Must Read Research
Also featuring commentary from our economists
September 21, 2025
Candace Browning, Head of BofA Global Research
Astronomers chart patterns in orbit, but markets remind us that even bright stars can mask black holes. This week we map disruptive growth in LEO satellites, balance AI revenue against expense reality, project a broadening in S&P earnings, and capture signals from our Global Real Estate conference.
The low-earth orbit (LEO) satellite industry is entering a new phase, with disruptive potential across wireless, wireline, defense, and many other markets.
While 2015 marked peak founding activity, many of the 374 announced constellations are now dormant. Costs to launch capacity have fallen 100x in two decades, and Communications, Communication Infrastructure and Comm Software Analyst Mike Funk cites committed capital, rapid tech evolution, a ~$200 billion TAM (total addressable market), and regulatory support as key drivers of growth. U.S. initiatives like the broadband equity access and deployment program (BEAD) further reinforce adoption. LEO complements wireless carriers by covering dead zones and enabling new subscription models, while spectrum-hosting deals (e.g. Starlink) create added opportunities. The technology poses threats to residential broadband, especially digital subscriber line (DSL), small fiber builds, and even cable in rural areas, though towers remain insulated given their adaptability. Notably, the addition of AWS-4 spectrum should boost Starlink’s competitiveness, and government contracts strengthen the business case. AI is an unknown potentially positive catalyst.
As higher depreciation costs loom, monetizing AI infrastructure investment is key to internet mega-cap performance.
Internet Analyst Justin Post believes capacity utilization will remain high until 2026 and the Street is potentially underestimating revenue upside. Historical revenue-to-capex ratios suggest strong revenue generation from infrastructure spend, yet consensus projections show the capex to forward revenue ratio falling to ~5x in 2026 vs. a 7-10x historical range. However, there are revenue timing and margin risks. First, depreciation and amortization (D&A) will ramp in 2026-27 before AI revenues fully materialize, and a mismatch could pressure margins. Second, AI servers may have shorter lifespans of 3-5 years, which could accelerate expense recognition. Third, capacity overbuild and converging large language model (LLM) performance could impact Cloud infrastructure pricing and ROI (return on investment).
Head of U.S. Equity and Quantitative Strategy Savita Subramanian raised her 2025 S&P 500 EPS (earnings per share) forecast to $271 (+12% y/y) after 2Q beats, upbeat guidance, and early signs of AI monetization.
Consensus sits at $269. Tariff mitigation outside Tech also drove strength, and she sees potential upside from trade deals and a capex cycle which could be pulled forward. She initiates 2026 EPS at $298 (+10% y/y), slightly below consensus but expects broader participation; the Mag 7’s contribution to S&P growth is expected to fall below 50% as cyclical sectors accelerate. Tail risks in 2026 skew positive, with a weaker USD (each 10% drop adds ~3ppt to EPS), robust AI/reshoring capex, and rising productivity – all supportive for earnings. Revenue per worker has already hit record highs after years of stagnation. Watch points include tariff costs, AI monetization, rates, and the labor market, but Savita sees a shift toward more evenly distributed earnings power.
Our REIT (real estate investment trust) team’s take from the BofA Global Real Estate conference in New York was positive with capital flowing, rates falling, fundamentals healthy and supply declining.
The risk is that tariffs boost consumer prices and thus we keep a balanced exposure to both risk-on and risk-off. Healthcare is seeing a pickup in investment opportunities suggesting ’25 acquisition guidance is still conservative. Job numbers have led to investor concern around apartment demand but less concern about office. A panel on NYC emphasized that “New York is back and as vibrant as ever,” and the mayoral race has so far had minimal impact on corporate leasing decisions. Performance around the last three Fed cutting cycles shows that three months after the first cut, the RMZ (MSCI U.S. REIT Index) outperformed the S&P by an average of 6%. The ’01 cycle is most similar to the current one and the RMZ outperformed by 6% over that 2+ year period.
Claudio Irigoyen, Head of Global Economics, BofA Global Research
Communication Breakdown
The Fed delivered a 25bp (basis point) cut last week, as widely expected. However, the surprise was the dovish shift in the SEP (Summary of Economic Projections). In our view, the evident dovish changes in the dots led Powell to sound hawkish and downplay their importance in the presser. The Fed's communication conundrum was evident in the whipsawing in the rates market. Our view remains that the Fed will cut 25bp in December, but the risks of an additional cut in October are high.
A lost decade for inflation?
Relative to our more cautious Fed path, market pricing is more consistent with a Fed that does not pay much attention to high inflation, pricing an increasingly lower Fed funds rate by end-2026 despite increasing inflation pricing. Perhaps the market is discounting a new dovish Fed chair and/or risks to Fed independence ahead. If that is the case, inflation will most likely remain above the target for the relevant forecasting horizon, potentially feeding into inflation expectations. According to our forecasts and even the Fed's dot plot, inflation will not have been at target for 7 years and counting.
