Jul 1, 2025 · 20m · tbpn

Apple Considers New AI partnership

John Coogan · 14m spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

The podcast hosts analyze reports that Apple is exploring partnerships with Anthropic and OpenAI to revamp Siri, examining the strategic shift toward third-party model licensing, internal talent retention challenges, and the organizational hurdles Apple faces in the generative AI race.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 78.8% of the talking time here. How this is scored →

The hosts as informed peer 4.7 Guest teaching 2.7 Guest disagreement 1.7 The hosts pushing back 0.7
05100:0010:0020:000:00–3:01 · The hosts as informed peer 6/10 Apple Explores Anthropic and OpenAI Partnerships for Siri Coogan breaks down Apple's private cloud infrastructure limitations and explains hardware constraints regarding large model inference compared to Meta's data centers. Jordy chimes in briefly and agreeably about Nvidia chips. The tone is highly collaborative and analytical.3:01–8:23 · The hosts as informed peer 5/10 Comparing Siri, Gemini on Android, and ChatGPT User Experience Coogan details the user experience difference between voice dictation in ChatGPT versus Gemini on iOS. Jordy playfully points out that Apple rarely works with early pre-seed startups and emphasizes compensation disparity challenges for top researchers.8:23–11:52 · The hosts as informed peer 5/10 Startup M&A Deliberations and the Fast-Paced Model Race Coogan speculates on Apple or Meta acquiring Thinking Machines or SSI, while Jordy pushes back on the plausibility of startup founders accepting lower corporate compensation than top AI researchers. Coogan counters by citing the structure of the Beats acquisition earnout.11:52–14:06 · The hosts as informed peer 3/10 Historical Apple Acquisitions and Third-Party AI Precedents Jordy educates the discussion by pulling revenue multiples for Apple's historical acquisitions of Beats, Intel's modem business, and Shazam to illustrate Apple's aversion to massive talent acqui-hires. Coogan reacts to the revenue multiple data in real time.14:06–18:21 · The hosts as informed peer 5/10 Foundation Model Team Morale and the Apple AI Identity Both hosts riff collaboratively on how Apple could brand its in-house foundation models and maintain its historical identity of delightful product design. Coogan connects this to Steve Jobs' philosophy of empowering creativity through tools like GarageBand.18:21–20:34 · The hosts as informed peer 4/10 Sponsor Break: Vanta Trust Management and Compliance After an ad transition for Vanta, the hosts discuss Bloomberg reports on recent Apple researcher departures and retention battles around the MLX team. Jordy highlights Meta's massive 10 to 40 million dollar annual compensation packages.0:00–3:01 · Guest teaching 1/10 Apple Explores Anthropic and OpenAI Partnerships for Siri Coogan breaks down Apple's private cloud infrastructure limitations and explains hardware constraints regarding large model inference compared to Meta's data centers. Jordy chimes in briefly and agreeably about Nvidia chips. The tone is highly collaborative and analytical.3:01–8:23 · Guest teaching 2/10 Comparing Siri, Gemini on Android, and ChatGPT User Experience Coogan details the user experience difference between voice dictation in ChatGPT versus Gemini on iOS. Jordy playfully points out that Apple rarely works with early pre-seed startups and emphasizes compensation disparity challenges for top researchers.8:23–11:52 · Guest teaching 3/10 Startup M&A Deliberations and the Fast-Paced Model Race Coogan speculates on Apple or Meta acquiring Thinking Machines or SSI, while Jordy pushes back on the plausibility of startup founders accepting lower corporate compensation than top AI researchers. Coogan counters by citing the structure of the Beats acquisition earnout.11:52–14:06 · Guest teaching 6/10 Historical Apple Acquisitions and Third-Party AI Precedents Jordy educates the discussion by pulling revenue multiples for Apple's historical acquisitions of Beats, Intel's modem business, and Shazam to illustrate Apple's aversion to massive talent acqui-hires. Coogan reacts to the revenue multiple data in real time.14:06–18:21 · Guest teaching 1/10 Foundation Model Team Morale and the Apple AI Identity Both hosts riff collaboratively on how Apple could brand its in-house foundation models and maintain its historical identity of delightful product design. Coogan connects this to Steve Jobs' philosophy of empowering creativity through tools like GarageBand.18:21–20:34 · Guest teaching 3/10 Sponsor Break: Vanta Trust Management and Compliance After an ad transition for Vanta, the hosts discuss Bloomberg reports on recent Apple researcher departures and retention battles around the MLX team. Jordy highlights Meta's massive 10 to 40 million dollar annual compensation packages.0:00–3:01 · Guest disagreement 1/10 Apple Explores Anthropic and OpenAI Partnerships for Siri Coogan breaks down Apple's private cloud infrastructure limitations and explains hardware constraints regarding large model inference compared to Meta's data centers. Jordy chimes in briefly and agreeably about Nvidia chips. The tone is highly collaborative and analytical.3:01–8:23 · Guest disagreement 2/10 Comparing Siri, Gemini on Android, and ChatGPT User Experience Coogan details the user experience difference between voice dictation in ChatGPT versus Gemini on iOS. Jordy playfully points out that Apple rarely works with early pre-seed startups and emphasizes compensation disparity challenges for top researchers.8:23–11:52 · Guest disagreement 3/10 Startup M&A Deliberations and the Fast-Paced Model Race Coogan speculates on Apple or Meta acquiring Thinking Machines or SSI, while Jordy pushes back on the plausibility of startup founders accepting lower corporate compensation than top AI researchers. Coogan counters by citing the structure of the Beats acquisition earnout.11:52–14:06 · Guest disagreement 2/10 Historical Apple Acquisitions and Third-Party AI Precedents Jordy educates the discussion by pulling revenue multiples for Apple's historical acquisitions of Beats, Intel's modem business, and Shazam to illustrate Apple's aversion to massive talent acqui-hires. Coogan reacts to the revenue multiple data in real time.14:06–18:21 · Guest disagreement 1/10 Foundation Model Team Morale and the Apple AI Identity Both hosts riff collaboratively on how Apple could brand its in-house foundation models and maintain its historical identity of delightful product design. Coogan connects this to Steve Jobs' philosophy of empowering creativity through tools like GarageBand.18:21–20:34 · Guest disagreement 1/10 Sponsor Break: Vanta Trust Management and Compliance After an ad transition for Vanta, the hosts discuss Bloomberg reports on recent Apple researcher departures and retention battles around the MLX team. Jordy highlights Meta's massive 10 to 40 million dollar annual compensation packages.0:00–3:01 · The hosts pushing back 0/10 Apple Explores Anthropic and OpenAI Partnerships for Siri Coogan breaks down Apple's private cloud infrastructure limitations and explains hardware constraints regarding large model inference compared to Meta's data centers. Jordy chimes in briefly and agreeably about Nvidia chips. The tone is highly collaborative and analytical.3:01–8:23 · The hosts pushing back 1/10 Comparing Siri, Gemini on Android, and ChatGPT User Experience Coogan details the user experience difference between voice dictation in ChatGPT versus Gemini on iOS. Jordy playfully points out that Apple rarely works with early pre-seed startups and emphasizes compensation disparity challenges for top researchers.8:23–11:52 · The hosts pushing back 2/10 Startup M&A Deliberations and the Fast-Paced Model Race Coogan speculates on Apple or Meta acquiring Thinking Machines or SSI, while Jordy pushes back on the plausibility of startup founders accepting lower corporate compensation than top AI researchers. Coogan counters by citing the structure of the Beats acquisition earnout.11:52–14:06 · The hosts pushing back 1/10 Historical Apple Acquisitions and Third-Party AI Precedents Jordy educates the discussion by pulling revenue multiples for Apple's historical acquisitions of Beats, Intel's modem business, and Shazam to illustrate Apple's aversion to massive talent acqui-hires. Coogan reacts to the revenue multiple data in real time.14:06–18:21 · The hosts pushing back 0/10 Foundation Model Team Morale and the Apple AI Identity Both hosts riff collaboratively on how Apple could brand its in-house foundation models and maintain its historical identity of delightful product design. Coogan connects this to Steve Jobs' philosophy of empowering creativity through tools like GarageBand.18:21–20:34 · The hosts pushing back 0/10 Sponsor Break: Vanta Trust Management and Compliance After an ad transition for Vanta, the hosts discuss Bloomberg reports on recent Apple researcher departures and retention battles around the MLX team. Jordy highlights Meta's massive 10 to 40 million dollar annual compensation packages.

