Did Apple Get AI Spending Right?, Microsoft & OpenAI’s New Reality, Where’s Stargate?

Summary of Did Apple Get AI Spending Right?, Microsoft & OpenAI’s New Reality, Where’s Stargate?

by Alex Kantrowitz

1h 0m•May 5, 2026
Listen to Episode

Overview of Big Technology Podcast

Alex Kantrowitz and MG Siegler dig into three major AI business stories: Apple’s unusually restrained AI spending, Microsoft and OpenAI’s newly loosened partnership, and the changing reality of OpenAI’s Stargate infrastructure plans. The conversation centers on whether Apple’s low CapEx is a smart long-term bet or a dangerous delay, how OpenAI is becoming less dependent on Microsoft, and whether “Stargate” is really a project or just a branding umbrella for OpenAI’s broader data center strategy.

Apple’s AI Spending: Smart Discipline or Dangerous Delay?

The core debate

  • Apple is spending far less on AI-related capital expenditures than its peers.
  • MG’s “binary bet” framing: Apple is either making a shrewd call that AI infrastructure will commoditize, or it’s falling dangerously behind.
  • Apple’s CapEx is roughly $9–10 billion this year, while peers like Meta, Google, Amazon, and Microsoft are spending well over $100 billion each.

Why this matters

  • The rest of Big Tech is treating AI as infrastructure they must own.
  • Apple is acting like AI may be something it can buy later, license, or run on-device instead of building from scratch.
  • This could work if:
    • frontier models commoditize,
    • open-source models catch up,
    • or on-device AI becomes “good enough.”

The risk to Apple

  • If AI becomes the primary interface for computing, Apple could be reduced to a hardware vendor dependent on others’ models.
  • The most bearish scenario for Apple is an AI-first device layer where the model, not the phone, becomes the main product.
  • The most bullish scenario is that the iPhone remains the central AI device, preserving Apple’s power even if it doesn’t own the frontier models.

The Tim Cook → John Ternus transition

  • Ternus is signaling a possible shift in Apple’s financial posture:
    • more flexibility on buybacks/dividends,
    • more room for R&D,
    • possibly larger acquisitions,
    • and potentially higher AI-related CapEx.
  • MG reads this as Apple building optionality for a more aggressive AI era.

Apple’s New Signal: More Flexibility, More Spending?

What changed

  • Apple’s leadership transition appears to be accelerating.
  • The company is no longer emphasizing strict “cash neutrality” in the same way Tim Cook once did.
  • That suggests Apple may be preparing for:
    • bigger M&A,
    • heavier R&D,
    • and possibly more AI infrastructure investment.

Why it matters

  • Apple has historically been reluctant to make massive bets.
  • Ternus’s early messaging suggests the next era may be more aggressive and more AI-focused.
  • The transcript suggests Apple’s coming Siri overhaul will be an important test of whether the company’s restraint was wisdom or weakness.

Microsoft & OpenAI: A More Mature, Less Romantic Partnership

What changed in the new agreement

  • Microsoft and OpenAI removed the AGI clause that had loomed over their relationship.
  • OpenAI can now work with any cloud provider.
  • OpenAI quickly announced availability on AWS, and Google Cloud seems likely next.

What the deal really means

  • The partnership is becoming more pragmatic and less ideologically tied to exclusivity.
  • Microsoft still keeps important advantages:
    • access to OpenAI IP through 2032,
    • OpenAI products still ship first on Azure unless Microsoft can’t or won’t support them,
    • and Microsoft still benefits from OpenAI’s growth through its equity stake.

The bigger strategic shift

  • OpenAI is no longer locked into Microsoft’s cloud.
  • This makes OpenAI more like Anthropic, which can sell across multiple clouds.
  • It also reflects Microsoft’s willingness to prioritize equity value over strict exclusivity.

Key takeaway

  • Microsoft may be saying, in effect:
    “Grow as big as possible, even if you work with our competitors.”
  • But the relationship is still laced with tension, and future disputes over model access, exclusivity, and launch priority seem inevitable.

Stargate: From Grand Announcement to Umbrella Brand

What Stargate was supposed to be

  • Stargate was announced as a $500 billion AI infrastructure initiative with major fanfare, including a Trump White House appearance.
  • The early vision sounded like a huge coordinated buildout of AI data centers.

What it has become

  • In practice, Stargate now looks more like a catch-all term for OpenAI’s infrastructure ambitions.
  • The original idea of a neatly packaged, massive centralized project has given way to a looser, more fragmented reality.
  • OpenAI increasingly appears to be securing capacity through partners rather than directly building everything itself.

Why the strategy shifted

  • OpenAI likely realized that owning infrastructure is necessary to compete with Google and other vertically integrated players.
  • But its financing model and cash burn make direct ownership difficult.
  • So instead, it has leaned on:
    • Oracle,
    • SoftBank,
    • NVIDIA,
    • and various neocloud partners.

Risk in the new setup

  • A lot of the infrastructure burden is now being pushed onto partners.
  • That makes companies like CoreWeave, Nebius, Oracle, and others more exposed to OpenAI’s execution and funding cadence.
  • The market reaction to reports of OpenAI slowing its buildout suggests investors are increasingly sensitive to that risk.

Big Themes and Takeaways

1. AI spending is becoming a strategic identity test

  • Big Tech is splitting into:
    • companies trying to own the whole AI stack,
    • and companies trying to wait, partner, or leverage existing hardware/software advantages.

2. Apple may be playing the long game

  • If AI models commoditize and on-device AI becomes enough, Apple’s restraint could look brilliant.
  • If AI becomes a new computing layer controlled by model makers, Apple could be in trouble.

3. OpenAI is becoming less dependent, but not less complicated

  • OpenAI is more flexible than before, but its business model is still tangled with cloud providers, infrastructure partners, and revenue-sharing arrangements.

4. Infrastructure is the real battleground

  • The conversation keeps returning to who controls compute, chips, power, and data centers.
  • That’s where the moat may ultimately be formed.

Notable Closing Thoughts

  • The hosts agree the Apple transition to John Ternus is real and meaningful.
  • They also agree that Apple’s Siri revamp will be a major near-term test.
  • The discussion ends with the sense that:
    • Apple may be preparing to spend more,
    • Microsoft and OpenAI are entering a more transactional phase,
    • and Stargate is less a single project than a symbol of the broader AI infrastructure scramble.