businesses.
“robotaxis and those are increasingly becoming real businesses”
Listen at 19:17
The United States leads the robotaxi market.
“The U.S. is a leader in this space”
Listen at 19:49
Several Chinese cities have fully operational commercial robotaxi deployments.
“in China, where there's kind of a handful of cities with what we'd classify as fully operational commercial deployments”
Listen at 19:57
China has several cities with fully operational commercial robotaxi deployments.
“there's kind of a handful of cities with what we'd classify as fully operational commercial deployments”
Listen at 19:59
The United States and China currently have the leading robotaxi operations.
“those are the two markets that have leading operations at the moment”
Listen at 20:21
Europe is a growing robotaxi market but remains at an earlier deployment stage.
“Europe is also an active growth market for this technology, but that's at an even earlier stage”
Listen at 20:34
One trillion dollars of commercial real-estate debt is maturing.
“there's a trillion dollars of commercial real estate debt that is maturing”
Listen at 24:10
Unstabilized, delayed, or unpopular projects will struggle to refinance.
“other projects that may not be stabilized yet that have had cost overruns and delays or are not in favor are going to have a really hard time refinancing”
Listen at 24:44
Higher financing costs will reduce the number of new real-estate projects started.
“fewer projects are going to get started”
Listen at 25:24
Higher financing costs will reduce new commercial real-estate project starts.
“fewer projects are going to get started.”
Listen at 25:24
CPACE financing can fund refinancing, rescue capital, and new construction.
“CPACE is long-term sustainable finance that can be used for refi and rescue capital, as well as new construction.”
Listen at 25:39
CPACE financing can provide projects with long-duration repayment terms.
“we can put money into projects, create a long duration for us to be paid back”
Listen at 26:31
New York City has 74 office-to-residential conversion starts representing 10,000 units this year.
“there are 74 starts this year alone. In New York City, which represents 10,000 units”
Listen at 27:52
New York City currently has 122 office-to-residential conversion projects underway.
“there's 122 projects in New York City going on right now”
Listen at 28:01
Office-to-residential conversions in New York City will continue growing.
“one that I think you'll see and continue to grow because New York City is very supply constrained”
Listen at 28:06
Office-to-residential conversions in New York City will continue growing.
“one that I think you'll see and continue to grow”
Listen at 28:06
Office-to-residential conversion is an effective way to create housing stock.
“this is a very effective way to create housing stock”
Listen at 28:15
Office-to-residential conversion is an effective way to create needed affordable and market-rate housing.
“this is a very effective way to create housing stock, not only affordable, but also market rate housing that's much needed”
Listen at 28:15
Office-conversion projects will face more scrutiny, take longer, and be harder to execute well.
“What I do believe we'll see is more scrutiny over these projects, more scrutiny from the DOB, potentially take longer. Fewer people will be able to execute them well.”
Listen at 28:24
Conversion projects will face greater scrutiny, longer timelines, and fewer capable executors.
“more scrutiny over these projects, more scrutiny from the DOB, potentially take longer. Fewer people will be able to execute them well.”
Listen at 28:25
Higher capital costs reduce project profitability.
“the cost of capital has increased, which really reduces the overall profitability of projects”
Listen at 29:08
Higher capital costs will reduce office-conversion project profitability.
“the profitability is going to be impacted because the cost of capital is now so expensive”
Listen at 29:44
Higher construction capital costs will reduce completed project volume.
“there's gonna be fewer projects actually getting done because the cost of capital to build these new projects is now increased”
Listen at 29:55
Higher capital costs will result in fewer conversion projects being completed.
“there's gonna be fewer projects actually getting done”
Listen at 29:55
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.