speaking balance: gold is the hosts, purple is the guest (3 minute bins)

0:00 · the hosts 97.1% · guest 2.9%0:00 · the hosts 97.1% · guest 2.9%3:00 · the hosts 95.8% · guest 4.2%3:00 · the hosts 95.8% · guest 4.2%6:00 · the hosts 74.3% · guest 25.7%6:00 · the hosts 74.3% · guest 25.7%9:00 · the hosts 69.5% · guest 30.5%9:00 · the hosts 69.5% · guest 30.5%12:00 · the hosts 76.7% · guest 23.3%12:00 · the hosts 76.7% · guest 23.3%15:00 · the hosts 71.1% · guest 28.9%15:00 · the hosts 71.1% · guest 28.9%18:00 · the hosts 64.1% · guest 35.9%18:00 · the hosts 64.1% · guest 35.9%
Sharpest disagreement ▶ 10:19 Mira Murati compensation pushback

Jordy directly challenges Coogan's acqui-hire premise by asking whether Mira Murati would ever accept an acquisition payout that results in lower annual comp than average AI researchers.

Hardest push from the hosts ▶ 10:35 Coogan counters with Beats acquisition precedent

Coogan rejects Jordy's compensation constraint argument by pointing out that acquisition earnouts historically allowed executives at Beats to earn more than the Apple CEO.

