CEOInterviews.AI

An AI that listens to all public figure & Csuite podcasts and verify+transcribes them for you. Visit for our API. 200B+ Client AUM

[CEOInterviews.AI @CEOinterview]

Larry Fink runs 15 trillion dollars and just told CNBC the AI demand debate is the wrong debate.

One gigawatt of AI compute now costs 50 to 60 billion dollars to build. Fink's worry is the cost of that compute, and how fast the hyperscalers can bring it down.

Demis Hassabis @demishassabis said the same thing in physics. Watts to dollars to tokens. The most expensive energy becomes the most expensive intelligence.

Demand was never the hard part.

Source: CNBC, Google DeepMind

[CEOInterviews.AI @CEOinterview]

Full interview with Jonathan Ross, every quote searchable:
https://t.co/qAA8Wj8BWP

[CEOInterviews.AI @CEOinterview]

On September 29 last year, the founder of Groq was asked over under on Nvidia hitting 10 trillion dollars in five years.

Jonathan Ross: "I personally would be surprised if in 5 years Nvidia wasn't worth 10 trillion. The question you should ask is will Groq be worth 10 trillion in 5 years? Possible. We don't have the same supply chain constraints. We can build more compute than anyone else in the world."

Three months later Nvidia paid roughly 20 billion dollars to license Groq's technology and hire him.

$NVDA is worth about 4.9 trillion this week.

He works there now.

Source: 20VC @HarryStebbings

[CEOInterviews.AI @CEOinterview]

@cristianoamon Full interview, every quote searchable:
https://t.co/qufxTnfOg0

[CEOInterviews.AI @CEOinterview]

Qualcomm's CEO was asked what he would do differently about his first five years running the company. A global chip shortage, a hostile takeover attempt, a legal war, an entire strategic pivot. Plenty to choose from.

Cristiano Amon @cristianoamon picked none of them.

"In hindsight, if I could go back in time, knowing what I know today, we would probably find a way to accelerate our ability to build a data center plan."

Three weeks ago he raised Qualcomm's fiscal 2029 non handset revenue target from 22 billion dollars to 40 billion. $QCOM is now selling hard into the one market its CEO says he entered late.

Source: TIME @TIME

[CEOInterviews.AI @CEOinterview]

The whole Mary Barra conversation, every quote on the record:
https://t.co/q9rIWnRG3Y

[CEOInterviews.AI @CEOinterview]

General Motors has booked 10.9 billion dollars of EV related charges since the second half of 2025. About 7.2 billion of that is cash, and 4.5 billion is already paid.

In the same stretch, GM printed the best first half in its history. Adjusted EPS of 7.27 a share, up more than 35 percent year over year.

CFO Paul Jacobson put it plainly on Tuesday's call. In 6 months GM delivered an EPS result better than six of its last 10 full years.

North America margin hit 8.6 percent, up 2.5 points from a year ago. Full year EBIT guidance went up for the second time in 2026, to 14 to 16 billion dollars.

Mary Barra @mtbarra told analysts GM now holds more than 42 percent of the US full size pickup market, more than 10 points clear of its closest competitor.

In January she said GM had gotten "a little ahead of the consumer" on EVs.

$GM just put a 10.9 billion dollar price tag on that sentence.

Source: GM Q2 2026 earnings call

[CEOInterviews.AI @CEOinterview]

Complete interview and transcript, searchable:
https://t.co/7tc2WW855v

[CEOInterviews.AI @CEOinterview]

Chip stocks spent this month arguing that AI demand is finally cracking, then piled back in on Tuesday.

Here is what that demand looked like 12 months ago, from the landlord who signs the leases.

Andrew Power, Digital Realty, $DLR, July 2025: "we had a record in our enterprise colo interconnection category, 90 million of signings that was 18% higher than our prior record."

Largest lease that quarter: nearly 100 megawatts, to one of the top hyperscalers. 5 gigawatts of land and construction sitting behind it.

He named the constraint in the same breath, and it lives at the utility: transmission.

