Cycle on Sep 14, 2026

I will begin the research cycle by listing the existing threads on this topic to see what questions are currently open or what watches have been registered.

I will now read the existing notes for this topic to understand what has been written so far and which slugs are already in use.

Let's pull a market view for the Real Estate sector and for some relevant tickers mentioned in the brief or in existing notes to see what the latest public market snapshot is. The brief mentions SL Green (SLG), Boston Properties (BXP), Vornado (VNO). Let's fetch the Real Estate sector view, and see if there are any specific tickers we should query. Let's do `market_view` with `kind='sector', id='real-estate'`. Let's also do `market_view` with `kind='macro'`. Let's start with these.

Let's also look at the macro backdrop to anchor our findings. We will call `market_view` with `kind='macro'`.

The 10-year Treasury yield is at 4.95% as of September 10, 2026, which is quite high and puts pressure on real estate valuations and refinancing costs. Let's look up some specific tickers: Boston Properties (BXP), SL Green (SLG), and Vornado (VNO) using `market_view` to understand their latest valuations and performance. Let's call them.

Let's also look at SL Green (SLG) and Vornado (VNO) to see their market views. This will give us a strong baseline.

Now let's check Vornado (VNO).

Great. Now let's do some web searching on the core elements of the brief: 1. "how the AI economy is splitting US metros apart — concentrating wealth, jobs, and housing demand in a handful of winners while office and commercial real estate bifurcates." 2. "metro winners vs laggards (SF/San Jose, Austin, Seattle vs older office-heavy cities)" 3. "office/CRE exposure (SL Green, Boston Properties, Vornado; the CMBS market and regional-bank CRE books)" 4. "residential exposure (homebuilders and for-sale/rental markets in AI-boom metros)" 5. "the data-center geography pulling investment into specific counties." 6. "CRE fundamentals by metro (office vacancy, CMBS delinquency, regional-bank CRE concentration), home-price and rent divergence across metros (Case-Shiller, Zillow/Redfin commentary), migration and job-posting data, and REIT/bank earnings commentary on geographic divergence." Let's run a search for recent housing and office market reports that discuss this divergence, or recent Zillow, Redfin, Case-Shiller, Trepp, or CBRE reports focusing on the AI metro boom vs laggards. We want to find concrete evidence, data, and quotes. Let's do a search.

Let's do some more searches. Let's look up CBRE office market reports for Q2 2026 or Q3 2026, or Trepp CMBS reports for 2026. Let's search for: `"office vacancy" "San Francisco" "Austin" "Seattle" CBRE 2026` or `"CMBS delinquency" Trepp office 2026`. Let's run a search.

Let's fetch the CBRE press release: "AI Investment Fuels Office Leasing Recovery Across Major Markets" (`https://www.cbre.com/press-releases/ai-investment-fuels-office-leasing-recovery-across-major-markets-us`). This looks like an incredibly relevant source for AI leasing and office recovery. Let's do a `web_fetch`.

That is a highly relevant press release from CBRE dated May 12, 2026. It notes: - Tech accounted for 16.8% of all U.S. office leasing in 2025, and jumped to 22.7% in Q1 2026. - AI companies leased 21 million sq. ft. of office space in San Francisco and Silicon Valley since 2019 (equivalent to 15 Salesforce Towers). - AI companies leased another 9.4 million sq. ft. in Manhattan, Boston, and Seattle over the same period. - "Since 2020, U.S.-based AI companies have attracted $578 billion in venture capital investment, with the San Francisco Bay Area accounting for roughly 80% of that total." This is a key quote showing extreme concentration of AI venture capital (80% in the Bay Area). - Overall office vacancy in Manhattan and San Francisco declined over the past two years, and net absorption exceeded 3% of total building inventory in 2025. Let's do some more searches to see the office vacancy and commercial real estate conditions in other metros (Austin, Seattle, older office-heavy cities) as of mid-2026, and the CMBS delinquency rates. Let's search for: `"Seattle" office vacancy Q2 2026` and `"Austin" office vacancy Q2 2026`. Let's do a search.

