DeepSeek's AI Breakthrough Benefits Microsoft, But Oracle Faces Cloud Growth Challenges: Analyst

China-based AI lab DeepSeek announced it released AI models (including R1) that took two months to build using less-powerful Nvidia Corp NVDA H800 chips, costing far less to build than LLMs today.

The exact cost of development and energy consumption of DeepSeek are not fully documented. However, BofA Securities analyst Brad Sills noted that the lower cost of running AI models benefits the general software group.

Sills noted the advancement of DeepSeek as a positive for Microsoft Corp MSFT.

Also Read: AMD Gets A Downgrade And It’s Not DeepSeek Related

Microsoft is already on the cutting edge of AI infrastructure efficiencies, given that the company is running inferencing workloads for the largest LLMs, ChatGPT, on its Azure platform via the OpenAI partnership.

If DeepSeek can generate efficiencies at a lower scale, this would bode well for Microsoft’s ability to continue lowering the cost of AI computing and drive scale on capital expenditures (fiscal 2025 of $82.2 billion, including capital leases, or 30% of total revenue), via its operational efforts.

Sills expects Microsoft to begin demonstrating more data center scale from here. In his preview, he cited the expectation for Azure acceleration in the second half of 2025 and a better margin outlook, largely due to data center efficiencies.

Sills noted the DeepSeek advancement as a potential negative for Oracle Corp ORCL.

Positive sentiment on Oracle’s shares has been driven by accelerating cRPO/backlog growth (+21% Y/Y in the second quarter of 2025, up from 7% in the second quarter of 2024), fueled by OCI traction.

According to Sills’ channel checks, the core driver of OCI/cloud growth has been migrating on-premise Oracle DB instances to OCI, unaffected by the DeepSeek breakthrough.

However, with Oracle’s access to GPUs via an Nvidia partnership and an industry-wide capacity shortage, some of the OCI growth has also been driven by GPU rentals. Easing capacity constraints could impact demand for GPU rentals and Oracle’s cRPO growth.

DeepSeek has positive implications for AI names in the application layer, like Microsoft’s M365 Copilot, ServiceNow Inc’s NOW Now Assist, Salesforce Inc’s CRM Agentforce, HubSpot Inc’s HUBS Breeze AI and Intuit Inc INTU Assist for QuickBooks and TurboTax.

A lower cost of computing could reduce the sales costs of AI computing and drive better margins on AI-enabled offerings.

More importantly, agentic application providers could pass savings on to enterprises through lower prices, catalyzing agents’ adoption and proliferation.

Feedback from SIs in the software ecosystem suggests that cost is the biggest hurdle to adopting AI applications.

Application providers have moved from upfront monthly user subscription pricing to pay-as-you-go consumption-based pricing.

Sills noted advancements in cost could drive prices even lower and, therefore, adoption higher.

Sills has a price forecast of $850 for HubSpot, $780 for Intuit, $510 for Microsoft, $195 for Oracle, $440 for Salesforce, and $1,280 for ServiceNow.

Also Read:

• DeepSeek Sell-off Offers Opportunities On Semicaps, Analyst Bumps Up Price Forecasts For KLA, Nova And Camtek

Photo via Midjourney and Shutterstock.