Bloomberg Tech
Bloomberg Tech

Aug 14, 2026 · 54 min

AI’s growth story collides with compute costs and bubble fears

OpenAI’s Explosive Growth Continues

The episode tests whether extraordinary AI revenue and investment can produce durable returns while reshaping labor, media, and technology markets.

3 key takeaways
  1. 1OpenAI’s accelerating revenue depends on scarce, extraordinarily expensive computing power, complicating comparisons with Anthropic and other AI firms.
  2. 2Investors debate whether semiconductor and hyperscaler valuations reflect durable demand, circular financing, or concentration risk among major AI customers.
  3. 3AI’s expansion reaches beyond models into robotics, delivery work, venture capital, and podcast media businesses facing difficult scaling economics.

Don't miss

The robotics discussion asks whether machines are approaching a ChatGPT moment while confronting the practical limits of humanoid designs.

The brief

OpenAI’s consumer and business revenue is accelerating, with Codex contributing to growth, but the discussion stresses that computing power remains scarce and enormously expensive.

The central investment question is whether semiconductor and hyperscaler valuations anticipate durable returns or a bubble built on long horizons, circular financing, and concentrated customers.

Google DeepMind’s Carolina Parada examines whether robotics is nearing a ChatGPT moment, weighing robots’ environmental understanding against the practical advantages of specialized machines.

Kate Bronfenbrenner explains how New York’s proposed Delivery Protection Act could affect Amazon’s delivery partners, covering pay, safety, liability, and possible operational workarounds.

The episode widens the lens to AI venture capital, where a new $100 million fund is targeting agents, robotics, infrastructure, and applications that perform real jobs.

Its final business lesson comes from Alex Cooper’s Unwell: a $500 million valuation and beverage ambitions show how difficult it is to scale a podcast-led media brand.

What was said on this episode

28 statements · 17 positive · 7 negative · 1 mixed · 3 neutral

  1. Rachel Metzon AnthropicPositive3:58

    Anthropic has outpaced OpenAI in coding capabilities in some respects.

    “on the coding front, they in some ways really outpaced OpenAI.”

    Listen at 3:58

  2. Rachel Metzon OpenAI compute costsNegative4:29

    Compute is a major expense for OpenAI.

    “Compute is really expensive. costs a lot.”

    Listen at 4:29

  3. Anna Rathbunon AI infrastructure investment returnsNegative6:05

    AI infrastructure returns may arrive later than investors expect.

    “we definitely have a mismatch in expectations of when we might actually see return on some of these investments.”

    Listen at 6:05

  4. Anna Rathbunon AI infrastructure investment returnsNeutral6:54

    The timing of AI infrastructure returns is unknown.

    “The answer, I think, is we don't really know.”

    Listen at 6:54

  5. Anna Rathbunon AI investment returnsPositive8:02

    AI investments will eventually generate returns.

    “I do believe that we will see the ROI.”

    Listen at 8:02

  6. Major AI companies are increasingly developing their own chips.

    “Now everyone is making their own chips.”

    Listen at 9:00

  7. AI companies need to reduce customer concentration risk.

    “concentration risk is something that these companies need to work on.”

    Listen at 9:42

  8. Anna Rathbunon Hyperscalers’ supplier dependencePositive10:29

    Hyperscalers are trying to reduce dependence on their leading supplier.

    “we have the hyperscalers trying to reduce their dependence on the number one supplier.”

    Listen at 10:29

  9. Ian Kingon Semiconductor industry growth projectionsMixed14:05

    Current semiconductor growth projections are unprecedented.

    “we've never seen anything like this.”

    Listen at 14:05

  10. China’s AI firms could benefit from faster domestic product adoption.

    “Chinese firms could have an advantage in having that more fertile ground to test new products and to allow much faster”

    Listen at 16:52

  11. Ian Kingon Delivery service partner modelNegative29:48

    Delivery-service-partner structures enable worker exploitation and liability avoidance.

