Goldman Sachs Updates Belief List, Focusing on Structural Winners in AI and the Rebound in Energy Capital Expenditures
AI summary card
Goldman Sachs Updates Belief List, Focusing on Structural Winners in AI and the Rebound in Energy Capital Expenditures
The report adds Siemens Energy and Smith & Nephew to the European belief list, emphasizing that AI enterprise applications rely on high-quality data moats, while also reminding investors to pay attention to oil service companies like TGS benefiting from the recovery in exploration capital spending.
- Siemens Energy is listed as a structural winner, benefiting from growing power demand for data centers and grid investments
- Smith & Nephew is expected to accelerate growth in the second half due to product innovation and operational simplification
- RELX scores highest (9/10) in the AI framework because it owns a proprietary database that is hard to replicate
- TGS (Buy rating) will benefit from the rebound in exploration spending by international oil companies; the industry has already consolidated and gained pricing power
- The effectiveness of enterprise AI implementation varies widely, and data structure and orchestration are key bottlenecks
- The overconcentration of economic value in the semiconductor sector is unsustainable
Report interpretation
Overview
This report is part of Goldman Sachs’ June 2026 ‘Connecting Points: The Belief List’ series, aiming to examine key research topics such as artificial intelligence (AI) and TGS plus from different perspectives. The report updates its regional ‘belief list,’ adding Siemens Energy and Smith & Nephew, and delves deeply into the practical benefits of AI in enterprises, the data moats of media and internet companies, and the overlooked opportunity of the recovery in oil and gas exploration capital expenditures.
Core views
The report argues that the current AI investment boom shows structural differentiation. On one hand, Jim Covello points out that although enterprise AI spending has increased significantly, actual results have been ‘at best mixed,’ and the overconcentration of economic value in the semiconductor sector is unsustainable. The real key to unlocking enterprise AI potential lies in ‘data structure and orchestration.’ On the other hand, Adam Berlin, the newly appointed European media and internet analyst, suggests that the impact of AI on industries is not unidirectional; many companies have ‘more resilient AI moats.’ The core criterion is whether they possess ‘proprietary, hard-to-replicate databases,’ especially in sensitive areas like law. Based on this framework, RELX scores highly at 9/10 and is listed as a ‘buy,’ despite being mistakenly categorized under ‘AI risk.’ In non-AI sectors, the report specifically reminds investors not to overlook investment opportunities in TGS (an offshore geophysical company). Analyst Michele della Vigna points out that international oil companies (IOCs) are gradually increasing exploration spending, and the seismic data services industry has become highly consolidated. As the leader, TGS will demonstrate strong pricing power during the upcycle, and historical experience shows its stock price will quickly re-rate. Additionally, the report updates views on several companies: Siemens Energy is seen as a structural winner driven by power demand for data centers and grid investments; Smith & Nephew is poised for a growth inflection point in the second half thanks to innovative product launches and operational optimization; Ferrari, meanwhile, continues to strengthen its ASP (average selling price) supported by personalized customization.
Analysis framework
Goldman Sachs adopts a multi-dimensional cross-validation approach to form investment views. First, through the ‘belief list’ mechanism, it screens stocks from a large pool for those with the greatest absolute return potential—a strategy that itself focuses on core contradictions. Second, when analyzing AI themes, the report does not stop at general technological trends but dives deep into specific bottlenecks in enterprise implementation (such as data quality) and business model moats (such as proprietary databases), building a quantifiable evaluation framework (like Adam Berlin’s AI framework). This methodology closely links macro themes with micro-level company fundamentals. Finally, when uncovering overlooked opportunities like TGS, the report looks back at industry cycles (exploration capital expenditure cycle), competitive landscape (industry consolidation), and historical stock performance (lumpy spending leading to rapid re-rates), reflecting a typical bottom-up and top-down combined cyclical analysis approach.
Methodology notes
Capital expenditures in the oil and gas exploration services industry have a ‘lumpy’ characteristic
Demand in this industry does not grow linearly but bursts sharply when exploration cycles start. Since the industry is highly consolidated, suppliers like TGS have strong pricing power. Once demand recovers, company profits and stock prices will quickly re-rate.
In the AI era, the moat lies in proprietary, hard-to-replicate data assets
The report argues that while algorithms may converge at the application level, high-quality, licensed, domain-specific proprietary databases constitute the real competitive barrier—especially in professional services like law and finance. This is the core logic behind why companies like RELX are favored.
AI computing power demand drives upstream investments in power and grid infrastructure
The report sees the explosive growth of AI data centers as a driving factor and traces its downstream transmission along the industry chain, identifying strong demand for grid technology and power equipment (such as Siemens Energy). This reflects a complete analytical chain—from end-use applications to upstream infrastructure.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Siemens Energy (ENR1n.DE)Benefiting from structural growth in power demand driven by AI data centers and grid investments
- Strengths
- Possesses an underappreciated grid technology business and is a structural winner
- Comparison
- Valuation still has room to rise compared to U.S. peer GE Vernova
- Smith & Nephew (SN.L)Benefiting from accelerated growth driven by innovative product launches and operational model simplification in the second half
- Strengths
- Growth momentum driven by innovation, reasonable valuation does not fully reflect new product potential
- RELX Plc (REL.L)Benefiting from its hard-to-replicate AI moat in the professional information services sector (proprietary database)
- Strengths
- Highest score in the AI framework, strong growth momentum across all business segments, 6% free cash flow yield
- Comparison
- Misclassified as ‘AI risk’ category, creating expectation gaps
- TGS ASA (TGS.OL)Benefiting from the rebound in exploration capital expenditures by international oil companies
- Strengths
- Industry is consolidated and has the ability to wait for the upcycle and exercise pricing power
Key data
- RELX AI Framework Score9/10Highest score in analyst Adam Berlin’s new AI evaluation framework
- European Power Demand Growth YTD+2% YTDAverage growth across Spain, Germany, Italy, UK, and Portugal, driven by heat pumps, electric vehicles, and manufacturing
- Siemens Energy’s Potential Capital Return€29 billionAnalysts expect this amount to be returned by 2030, far exceeding the current commitment of €10 billion
Impact & implications
The report argues that the market’s understanding of AI is overly focused on the semiconductor hardware layer, overlooking structural opportunities in the data and application layers. Companies with high-quality data assets (such as RELX) and those benefiting from the expansion of AI infrastructure (such as Siemens Energy) are undervalued. Meanwhile, in the energy sector, the market focus is too narrow, ignoring the exploration services segment, which is about to enter an upcycle. As the industry leader, TGS will significantly benefit. These insights provide investors with allocation strategies that go beyond market consensus.