Let's fetch the Colliers "Puget Sound Office Report | 2026 Q2" (`https://www.colliers.com/en/research/puget-sound/puget-sound-office-report-2026-q2`) and the Colliers "Seattle Office Market Report | 2026 Q2" (`https://www.colliers.com/en/research/puget-sound/seattle-office-market-report-2026-q2`). Let's do a `web_fetch` for both. This will give us excellent primary data for Seattle. Let's start with `https://www.colliers.com/en/research/puget-sound/puget-sound-office-report-2026-q2`.

The Colliers Q2 2026 Puget Sound report shows: - Regional office vacancy decreased 30 bps to 26.0%. - Office vacancy in the Seattle CBD ended the quarter at 36.2%, amongst the highest in the U.S. - By contrast, Bellevue CBD (on the Eastside) vacancy is at 22.4%. - Regional net absorption was 497,825 SF, but Seattle net absorption was just 12,807 SF while Eastside net absorption was 411,242 SF. - This is a massive divergence between Seattle CBD (36.2% vacancy) and Bellevue CBD (22.4% vacancy), and shows Seattle CBD is still lagging while Eastside (where tech giants like Microsoft, Amazon, and others have massive footprints, and where AI firms are expanding) is seeing much stronger absorption. Let's fetch the Seattle Office Market Report | 2026 Q2 (`https://www.colliers.com/en/research/puget-sound/seattle-office-market-report-2026-q2`) to see if we can get more details on Seattle. Let's do a `web_fetch`.

This is perfect. It explicitly states: - "Availability of 32.1% has decreased due to larger leases reported this quarter led by DocuSign and Anthropic, respectively. Generative A.I. firms are now actively expanding around the region." - This is a critical piece of evidence. It shows Anthropic is expanding in Seattle as well. Let's look at Austin's office market reports for Q2 2026. Let's do a search.

Let's fetch the CBRE Austin Office Figures Q2 2026 report or Colliers Austin Office Market Report Q2 2026. Let's try fetching `https://www.colliers.com/en/research/austin/austin-office-market-report-2026-q2`. This will give us excellent primary data for Austin. Let's do a `web_fetch`.

This is very helpful. It shows that in Austin: - Vacancy rate edged down to 20.6% in Q2 2026. - Asking rents continued their downward adjustment, settling at $44.79 per square foot. - The market has shown signs of stabilization, but rents are softening. Let's do some research on the residential side: - "home-price and rent divergence across metros (Case-Shiller, Zillow/Redfin commentary)" - "migration and job-posting data" - Let's search for: `Case-Shiller home price index San Francisco Seattle Austin 2026` or `Zillow home price divergence 2026` or `Redfin housing market 2026 San Francisco Austin Seattle`. Let's run a search.

Let's fetch the PR Newswire article on the Case-Shiller Index for June 2026: `https://www.prnewswire.com/news-releases/sp-cotality-case-shiller-index-reports-annual-gain-in-june-2026-302859486.html`. Let's do a `web_fetch`. This is a fantastic primary source with exact Case-Shiller figures for June 2026.

This is incredibly rich data. Let's analyze the June 2026 Case-Shiller figures: - **San Francisco** annual change: **+3.22%** (June 2026 level: 371.01). - **Seattle** annual change: **-1.95%** (June 2026 level: 393.21). It is noted as the weakest market in the 20-city index, falling 1.95% or 2.0% YoY. - **National** index annual gain: **+1.52%**. - This shows a stark divergence between San Francisco (+3.22% YoY) and Seattle (-1.95% YoY). Note that San Francisco is seeing a housing boom driven by AI wealth (as noted in prior findings), whereas Seattle is sliding (falling -1.95% YoY, the worst-performing city in the index). - Let's check Austin. Austin is not in the Case-Shiller 20-City index (it is not one of the 20 cities, though it is often covered by other sources). Let's search for Austin home prices in 2026. - Let's search for: `"Austin" home prices Redfin Zillow 2026` or `"Austin" home prices decline 2026`. Let's do a search.

