It’s four years since the start of the AI-fueled bull market, and Société Générale is urging investors to stick with all parts of it.
In a recent client note, the bank said that new legs of the trade are unfolding, which should continue to propel returns upward, even for the AI infrastructure and application trades that have already seen huge upside.
“After three years of massive infrastructure buildout, AI value creation is beginning to broaden downstream,” Manish Kabra, SocGen’s chief US equity strategist, wrote in the note.
“Rapidly evolving use cases and an expanding ecosystem of software, agents and physical AI applications are widening the opportunity set, from infrastructure providers to companies directly monetising AI products and, ultimately, those capturing productivity gains from adoption.”
There’s perhaps no better illustration of Kabra’s point than Meta’s agentic AI model, Muse.
After worries that AI buildout stocks had run their course, the product has sparked new optimism that AI can indeed be monetized and that other firms are likely to keep spending to develop their own agents. The agentic AI boom also promises to boost productivity of firms across sectors of the economy.
To keep playing the AI theme, SocGen has released a three-part investing playbook. Here’s what to know.
AI builders
This is the AI infrastructure trade. Think chipmakers, data center stocks, and power companies. These stocks are the recipients of massive hyperscaler capex.
Kabra laid out five parts of the supply chain, and stock picks to play each one. We’ve listed an example stock from each list each below:
Chipmaking enablers: Applied Materials (AMAT)
Chipmakers: Nvidia (NVDA)
Server and networking providers: Arista Networks (ANET)
Power Suppliers: GE Vernova (GEV)
Cooling: Vertiv (VRT)
AI sellers
This basket capitalizes on companies monetizing the AI buildout through their models.
“AI is part of what customers pay for, directly translating adoption into revenue,” Kabra wrote.
“Frontier-model partnerships are further accelerating monetisation,” he continued. “Microsoft and Oracle remain closely tied to OpenAI, while Amazon and Google Cloud partner with Anthropic. Beyond hyperscalers, Salesforce partners with both OpenAI and Anthropic, while ServiceNow has strategic collaborations with both to embed frontier models directly into enterprise workflows.”
Here are four parts of this basket, and an example stock pick from each:
Hyperscalers: Microsoft (MSFT)
Cybersecurity: CrowdStrike (CRWD)
AI software and data: Salesforce (CRM)
Physical AI: Tesla (TSLA)
AI users
This basket is made up of companies “using AI to improve their underlying economics,” Kabra said. It’s the AI productivity beneficiaries, or the companies that will use AI technology to grow their margins.
The bank screened for stocks by looking at firms with both strong future revenue potential and rising earnings per employee.
The three industries leading the way in AI productivity gains are financials, healthcare, and industrials, the bank said.
SocGen shared stock picks for each sector — an example for each one includes:
Financials: JPMorgan (JPM)
Healthcare: Eli Lilly (LLY)
Industrials: GE Aerospace (GE)