Written by: Rita

Trends Guide

In the past two weeks, a series of seemingly isolated events occurred in the software industry: Starbucks built its own AI tools to replace Microsoft and IBM software, Meta is planning to sell cloud AI computing power, and Microsoft replaced OpenAI and Anthropic with its self-developed MAI model. JPMorgan Chase read these events together and concluded that they point in the same direction: the profit pool in the software industry is being reallocated. AI has changed the scale and speed of building in-house tools. To know who will be hurt the most and who will benefit from this, we need to break down and examine each layer of the chain reactions.

Starbucks Builds Its Own AI Tools, Enterprise Software Faces "Cost Cutting"

Starbucks spends about $400 million annually on software, and the CTO is reviewing every contract and service. The plan is to replace some applications purchased from Microsoft (inventory tracking) and IBM (operations management) with self-developed AI-assisted tools, with some going live possibly next year. JPMorgan Chase believes this is a specific case of the "build vs buy" and "software will be free" thesis. Building in-house is not a new trend, but AI has significantly lowered the thresholds for self-development, giving companies greater motivation to do it themselves when facing cost pressures. The complexity and cost of maintaining self-developed software are often underestimated, but for a company like Starbucks, it is still cost-effective after doing the math.

Meta Selling Cloud Computing Power, AI Computing Demand Remains Strong

Meta is developing "Meta Compute" cloud infrastructure to sell AI computing power and model access to external developers, similar to AWS Bedrock. The market's first reaction was that Meta has excess computing power, but Morgan Stanley offered the opposite judgment: the demand for AI computing power is so strong that Meta believes this business is worth pursuing. Meta plans to double its data center capacity from 7GW to 14GW in the next 12 months. Morgan Stanley believes that Meta's sale of computing power will not reduce its contracts with CoreWeave, Nebius, and other NeoCloud providers because the demand is simply too great.

DigitalOcean Secures Billion-Dollar Contracts, AI Inference Demand is Rising

DigitalOcean announced its Q2 performance ahead of schedule, with remaining performance obligations (RPO) expected to exceed $800 million, a tenfold year-over-year increase, and the weighted average contract term extended from about 1.6 years to over 3 years. The company attributes this to signing multiple nine-digit dollar annual customer commitments in Q2, with expected revenue growth of about 29% in Q2, up from 22% in Q1. JPMorgan Chase emphasizes two points: AI inference now accounts for a significant proportion of contracts, and AI workloads are shifting towards inference; large long-term contracts are changing DigitalOcean's customer structure and business fundamentals, which is a sign of maturity in the NeoCloud space.

Microsoft Replaces OpenAI with Self-Developed Models, Cutting-Edge Models Are No Longer Essential

Microsoft has begun replacing OpenAI and Anthropic models with its self-developed MAI model in products like Excel and Outlook, with tens of thousands of AI prompts running on MAI each week. Morgan Stanley judges that this serves as a margin lever for Microsoft, having cost-reduction effects in both the short and long term. More importantly, the industry signal is: software companies are proving that AI applications can run without being tied to the most cutting-edge large models, as the industry is shifting from a "model arms race" to a "cost optimization phase".

Cloudflare Sets Pricing for AI Crawlers, Information Intermediaries are Earning New Revenue

Cloudflare launched a two-way bot paywall that allows content owners to charge AI crawlers. The new default rules will take effect on September 15, blocking training and proxy bots by default on ad-supported pages. More than half of Cloudflare's network requests now come from AI agents. Morgan Stanley believes this is a new source of revenue for Cloudflare, and with its extensive network covering a large number of websites, Cloudflare is positioning itself as a settlement layer between publishers and AI model providers.

Trend Perspective

What Morgan Stanley really wants to highlight is that these five events collectively reveal a structural migration that is occurring: the AI profit pool is shifting from the model layer to downstream. The upstream model layer (OpenAI, Anthropic) faces pressure to be replaced, and Microsoft's use of its self-developed MAI is evidence of this. The midstream infrastructure layer (AWS, CoreWeave, DigitalOcean) continues to see strong demand, but the growth rate and contract structure are changing. The bargaining power of downstream application layer enterprise customers (Starbucks) is increasing, and self-development is replacing external procurement. Do not equate AI beneficiaries with model providers. Models may become commoditized, while inference infrastructure and intermediaries that can charge tolls from traffic (Cloudflare) may have more lasting pricing power. This is the signal that Morgan Stanley extracts from the noise: the stock selection logic must follow the profit pool.

Disclaimer

This article is an arrangement and interpretation of a third-party brokerage research report (JPMorgan Chase, July 13, 2026) by Trends Research. The ratings, target prices, earnings forecasts, and related judgments quoted in the text are the views of the brokerage's analysts and represent only the positions of their respective institutions, not the views of Trends Research, and do not constitute any investment advice.

Markets are risky, and decisions must be independent. This article should not be used as the basis for buying or selling any securities.