Let's fetch the article "Austin Real Estate Market Forecast 2026–2027 | See Trends" (`https://kumarawilcoxon.com/blog/austin-real-estate-market-prices-trends-and-forecast-2026-to-2027`) and "Austin Home Prices Drop in 2026: Buyer Guide" (`https://lrgrealty.com/lrg-blog/austin-home-prices-drop-2026/`). Let's do a `web_fetch` for both. This will give us excellent quantitative data for Austin's housing correction. Let's do `web_fetch` for `https://lrgrealty.com/lrg-blog/austin-home-prices-drop-2026/` first.

This is an extraordinary and comprehensive update on Austin's housing market in 2026! Let's synthesize the key figures: - **Austin Peak (May 2022):** Median home price near **$550,000**. - **Austin February 2026:** Median home price hit **$414,950** (a **24.55% decline** from peak). - **Austin March 2026 (City proper):** Median home price fell to **$399,000** (a **27.5% decline** from peak, falling nearly 7% YoY). - **National Ranking:** Austin posted the **4th largest price drop** among major U.S. cities since 2025. - **Demand Response:** Pending home sales jumped over **15%** in March 2026 across the Austin metro, indicating that buyers are returning as prices approach a structural floor. - **Underlying Forces:** Extreme oversupply (multifamily permits hit record highs in 2021-2023, active listings up 40%+ YoY by mid-2025), a cooling tech sector (layoffs, hiring freezes), return-to-office mandates (slowing inbound migration from coastal metros), and interest rate pressure (30-year mortgage rates holding near 6.5% compressing buying power). - This is a phenomenal case study of Austin's housing correction, showing a stark bifurcation compared to San Francisco's AI-driven surge. Let's look at what's happening with data centers and regional banks. The brief mentions: - "the data-center geography pulling investment into specific counties." - "CRE fundamentals by metro (office vacancy, CMBS delinquency, regional-bank CRE concentration)" - "REIT/bank earnings commentary on geographic divergence." Let's search for: `data center geography county level real estate 2026` or `data center development Virginia Ohio Georgia 2026` to see what is happening in specific counties. Let's do a search.

These are incredibly fresh and highly relevant results! Let's look at: 1. Realtor.com's report on "Data Centers and the American Housing Market" (August 2026): `https://www.realtor.com/research/august-2026-data-centers/` 2. NAR Realtor article: "Is There a 'Data-Center Effect' on Real Estate?": `https://www.nar.realtor/news/real-estate-news/is-there-a-data-center-effect-on-real-estate` 3. NAHB article: "AI Data Centers Are Outbidding Home Builders for America's Land": `https://www.nahb.org/blog/2026/07/servers-over-shelter-ai-data-centers-outbidding-home-builders-for-land` 4. JLL report: "North America Data Center Report Midyear 2026" Let's fetch `https://www.realtor.com/research/august-2026-data-centers/` first. This is a very recent report from August 2026. Let's do a `web_fetch`.

This Realtor.com report from August 2026 is an absolute goldmine! It provides quantitative evidence on the physical and geographic expansion of data centers and their intersection with local housing markets: - **Rapid Siting Expansion:** The share of American home sales near large (50+ megawatt) data centers has more than doubled since 2018 to roughly 1.5% of all home sales. The number of large data centers grew more than sevenfold over the same period, to nearly 350 large facilities. - **Siting Shift to Rural and Lower-Income Areas:** Data centers are pushing farther into rural and lower-income areas due to land and power availability. The typical new large data center opened about 23 miles from the nearest major city center from 2019-2025, but this distance rose to 27 miles in 2026 and is projected to reach 34 miles in 2027. - **Vast Buildout Scale:** Large data center activations jumped from 58 in 2024, to 83 in 2025, to **178 in 2026**, and are projected to reach **212 in 2027** (based on the construction pipeline). - **Backlash and Moratoriums:** Seattle, Nashville, Cleveland, and Durham have imposed moratoriums on new data centers. Ohio paused tax incentives. In June 2026, Monterey Park, California became the first U.S. city to outright ban data centers. In Georgia, public outcry over rising electricity bills linked to data center demand forced OpenAI to pledge that its new $30 billion campus in Effingham County would not raise local rates. - This shows a massive county-level bifurcation: data centers are pulling massive capital into specific counties (like Effingham County, GA, and rural/suburban counties 27-34 miles outside major cities), while encountering intense local backlash, bans, and moratoriums. Let's do a quick search on the banking sector and commercial mortgage-backed securities (CMBS). The brief mentions: - "office/CRE exposure (SL Green, Boston Properties, Vornado; the CMBS market and regional-bank CRE books)" - "CRE fundamentals by metro (office vacancy, CMBS delinquency, regional-bank CRE concentration)" - Let's search for: `"CMBS delinquency" Trepp office August 2026` or `Trepp CMBS delinquency rate September 2026` or `regional bank "commercial real estate" delinquency 2026`. Let's do a search.

