Must Read Research
Also featuring commentary from Global Economic Weekly
June 29, 2025
Candace Browning, Head of BofA Global Research
From the battlefield to the server farm, some questions can no longer wait. This week, we explore NATO’s push toward 5% defense spending and what it means for European and U.S. defense companies, examine the role of small modular reactors in meeting surging electricity demand, unpack the potential of AI in revolutionizing personalized medicine and look at the critical pieces behind the surge in AI infrastructure buildout.
Nuclear power offers a number of advantages, but large reactors also come with challenges including high upfront costs and long construction timelines.
Small Modular Reactors (SMR) offer an alternative and 80+ different designs are in development globally. With policy tailwinds and commercialization in 2029-2030, Head of Korea Research Joon-Ho Lee sees 2025 as an inflection point for SMR component order momentum. The U.S. Department of Energy anticipates costs can be cut by 40% via standardization, supply chain development and modularization. Korea is a leader in nuclear exports – the country maintained its industrial base even as supply chains in the U.S. and France weakened during past nuclear phase-outs.
NATO (North Atlantic Treaty Organization) allies last week agreed to spend 5% of GDP (gross domestic product) on defense by 2035.
We see $371 billion of incremental NATO spending on core defense if the target is met, relative to $507 billion in 2024 spend. This amounts to a 5% CAGR (compound annual growth rate) over the next 10 years, but European Defense companies should grow closer to 12.5% medium-term as they garner more of the incremental spend. France, for example, has a strong enough defense industry to be almost 100% self-sufficient according to EMEA Aerospace & Defense Analyst Ben Heelan. Defense electronics (radar, sensors) and defense systems (integrated air, missile defense and precision strike) will be the areas with growth well beyond 2030. Land systems will also see significant expansion, but rates of growth should slow after 2030.
U.S. SMID Cap Biotech Analyst Alec Stranahan sees AI as the breakthrough force set to transform personalized medicine.
Traditional care often overlooks individual patient differences and while some patients benefit with standard-of-care treatment, there is rarely a one-drug-fits-all scenario (e.g., the best cancer drugs can be ineffective in up to 75% of patients, Alzheimer's drugs in 70%, diabetes drugs in 43%, etc.). Precision and personalized therapies are more effective for targeted and rare diseases but are costly, labor-intensive and time-consuming. AI’s edge lies in analyzing vast, complex datasets to identify predictive biomarkers that can dramatically improve outcomes in a cost-effective way. In one study of 346 phase 1 trials, biomarker-based selection strategy for patients led to response rates of 30.6% vs. just 4.9% under random selection, and progression-free survival nearly doubled (5.7 vs. 2.95 months). While full integration into clinical protocols is still far out, Alec expects siloed uses to increase near term.
U.S. Semis Analyst Vivek Arya predicts a $1 trillion AI infrastructure boom by 2030—driven by mission-critical data center investments.
While stocks exposed to the theme have rallied, there’s plenty of runway left. Today’s $420 billion data center spend (out of $5.4tn in global IT (information technology)) could more than double to $980 billion by 2030, growing 19% annually, far outpacing broader IT at 8% CAGR. Within that, AI infrastructure is set to grow 26% annually, reaching $823bn, led by AI servers (~$700bn). Networking ($74bn) and storage ($39bn) round out the stack. Vivek notes power is the new bottleneck, with data center energy usage forecast to grow 11% globally through 2030.
Claudio Irigoyen, Head of Global Economics, BofA Global Research
The U.S. and China announced a deal was reached
U.S. and China officials have stated that a trade deal has been signed. Overall, we interpret these developments as further confirmation of the trade truce, which still implies high effective tariffs on China of around 40%, but lower than feared. Focus in the coming days should shift to U.S.-E.U. negotiations, with both sides signaling optimism around reaching an agreement.
Tariff de-escalation is in line with our new forecasts
We recently upgraded our global GDP growth forecast for 2025 from 2.8% to 3.0%, largely predicated on a faster and broader than expected tariff de-escalation between the U.S. and China. In China, we expect growth at 4.7% this year (vs 4.0% before), reflecting the trade truce and stronger-than-expected 4Q24 and 1Q25 data. In the U.S., we now see a lower core PCE peak at 3.1% (vs. 3.5% before) as tariffs set in. But we still see no Fed cuts this year.
U.S. effective tariffs may be converging to the new normal
With a deal sealed between the U.S. and China, we think U.S. effective tariffs will converge to around 10%, or half of what was implied by the "Liberation Day" announcements. In our baseline, we assume U.S. tariffs stay at around 40% for China and stabilize at around 5% for Canada and Mexico, and 10% for the rest of the world. And based on the latest data, U.S. customs revenue show an effective tariff rate of around 9.5% for May imports, which could signal that effective tariffs are converging to their new normal.