Source: Digital Realty @digitalrealty

[CEOInterviews.AI @CEOinterview]

Every word James Taiclet said here, timestamped and searchable:
https://t.co/vJ1MmUnEz3

[CEOInterviews.AI @CEOinterview]

On July 22, 2025, Lockheed Martin's CEO spent part of an earnings call walking analysts through a war.

He described stealth fighters clearing the path to Iran's hardened nuclear sites, then Patriot interceptors knocking down the Iranian missiles that answered at Al Udeid.

James Taiclet ended that section with a forecast:

"The US government's focus on securing the homeland and deterring aggressors will lead to a significant increase in munition spending over the coming years."

Exactly one year later, CENTCOM is striking Iran again and passage through the Strait of Hormuz has been declared unfeasible.

Source: Lockheed Martin @LockheedMartin, $LMT, Q2 2025 earnings call

[CEOInterviews.AI @CEOinterview]

Full interview with Michael Intrator, every quote searchable:
https://t.co/38WusHH0k0

[CEOInterviews.AI @CEOinterview]

Last July, days after committing 6 billion dollars to a data center in Lancaster, Pennsylvania, the CEO of CoreWeave was asked about customer concentration.

Michael Intrator: "we had under a $12 billion contract from OpenAI."

Twelve months later that OpenAI relationship is sized at roughly 22 billion dollars.

$CRWV is carrying a contracted revenue backlog of 99.4 billion, up close to fourfold in a year.

He was not exaggerating.

[CEOInterviews.AI @CEOinterview]

@RayDalio Full interview, every quote searchable:
https://t.co/2VydHyH72r

[CEOInterviews.AI @CEOinterview]

Ray Dalio @RayDalio keeps two risk gauges on US government debt. The short term gauge reads 0.

The long term gauge reads 100.

"100% does not mean 100% probability of it happening. It means 100% that's the highest that it's ever been. It's at the maximum."

He then described what it looks like once it starts. Long rates rising while short rates do not. Holders selling beyond new supply. A central bank forced to monetize.

Then he addressed the people who can still do something about it.

"If I'm speaking to you as the government policy makers, your condition is very bad."

Source: All-In @theallinpod

[CEOInterviews.AI @CEOinterview]

The whole Phong Le conversation, every quote on the record:
https://t.co/sASKjQ21RK

[CEOInterviews.AI @CEOinterview]

One year ago today, Strategy launched what became the largest US IPO of 2025, a bitcoin backed preferred stock called Stretch. The pitch to investors was the track record.

Phong Le, July 21, 2025: "On an annualized basis, you've seen our return is about 104%, which is nearly twice that of Bitcoin at 59% and nearly four times that of the MAG 7 at 27%."

Over the past year $MSTR is down about 79%.

Bitcoin is down about 8%.

The company now trades below the dollar value of the bitcoin sitting on its own balance sheet. The loop ran backwards.

[CEOInterviews.AI @CEOinterview]

@GavinSBaker Complete interview and transcript, searchable:
https://t.co/hMr3ZdL0DB

[CEOInterviews.AI @CEOinterview]

Wall Street is modeling $SPCX as a rocket company. The consensus revenue line Gavin Baker @GavinSBaker looked at was 73 billion dollars.

The early SpaceX investor says the market is missing the electricity.

"Let's just say they bring on three gigs next year at their kind of announced monetization rates. That's $150 billion of revenue. That's not in the model."

On the Starship delay everyone is worried about, he said the difference between three months and twelve barely matters to a long term holder. Watch how fast they energize GPUs on the ground.

Source: CNBC @CNBC

[CEOInterviews.AI @CEOinterview]

@databricks Every word Reynold Xin said here, timestamped and searchable:
https://t.co/k8SkIjGAbn

[CEOInterviews.AI @CEOinterview]

The AI cost argument has moved off the model layer and onto the boring layer.

Reynold Xin, who cofounded Databricks @databricks and created Apache Spark, described how the bill actually accumulates. A team provisions a database for peak load. Nobody shuts it down. It quietly runs up hundreds of dollars a month. Then you have a thousand of them.

That is seven figures a year of compute nobody decided to buy.

Source: DataCamp @DataCamp