This is incredibly precise! Let's fetch the Trepp reports: 1. `https://www.trepp.com/trepptalk/overall-cmbs-delinquency-rate-decreased-in-august-2026` 2. `https://www.trepp.com/trepptalk/cmbs-delinquency-report-july-2026` Let's do a `web_fetch` for `https://www.trepp.com/trepptalk/overall-cmbs-delinquency-rate-decreased-in-august-2026` first. This will give us the exact delinquency rates across sectors, especially office.

This is incredible! Let's examine the Trepp CMBS Delinquency data as of August 2026: - **Overall CMBS Delinquency Rate:** **7.85%** (down 1 basis point from 7.86% in July 2026). - **Office CMBS Delinquency Rate:** **12.00%** (up 9 basis points from 11.91% in July 2026). This is a massive figure, continuing to rise. - **Multifamily Delinquency Rate:** **7.69%** (unchanged). This is also elevated, reflecting refinancing friction in the multifamily sector. - **Retail Delinquency Rate:** **7.20%** (up 24 basis points). - **Lodging Delinquency Rate:** **5.84%** (up 49 basis points). - **Industrial Delinquency Rate:** **1.14%** (up 1 basis point). - **Newly Delinquent Properties:** The five largest newly delinquent loans included office towers in Chicago, Los Angeles, and Washington, D.C., and an office portfolio spanning Washington, D.C. and Northern Virginia. This shows continued severe distress in traditional, non-AI-boom office-heavy cities. Let's check the July 2026 Trepp report to see what the numbers were and get more context on the July spike. Let's do a `web_fetch` for `https://www.trepp.com/trepptalk/cmbs-delinquency-report-july-2026`. This will help us trace the trajectory.

This is extremely valuable. The July 2026 report shows: - **July CMBS Delinquency Rate:** **7.86%** (up 51 basis points from 7.35% in June 2026). - **Office CMBS Delinquency Rate:** **11.91%** (up 34 basis points). - **Seattle office portfolio:** A large Seattle office portfolio became a non-performing matured balloon loan in July 2026, which is a major driver of the delinquency spike and correlates with Seattle's high CBD vacancy rate of 36.2%. - **Chicago office tower:** Also became a non-performing matured balloon. - **Multifamily:** Increased 46 basis points in July to 7.69% as a wave of Ohio, Texas, and New York multifamily loans became 30 days delinquent. Let's do a search on regional banks and their commercial real estate exposure, or any recent regional bank earnings or regulatory filings that discuss CRE concentration or credit migration. Let's search for: `regional bank CRE concentration "delinquency" FDIC Q2 2026` or `regional bank CRE nonperforming assets 2026`. Let's do a search.

Let's fetch the FDIC speech on the Quarterly Banking Profile Second Quarter 2026: `https://www.fdic.gov/news/speeches/2026/fdic-quarterly-banking-profile-second-quarter-2026`. This is an official speech delivered recently for Q2 2026 and will give us very precise numbers on non-owner-occupied CRE past-due/nonaccrual (PDNA) rates and credit quality. Let's do a `web_fetch`.

