Apple (AAPL) doesn’t have to win the artificial intelligence model competition to win the next phase of AI.

That’s the bet underlying Bank of America’s latest Apple call.

Wamsi Mohan, an analyst at BofA, increased his price target on Apple to $380 from $330 and reiterated a Buy rating on May 26. The new target suggests 23% upside from Apple’s price of $308.82 as of May 26, according to the article.

The argument is not that Apple is suddenly like OpenAI, Anthropic or Google.

The thing is, the AI sector may be heading into Apple’s home turf, the iPhone.

BofA says the next phase of AI will shift value from the company with the best model to the company that controls the trusted endpoint where user intent, identity, payments, apps, permissions and privacy already meet.

For Apple that goal is the iPhone.

But that complicates Apple’s next AI test beyond just making Siri smarter.

For the bull case to work, Apple needs Siri to become the execution layer of the iPhone. And it needs developers to make their applications callable by that system.

“To win, Apple needs to evolve Siri into the orchestration layer of the iPhone,” Mohan wrote in the report.

Apple’s Siri plan depends on app developers

BofA’s analysis is on agentic AI, a move from AI that answers questions to AI that does things.

That difference matters to investors.

A chatbot can recommend a restaurant. An AI agent may check a user’s schedule, discover a nearby reservation, make the reservation, text a contact and confirm payment.

BofA says that smartphones are the natural control point for that kind of AI, as they already have identification, authentication, payments, location, camera, microphone, contacts, calendar, messages, apps and real-time context integrated.

Apple is in a very strong position with its control of the iPhone, Apple silicon, iOS, App Store distribution, Apple ID, Face ID, Apple Pay, Wallet and privacy architecture.

The paper says Apple’s moat is built on two layers: silicon for local inference and iOS to control context, app access, permissions, authentication, payments and trust.

That is where App Intents becomes crucial.

App Intents offers developers an opportunity to expose app functionality to Siri, Spotlight, Shortcuts and system experiences, says BofA. For Apple to succeed, that foundation needs to evolve into a comprehensive agent-action system covering travel, commerce, productivity, payments, messaging, files, calendar, media, and enterprise workflows.

In other words, Apple wants apps to be machine-callable.

Related: Bank of America resets Apple stock price target on AI update

Siri is more powerful if developers expose helpful actions. If not, Siri will stay confined to Apple’s own apps and a constrained range of workflows.

That’s the true execution risk.

Apple may own the phone, the OS and the payment layer. But agentic AI only works if third-party apps become fulfillment endpoints that Siri can call safely and reliably.

Bank of America sees a large Apple revenue opportunity

The dollar figures in BofA’s study make the point clear.

In a base case, Mohan predicts agentic Siri could bring $15 billion to $30 billion in incremental fiscal 2030 revenue. In a bull situation, the opportunity grows to $40 billion to $65 billion.

The firm divides that opportunity into six potential sources of revenue: a paid Apple Intelligence Pro subscription, default model placement or routing costs, App Store agent-commerce take rates, Apple Pay and Wallet routing fees, advertising in agent results and an iCloud+ AI layer.

In the basic case, BofA predicts 50 million to 75 million Apple Intelligence Pro members at $15 per month, or $9 billion to $14 billion in revenue.

It also projects $200 billion to $300 billion in agent-mediated gross merchandise value at a take rate of 1% to 2%, creating another $2 billion to $6 billion opportunity.

The bull case is bigger. BofA predicts 100 million to 125 million Apple Intelligence Pro subscribers may generate $18 billion to $23 billion, while $400 billion to $600 billion in agent-mediated gross merchandise value could yield$8 billion to $18 billion.

That’s why developer adoption is important.

The new App Store could not only be about what app gets downloaded. Perhaps it is which app Siri uses to respond when a user asks for something to be done.

Key Apple AI figures from BofA

• New Apple price target: $380, up from $330.
• Rating: Buy.
• Implied upside: 23% from $308.82.
• Base-case agentic Siri revenue opportunity: $15 billion to $30 billion by fiscal 2030.
• Bull-case agentic Siri revenue opportunity: $40 billion to $65 billion by fiscal 2030.
• Base-case Apple Intelligence Pro assumption: 50 million to 75 million subscribers at $15 a month.
• Bull-case Apple Intelligence Pro assumption: 100 million to 125 million subscribers at $15 a month.
• Base-case agent-commerce gross merchandise value: $200 billion to $300 billion.
• Bull-case agent-commerce gross merchandise value: $400 billion to $600 billion.

BofA also calculates the infrastructural load underlying that opportunity.

The company thinks a typical agentic iPhone request would consume about 1,900 tokens weighted. A moderate agentic task could take 1,000 to 3,000 tokens, a heavy workflow 3,000 to 12,000 tokens.

More AI:

• Micron sits at the center of a red-hot chip rally
• IBM CEO sends blunt message on AI and quantum computing
• Anthropic CEO makes shocking admission about AI

In its base assumption, BofA expects about 1.36 billion AI iPhone users making 10 requests every day with about 2,000 tokens per request by fiscal 2030. That’s almost 9.9 quadrillion tokens a year, with nearly half processed through the cloud.

Hence the significance of Apple’s mixed strategy. Apple will need to rely on on-device processing, Private Cloud Compute and third-party compute to create a powerful AI experience, says the research.

The economics are very different depending on where those requests run.

But if Apple depends too much on third-party models, AI might become a variable-cost headache. Apple could safeguard margins and control more of the user experience by doing more work on devices or via its own cloud infrastructure.

Apple faces an AI challenge Wall Street may be missing.

Photo by picture alliance on Getty Images

Apple’s AI opportunity comes with a catch

The biggest danger to BofA’s argument isn’t whether Apple has enough consumers.

The question is whether Apple is able to modify user behavior fast enough.

BofA estimates Apple’s iPhone installed base might hit 1.5 billion by 2026 and 2 billion or so by 2030. The company also possesses the building blocks for a trusted AI layer, such as hardware, software, app distribution, identity, payments and privacy.

But Siri has missed expectations for years and BofA admits that market mistrust is legitimate.

If consumers develop daily routines with ChatGPT, Gemini, Claude or other huge language models, Apple may still hold the hardware but steal less of the AI value pool.

This puts pressure on Apple’s plan for developers.

App Intents have to be more than a technological framework for the firm. It needs to be the agentic version of App Store distribution, where developers fight not just for downloads, but for the right to be called up by Siri when customers want stuff done.

If Apple succeeds, Siri is more than a voice-based assistant. It becomes the control layer for apps, payments, commerce and personal context in the iPhone.

If Apple misses, the $380 price target would look less like a projection and more like a reminder of how much execution remains.

Related: Apple makes quiet AI move that will change iPhones, Vision Pro