Apple pays Google for AI: intelligence on demand arrives
The hosts opened with news that Apple is reportedly paying Google $1 billion to access Gemini for features like Siri. Andy called it an admission that Apple Intelligence, launched in 2024, has not taken hold. Separately, Google paid Apple $38 billion between 2021 and 2022 to secure default search placements, illustrating the scale of money flowing between these companies.
Apple didn’t need to develop their own version. They just bought it on the open market.
Andy Hoar, Master B2B
The implication is that AI is becoming commoditized. Apple chose to purchase intelligence rather than build it, which Andy described as intelligence on demand. If the largest consumer technology company in the world can buy AI off the shelf, the question becomes: what does that mean for the companies building it, and what does it mean for B2B firms deciding how to use it?
The protocol wars: OpenAI versus Google
Behind the consumer-facing products is a deeper battle over infrastructure. The hosts described what they called protocol wars, competing standards for enabling commerce through AI chatbots. OpenAI launched ACP in September 2025 in partnership with Stripe, creating what the hosts called a closed ecosystem similar to Apple’s app approach. Google responded at NRF with Universal Commerce Protocol, partnering with Shopify, Target, Walmart, and Etsy in a more open, Android-style model.
The stakes are significant. If a user asks an AI chatbot to buy a product, the chatbot needs access to price, inventory, and product information from vendors. Whichever protocol wins will determine how that connection works. Brian summarized the dynamic: one of them will win, and eventually it becomes the standard.
A $3 trillion investment for a $300 billion market
The hosts identified eight companies spending heavily on AI infrastructure: Google, OpenAI (partnered with Microsoft), xAI (Elon Musk’s Grok), DeepSeek, Claude, Perplexity, Amazon’s Titan, and Meta. Collectively, these players are reportedly investing around $3 trillion in chips, compute, data centers, and people.
Google’s entire search business is like a $300 billion business. Do the math. Who’s going to recoup their $3 trillion investment?
Andy Hoar, Master B2B
The bet is that the market will expand dramatically as AI becomes embedded in more applications and processes. If the $300 billion market stays roughly that size, the math does not work. If AI spreads through the enterprise the way some expect, the market could grow enough to justify the investment. The outcome will determine how many of these eight players survive.
Why this looks like an oligopoly
Andy drew on his early career at an antitrust economics consulting firm to define oligopoly: a market structure where five firms control more than 60% of sales. The current AI chatbot market fits that definition. ChatGPT reportedly holds around 68% share, down from 85% a year ago, with Google Gemini at roughly 18% and growing. The rest is split among smaller players.
The barrier to entry is capital. Building and training foundational AI models requires trillions of dollars in infrastructure, access to massive datasets, and specialized talent. These are the same dynamics that created oligopolies in telecommunications, railroads, and airlines. The hosts expected the market to consolidate further, perhaps to three or four major players at the foundation layer, with others licensing or reselling capacity.
What this means for B2B companies
Most B2B buyers are not yet purchasing through AI chatbots, but the trajectory points in that direction. Brian noted that B2C companies are racing to adopt commerce protocols so their products appear when a consumer asks a chatbot for a recommendation. The same dynamic will eventually reach B2B, particularly for commodity products and straightforward reorders.
The hosts offered two near-term takeaways. First, pay attention to the protocol wars. The winning standard will shape how products are discovered and purchased through AI. Second, recognize that the level of investment required to build foundational AI means B2B companies will be consumers of AI, not builders. The question is which platform to adopt and how to ensure product data is structured for AI discovery.
B2B companies are going to have to decide which one of these protocols to be a part of. And over the horizon, it may displace you to some extent, just like Amazon has done with search shifting there.
Andy Hoar, Master B2B
How many foundation models will survive?
The hosts polled their community on how many foundational AI models would power the market in five years. The results were evenly split: roughly a third said two to three, a third said four to five, and a third said more than six. The uncertainty reflects how early the market is.
Brian predicted consolidation to a limited number, similar to how the telecom industry evolved after the breakup of Bell. Andy agreed, expecting perhaps three or four foundational players with others reselling capacity. Both noted that regulation may eventually apply if the market becomes as concentrated as they expect.