This FDIC Q2 2026 report is an outstanding primary source for banking data! Let's pull out the critical figures: - **Non-owner occupied CRE PDNA rate for banks with assets > $250 billion:** **3.08%** in Q2 2026. This has declined for the 7th consecutive quarter from a peak of **4.99%** in Q3 2024, but is still **well above the pre-pandemic average of 0.59%**. - **Non-owner occupied CRE PDNA rate for banks with assets $100M - $1B (Community/Regional banks):** **1.48%** (down from 1.52% in Q1 2026, but up from pre-pandemic levels). - **Non-owner occupied CRE PDNA rate for banks with assets $10M - $100M:** **2.10%** in Q2 2026. - **Overall PDNA rate:** **1.44%**. - **Total Unrealized Losses on Investment Securities:** **$326.7 billion** in Q2 2026 (up slightly from $325.1 billion in Q1 2026), sitting at 5.5% of amortized cost. This is a massive overhang on bank balance sheets. - **Reserve Coverage Ratio at Community Banks:** Declined to **145.6%** in Q2 2026, driven by an increase in noncurrent loans that exceeded the increase in allowance for credit losses. This is in contrast to the broader industry (where the reserve coverage ratio rose to **172.7%**). This highlights a credit divergence where community/regional banks are experiencing worsening credit metrics relative to their reserves. Let's do a search on regional banks' CRE concentration. The brief mentions: "regional-bank CRE concentration" and "REIT/bank earnings commentary on geographic divergence." Let's search for: `Truist OR "US Bancorp" OR KeyCorp OR Huntington CRE Q2 2026` or `KRE regional bank ETF 2026`. Let's do a search.

This is a highly relevant Trepp report: "Super-Regional Bank 2Q 2026 Earnings: Commercial Loan Growth Returns as CRE Nonperformers Diverge"! Let's fetch `https://www.trepp.com/trepptalk/super-regional-bank-2q-2026-earnings-commercial-loan-growth-returns-as-cre-nonperformers-diverge`. This will give us the exact numbers and quotes on super-regional bank CRE nonperforming assets and credit migration as of Q2 2026. Let's do a `web_fetch`.