    “It allows them to exploit. the workers and deny responsibility for accidents,”

    Listen at 29:48

  12. Ian Kingon Amazon New York City delivery operationsPositive31:35

    Amazon is unlikely to move deliveries outside New York City.

    “It is not in their interest to move outside the city, and they don't.”

    Listen at 31:35

  13. Ian Kingon New York City delivery accidentsNegative32:15

    Delivery-related accidents have increased dramatically in New York City.

    “There have been a dramatic increase in the number of accidents.”

    Listen at 32:15

  14. Anna Rathbunon Amazon worker treatmentPositive33:50

    Improved Amazon worker treatment would help competitors improve worker treatment.

    “if Amazon has to treat its workers better, it will make it easier for the other companies to treat their workers better and be competitive.”

    Listen at 33:50

  15. Ian Kingon Amazon delivery-worker liabilityPositive35:27

    Amazon liability would improve delivery safety and consumer protection.

    “then it's going to make conditions safer, consumers more protected, and enforce the laws of the city”

    Listen at 35:27

  16. Tim Stenovecon Reddit S&P 500 inclusionPositive36:03

    S&P 500 inclusion requires passive funds to purchase Reddit shares.

    “passive investment funds have to buy shares of Reddit.”

    Listen at 36:03

  17. Ian Kingon Non-consensus AI startupsPositive42:32

    Non-consensus startup theses could produce world-changing market leaders.

    “we recognize that there are a series of things that could be proven to be true. that would change the world and make that company, in fact, a market leader?”

    Listen at 42:32

  18. Ian Kingon AI future of workPositive43:31

    AI creates a major opportunity in the future of work.

    “There's a massive opportunity in the future of work.”

    Listen at 43:31

  19. Ian Kingon AI-driven productivityPositive44:34

    AI-driven productivity will not be a zero-sum game.

    “We believe it's not a zero-sum game.”

    Listen at 44:34

  20. Ian Kingon AI-driven employmentPositive44:46

    AI adoption will create many additional jobs.

    “There are going to be many more jobs.”

    Listen at 44:46

  21. Ian Kingon AI-driven labor transitionNeutral45:21

    Workers will need reskilling as AI changes jobs.

    “We're going to have to reskill.”

    Listen at 45:21

  22. Ian Kingon 224 Ventures fundraisingPositive45:58

    224 Ventures raised its initial capital in two months.

    “We raised that capital across just two months.”

    Listen at 45:58

  23. Ian Kingon AI investment hotspotsPositive46:48

    Several global cities are important AI investment hotspots.

    “Predominantly Silicon Valley, New York, Toronto, Paris, London, Tel Aviv are all very interesting hotspots for AI.”

    Listen at 46:48

  24. Ian Kingon San Francisco AI ecosystemPositive47:01

    San Francisco is currently the leading AI hub.

    “There is no question.”

    Listen at 47:01

  25. Ian Kingon United States AI research leadershipPositive47:28

    The United States should attract global researchers to maintain AI leadership.

    “enabling researchers from all over the world to come to the United States, to push the frontier, to keep the United States at the bleeding edge is a top priority.”

    Listen at 47:28

  26. Ian Kingon Jan LeCun and Oriel VignolsPositive48:09

    Jan LeCun and Oriel Vignols provide strong AI ecosystem connections.

    “Jan and Oriel are embedded in the ecosystem.”

    Listen at 48:09

  27. Anna Rathbunon Beverage businessNegative50:26

    The beverage market is highly competitive.

    “the beverage business is wildly competitive.”

    Listen at 50:26

  28. Anna Rathbunon UnwellNegative51:35

    Unwell shows beyond Call Her Daddy have not achieved major mainstream success.

    “that have broken through into the mainstream in a major way.”

    Listen at 51:35

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.

Weekly email

State of the Pod, every Monday

The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.

Listen to the full episode and explore every guest, topic, and moment on PodLume.