JPMorgan Research Report Analysis: Self-Built AI Replaces Enterprise Software, Software Profit Pool Shifts from Model Layer to Infrastructure
TechFlow Selected TechFlow Selected
JPMorgan Research Report Analysis: Self-Built AI Replaces Enterprise Software, Software Profit Pool Shifts from Model Layer to Infrastructure
Stock selection logic should follow the profit pool.
Navigating Web3 tides with focused insightsBy: Rita
TechFlow Guide
Over the past two weeks, a series of seemingly isolated events have occurred in the software industry: Starbucks building its own AI tools to replace software from Microsoft and IBM, Meta planning to sell AI computing power via its cloud business, and Microsoft replacing OpenAI and Anthropic with its self-developed MAI model. JPMorgan reads these events together and believes they point in the same direction: the profit pool of the software industry is being redistributed. AI has changed the scale and speed of in-house development. To know who will be hurt the most and who will benefit the most next, we need to break down the chain reaction at each layer.
Starbucks Builds Its Own AI Tools, Enterprise Software Faces "Cost Cutting Blade"
Starbucks spends approximately $400 million annually on software. The CTO is reviewing every contract and service, planning to replace some applications procured from Microsoft (inventory tracking) and IBM (operations management) with self-developed AI-assisted tools, with some going live possibly next year. JPMorgan believes this is a concrete case of the "build vs. buy" and "software will be free" narratives. Building in-house is not a new trend, but AI has significantly lowered the barrier to entry, giving enterprises greater motivation to do it themselves when facing cost pressures. The complexity and cost of maintaining self-developed software are often underestimated, but for an enterprise of Starbucks' scale, it still pays off after calculating the total cost.
Meta Builds Cloud to Sell Computing Power, AI Compute Demand Remains Hot
Meta is developing the "Meta Compute" cloud infrastructure business to sell AI computing power and model access to external developers, similar to the AWS Bedrock model. The market's first reaction was that Meta has excess computing power, but Morgan Stanley gave the opposite judgment: AI compute demand is so strong that Meta feels this business is worth doing. Meta plans to double its data center capacity from 7GW to 14GW over the next 12 months. Morgan Stanley believes Meta selling computing power will not reduce its compute contracts with NeoClouds such as CoreWeave and Nebius, because the demand is simply too large.
DigitalOcean Secures Billion-Level Contracts, AI Inference Demand is Scaling Up
DigitalOcean announced its Q2 results early, with Remaining Performance Obligation (RPO) expected to exceed $800 million, a year-over-year increase of over 10 times, and the weighted average contract term extended from about 1.6 years to over 3 years. The company attributes this to signing multiple nine-figure dollar annual customer commitments in Q2. Q2 revenue growth is expected to be about 29%, higher than Q1's 22%. JPMorgan emphasizes two points: AI inference already accounts for a significant proportion of contracts, and AI workloads are shifting towards the inference side; large-value long-term contracts are changing DigitalOcean's customer structure and business foundation, which is a sign of the maturity of the NeoCloud sector.
Microsoft Replaces OpenAI with Self-Developed Model, Frontier Models No Longer a Necessity
Microsoft has begun replacing models from OpenAI and Anthropic with its self-developed MAI model in products such as Excel and Outlook, with tens of thousands of AI prompts running on MAI every week. Morgan Stanley judges this to be a profit margin lever for Microsoft, with cost-reduction effects in both the short and long term. The more important industry signal is: software companies are proving that AI applications can run without being tied to the most frontier large models, and the industry is moving from a "model arms race" into a "cost optimization phase".
Cloudflare Prices AI Crawlers, Information Intermediaries are Making New Money
Cloudflare has launched a two-way bot paywall, allowing content owners to charge AI crawlers. The new default rules will take effect on September 15, defaulting to blocking training and agent bots on pages containing ads. More than half of Cloudflare network requests now come from AI agents. Morgan Stanley judges this to be a new revenue source for Cloudflare; leveraging its network scale covering a large number of websites, Cloudflare is positioning itself as a settlement layer between publishers and AI model providers.
TechFlow Perspective
What Morgan Stanley really wants to say is that these five events jointly reveal a structural shift happening: the AI profit pool is shifting downstream from the model layer. The upstream model layer (OpenAI, Anthropic) faces pressure of being replaced, with Microsoft using its self-developed MAI as clear proof. The midstream infrastructure layer (AWS, CoreWeave, DigitalOcean) continues to see strong demand, but growth rates and contract structures are changing. Downstream application layer enterprise customers (Starbucks) have rising bargaining power, and in-house development replacing external procurement is happening. Do not equate AI beneficiaries with model providers. Models may become commoditized, while inference infrastructure and the intermediary layer capable of collecting tolls from traffic (Cloudflare) may possess more enduring pricing power. This is the signal Morgan Stanley extracted from the noise: stock selection logic must follow the profit pool.
Disclaimer
This article is a compilation and interpretation by TechFlow Research of a third-party brokerage research report (JPMorgan, July 13, 2026). The ratings, target prices, earnings forecasts, and related judgments cited in the text are the views of the brokerage's analysts, represent only the position of their affiliated institution, do not represent the views of TechFlow Research, and do not constitute any investment advice.
The market carries risks; decisions must be made independently. This article should not be used as a basis for buying or selling any securities.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@chaoxiangooo