This is an outstanding and highly detailed report on super-regional banks in Q2 2026. Let's analyze the core findings: - **CRE Nonperformers Diverging:** While total charge-offs declined and credit costs improved, CRE nonperforming assets are diverging. CRE nonperformers increased at **Truist (TFC), U.S. Bancorp (USB), and KeyCorp (KEY)**. - **Asset Quality metrics:** Huntington's (HBAN) nonperforming asset ratio rose **13 basis points**, KeyCorp's (KEY) rose **11 basis points**, and Fifth Third's (FITB) portfolio nonperforming assets (NPA) ratio increased **3 basis points**. - **The Core Divergence:** The super-regional bank credit picture averages over two portfolios moving in opposite directions: C&I loans are robust and growing, which helps keep aggregate credit measures benign, while CRE nonperforming assets are quietly migrating higher and diverging across institutions. Let's check our threads to see if we can resolve any of them. The threads list has: 1. `NYC Rent-Stabilized Lease Rent Freeze Litigation Outcome` (one_shot watch, fire_date 2026-10-01). This has not fired yet (today is 2026-09-14). 2. `Q3 2026 Regional Bank Earnings and CRE Credit Migration` (ongoing watch, fire_date 2026-10-31). This has not fired yet (Q3 earnings are in mid-to-late October 2026). 3. `Q4 2026 CMBS Hard Maturity Refinancing and Payoff Rates` (one_shot watch, fire_date 2027-01-15). This has not fired yet. Wait, are there any other threads? No, those are the only three threads. Let's see if we have any other open threads or if we need to record a new thread or watch. Wait, let's look at the topic brief again: - "Track how the AI economy is splitting US metros apart — concentrating wealth, jobs, and housing demand in a handful of winners while office and commercial real estate bifurcates." - "Flag metros where office collapse and housing bifurcation happen at once, and any divergence between national averages and the metro-level reality." - "The thesis: 'the US housing/CRE market' is now several diverging markets — the AI map is the dividing line." Let's review our existing notes to see which ones we should update. The existing notes are: 1. `office-vacancy-metro-divergence-sf-seattle-austin` ("Tech Metros Diverge: San Francisco Office Rebounds via AI While Seattle and Austin Stabilize") 2. `us-housing-market-regional-fracture-case-shiller` ("Localized Housing Cycles Fracture the National Real Estate Market: Seattle and Austin Slide While San Francisco Surges") 3. `office-cmbs-delinquency-all-time-high` ("Office CMBS Delinquency Rates Hover Near Record Highs Amid Maturity Wall Crisis") 4. `regional-bank-cre-concentration-risks` ("Regional Bank Performance Diverges from Embedded Commercial Real Estate Risks") 5. `data-center-geography-county-level-bifurcation` ("Data Center Geography: County-Level Economic and Housing Bifurcation") Let's update these notes with our fresh Q2/mid-2026 data! Wait, let's double check if we can write highly targeted updates for each of these notes. Let's first review the exact content of `office-vacancy-metro-divergence-sf-seattle-austin` and `us-housing-market-regional-fracture-case-shiller`. Ah, we don't need to fetch them if we can update them directly, but let's read them first using `read_notes` to see their exact content if we can, or we can just write notes with the same slugs to update them. Wait, `read_notes` only returned the titles and previews. To be absolutely precise, let's fetch the full content of these notes. Ah, wait, there is no tool to fetch a single note's full body except that we can query `read_notes` or we can just update them based on our fresh findings. Let's do a quick read of the titles and previews: - `office-vacancy-metro-divergence-sf-seattle-austin` -> "Tech Metros Diverge: San Francisco Office Rebounds via AI While Seattle and Austin Stabilize" - `us-housing-market-regional-fracture-case-shiller` -> "Localized Housing Cycles Fracture the National Real Estate Market: Seattle and Austin Slide While San Francisco Surges" - `office-cmbs-delinquency-all-time-high` -> "Office CMBS Delinquency Rates Hover Near Record Highs Amid Maturity Wall Crisis" - `regional-bank-cre-concentration-risks` -> "Regional Bank Performance Diverges from Embedded Commercial Real Estate Risks" - `data-center-geography-county-level-bifurcation` -> "Data Center Geography: County-Level Economic and Housing Bifurcation" Let's write updates for these notes. ### Note 1: `office-vacancy-metro-divergence-sf-seattle-austin` We have new Q2 2026 data for Seattle and Austin office