Biggest teaching moment ▶ 11:52 Historical acquisition multiples breakdown

Jordy walks through specific historical transaction figures for Beats and Intel's modem division to prove Apple has virtually no precedent of doing massive pre-revenue talent acqui-hires.

The host holds their own ▶ 0:52 Hardware architecture and inference analysis

Coogan demonstrates deep domain knowledge by detailing how Apple's TSMC relationship and data center infrastructure differ fundamentally from Meta's large-scale recommendation and inference clusters.

the scores for every segment, with the reasoning behind each
ChapterTopicThe hosts as informed peerGuest teachingGuest disagreementThe hosts pushing backWhy
Apple Explores Anthropic and OpenAI Partnerships for Siri 6110 Coogan breaks down Apple's private cloud infrastructure limitations and explains hardware constraints regarding large model inference compared to Meta's data centers. Jordy chimes in briefly and agreeably about Nvidia chips. The tone is highly collaborative and analytical.
Comparing Siri, Gemini on Android, and ChatGPT User Experience 5221 Coogan details the user experience difference between voice dictation in ChatGPT versus Gemini on iOS. Jordy playfully points out that Apple rarely works with early pre-seed startups and emphasizes compensation disparity challenges for top researchers.
Startup M&A Deliberations and the Fast-Paced Model Race 5332 Coogan speculates on Apple or Meta acquiring Thinking Machines or SSI, while Jordy pushes back on the plausibility of startup founders accepting lower corporate compensation than top AI researchers. Coogan counters by citing the structure of the Beats acquisition earnout.
Historical Apple Acquisitions and Third-Party AI Precedents 3621 Jordy educates the discussion by pulling revenue multiples for Apple's historical acquisitions of Beats, Intel's modem business, and Shazam to illustrate Apple's aversion to massive talent acqui-hires. Coogan reacts to the revenue multiple data in real time.
Foundation Model Team Morale and the Apple AI Identity 5110 Both hosts riff collaboratively on how Apple could brand its in-house foundation models and maintain its historical identity of delightful product design. Coogan connects this to Steve Jobs' philosophy of empowering creativity through tools like GarageBand.
Sponsor Break: Vanta Trust Management and Compliance 4310 After an ad transition for Vanta, the hosts discuss Bloomberg reports on recent Apple researcher departures and retention battles around the MLX team. Jordy highlights Meta's massive 10 to 40 million dollar annual compensation packages.

Statements from this episode (8)

Opinion
Coogan: Apple could produce custom data center AI chips with TSMC
“Apple has enough of a relationship with TSMC that they could go and do kind of like, ah, you know, the A three, the A, or the, ah, the Apple Silicon, you know, the M two chip, the M three chip, like the M four chip, they could do a different run with TSMC that…”
John Coogan Jul 1, 2025 ▶ 0:57
Assertion Not checkable as stated
Coogan: Apple cloud infrastructure lacks Meta's scale in ML inference
“They don't do a lot of machine learning inference versus, say, Meta, that already has a, you know, a hundred billion dollar data center that just does recommendation algorithms and machine learning and inference.”
John Coogan Jul 1, 2025 ▶ 1:31
Opinion
Coogan: $2 Billion Is Enough to Train a Near-Frontier AI Model
“I think the fact of the matter is like two billion dollars is enough to do a GPT four level run to do a deep seek style run. And so you, if you take all the best practices from deep seek V three and R one, like the reasoning model, and you spend the money to g…”
John Coogan Jul 1, 2025 ▶ 7:02
Assertion Open · timeframe Dec 2026
Coogan: Apple approved a multi-billion dollar 2026 cloud AI model budget
“Apple has already improved, approved a multi-billion dollar budget for two, 2026 for running its own models via the cloud, but plans For that beyond remain, for beyond that remain murky.”
John Coogan Jul 1, 2025 ▶ 11:01
Opinion
Coogan: Apple struggles with software requiring rapid, continuous updates
“Apple is not great at, ah, at software that needs to be updated, you know, 25 times a year.”
John Coogan Jul 1, 2025 ▶ 11:28
Assertion Supported
Senior LLM researcher Tom Gunter departs Apple after eight years
“One of Apple's most senior large language model researchers, Tom Gunter, left last week. He had worked at Apple for about eight years”
John Coogan Jul 1, 2025 ▶ 18:44
Assertion Supported
Apple made counteroffers to retain MLX team after departure threats
“Apple this month also nearly lost the team behind MLX, It's key open source system for developing machine learning models on the, on, on the latest Apple chips. After the engineers threatened to leave, Apple made counter offers to retain them, and they're stay…”
John Coogan Jul 1, 2025 ▶ 18:59
Assertion Supported
Apple requested custom AI models from OpenAI and Anthropic for Private Cloud
“In his discussions with both Anthropic and OpenAI, the iPhone maker requested a custom version that could run on Apple's private cloud computer.”
John Coogan Jul 1, 2025 ▶ 19:18
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 500 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.