vacancy and leasing: - Seattle: Colliers Q2 2026 Puget Sound report shows regional office vacancy at 26.0%. Office vacancy in the Seattle CBD ended the quarter at **36.2%**, amongst the highest in the U.S. Bellevue CBD is at 22.4%. - Seattle leasing: Availability of 32.1% has decreased due to larger leases reported in Q2 2026 led by DocuSign and Anthropic. Generative AI firms are actively expanding around the region. - Austin: Colliers Q2 2026 Austin report shows vacancy rate edged down to **20.6%**, remaining below the peak reached during the same period last year, showing signs of gradual stabilization. Rents continued their downward adjustment, settling at $44.79 per sq. ft. - CBRE report (May 2026): "AI Investment Fuels Office Leasing Recovery Across Major Markets" shows tech accounted for 22.7% of all U.S. office leasing in Q1 2026. AI companies leased 21 million sq. ft. of office space in San Francisco and Silicon Valley since 2019 (equivalent to 15 Salesforce Towers). Let's write this note. Let's draft it. ```markdown # Tech Metros Diverge: San Francisco Office Rebounds via AI While Seattle and Austin Stabilize The commercial office market in major U.S. technology hubs has entered a phase of sharp geographic and structural divergence as of Q2 2026. San Francisco is experiencing a massive, AI-driven leasing surge, while Seattle and Austin are navigating elevated vacancy rates with early signs of stabilization. The AI-driven structural bifurcation is highly visible in office leasing data. According to CBRE's May 2026 analysis, artificial intelligence companies have leased 21 million square feet of office space in San Francisco and Silicon Valley since 2019—equivalent to 15 Salesforce Towers. Large AI leases helped push tech's share of total U.S. office leasing to 22.7% in Q1 2026, outpacing all other industries. ### Seattle CBD Vacancy Reaches 36.2% Amid Eastside Divergence While San Francisco rebounds, Seattle is grappling with some of the highest vacancy rates in the nation, though AI expansion is beginning to provide a cushion. According to the Colliers Q2 2026 Puget Sound Office Report, regional vacancy ended the quarter at 26.0%, but the Seattle Central Business District (CBD) recorded a vacancy rate of **36.2%**, amongst the highest in the country. By contrast, the Bellevue CBD on the Eastside recorded a vacancy rate of 22.4%. Regional net absorption rebounded sharply to 497,825 square feet in Q2 2026, but this was heavily skewed: Seattle CBD net absorption was just 12,807 square feet, while Eastside net absorption reached 411,242 square feet. Despite these headwinds, Colliers notes that Seattle CBD availability has decreased slightly due to major leases signed by DocuSign and generative AI firm Anthropic, indicating that "Generative A.I. firms are now actively expanding around the region." ### Austin Shows Gradual Stabilization at 20.6% Vacancy Austin's office market is navigating a challenging post-boom oversupply but showed clear signs of stabilization in mid-2026. The Colliers Q2 2026 Austin Office Market Report shows that overall vacancy edged down to **20.6%** in Q2 2026, down from its peak last year. Net absorption rebounded into positive territory, though leasing momentum remained soft. Landlords continue to adjust pricing to attract tenants, with average direct asking rents softening to $44.79 per square foot. ## Sources - [AI Investment Fuels Office Leasing Recovery Across Major Markets](https://www.cbre.com/press-releases/ai-investment-fuels-office-leasing-recovery-across-major-markets-us) — CBRE press release on AI venture capital and office leasing, May 12, 2026. - [Puget Sound Office Report | 2026 Q2](https://www.colliers.com/en/research/puget-sound/puget-sound-office-report-2026-q2) — Colliers report on Puget Sound regional office vacancy and absorption, July 9, 2026. - [Seattle Office Market Report | 2026 Q2](https://www.colliers.com/en/research/puget-sound/seattle-office-market-report-2026-q2) — Colliers report on Seattle leasing, Anthropic expansion, and construction activity, July 9, 2026. - [Austin Office Market Report | 2026 Q2](https://www.colliers.com/en/research/austin/austin-office-market-report-2026-q2) — Colliers report on Austin office vacancy, net absorption, and rents, July 20, 2026. ``` Let's call `write_note` on this. I will state my intent first. I am going to update the `office-vacancy-metro-divergence-sf-seattle-austin` note with the latest Q2 2026 office market data for Seattle, Bellevue, and Austin, citing Colliers and CBRE reports.

That was successful. Now, let's write an update for the note: `us-housing-market-regional-fracture-case-shiller`. We have new data: - S&P Case-Shiller June 2026 results released August 25, 2026: - National Home Price NSA Index posted a 1.52% annual gain for June 2026. - San Francisco Case-Shiller index posted a **+3.22%** annual gain in June 2026 (June level: 371.01). - Seattle Case-Shiller index posted a **-1.95%** annual decline in June 2026 (June level: 393.21), recorded as the weakest return among the 20 cities in the index. - This shows a stark geographic divide: San Francisco continues to surge (+3.22% YoY) while Seattle slides (-1.95% YoY). - Austin Housing Market (as of mid-2026): - According to Zillow and local market reports, Austin home prices have dropped **24.55%** from their May 2022 peak (near $550,000) to **$414,950** in February 2026. - Within Austin city limits, the median home price fell to **$399,000** in March 2026, representing a **27.5% decline** from peak and a nearly 7% year-over-year decline. - However, pending sales across the Austin metro surged by over **15%** in March 2026, indicating that buyers are returning to absorb discounted inventory as prices approach a structural floor. - This highlights the thesis: the U.S. residential market is deeply fractured, and national averages (+1.52% YoY) mask severe regional corrections (Austin down 24.5% from peak, Seattle down 1.95% YoY) and AI-wealth-driven booms (San Francisco up 3.22% YoY). Let's compose the note body for `us-housing-market-regional-fracture-case-shiller`. ```markdown # Localized Housing Cycles Fracture the National Real Estate Market: Seattle and Austin Slide While San Francisco Surges The U.S. residential housing market is fracturing into highly localized, diverging regional cycles, rendering national averages increasingly deceptive. While the national home price index shows modest, stable growth, individual tech and AI hubs are moving in opposite directions: San Francisco is surging on liquid AI wealth, while Seattle and Austin are undergoing deep price corrections. ### Case-Shiller June 2026: Seattle Slides as San Francisco Outpaces National Average According to the S&P Case-Shiller Index results for June 2026 (released August 25, 2026), the U.S. National Home Price Index posted a **1.52% annual gain**. However, this average masks a massive "nearly nine percentage point gap" separating the strongest and weakest markets in the country. Within the technology and AI gateway metros, the divergence is stark: * **San Francisco** recorded a robust **+3.22% annual gain** in June 2026 (Index level: 371.01), driven by a massive wave of liquid wealth generated by the local artificial intelligence boom. * **Seattle** recorded a **-1.95% annual decline** in June 2026 (Index level: 393.21), making it the worst-performing market among the 20 major metros in the Case-Shiller index. ### Austin's Deep Correction Reaches 24.5% Off Peak, But Buyers Return Austin, which experienced an unsustainable housing boom during the pandemic, has undergone one of the deepest residential corrections in the nation. According to local real estate and Zillow data, the typical Austin home value sitting at **$414,950** in February 2026 represents a **24.55% decline** from the May 2022 peak of $550,000—a drop of roughly $135,000 in typical home value. Within the city limits proper, the median home price fell nearly 7% year-over-year in March 2026, settling at **$399,000** (a 27.5% decline from peak). This correction was driven by a structural oversupply (with multifamily permits hitting record highs from 2021-2023), tech sector cooling, and a reversal of remote work migration as return-to-office mandates spread. However, the correction is showing signs of finding a floor. Pending home sales across the Austin metro surged by more than **15%** in March 2026 despite the price declines. This divergence—rising sales volume meeting falling prices—suggests that discounted inventory is being absorbed rapidly by sidelined buyers waiting for an entry point. ## Sources - [S&P Cotality Case-Shiller Index Reports Annual Gain in June 2026](https://www.prnewswire.com/news-releases/sp-cotality-case-shiller-index-reports-annual-gain-in-june-2026-302859486.html) — S&P Dow Jones Indices press release with June 2026 Case-Shiller data, August 25, 2026. - [Austin Home Prices Drop in 2026: Buyer Guide](https://lrgrealty.com/lrg-blog/austin-home-prices-drop-2026/) — Detailed analysis of Austin's 24.55% price drop, oversupply factors, and March 2026 pending sales surge, updated July 28, 2026. ``` Let's call `write_note` on this. I will state my intent first. I am going to update the `us-housing-market-regional-fracture-case-shiller` note with the June 2026 Case-Shiller figures and Austin's 2026 price correction and pending sales data.

Excellent! Now, let's write an update for the note: `office-cmbs-delinquency-all-time-high`. We have new data: - Trepp CMBS Delinquency Reports for July and August 2026: - Overall CMBS Delinquency Rate was **7.86%** in July 2026 (a 51 basis point spike from June's 7.35%) and ended August 2026 at **7.85%**. - **Office CMBS Delinquency Rate** reached **11.91%** in July and climbed to **12.00%** in August 2026. This is an extremely high level, reflecting severe distress in the office sector. - July delinquency spike was heavily driven by a group of very large loans that became non-performing matured balloon or foreclosure, including a **Seattle office portfolio** and a Chicago office tower. - August newly delinquent loans included office towers in Chicago, Los Angeles, and Washington D.C., and an office portfolio spanning Washington D.C. and Northern Virginia. - Multifamily CMBS Delinquency Rate stood at **7.69%** in July and August, driven by a wave of Ohio, Texas, and New York multifamily loans becoming 30 days delinquent. Let's compose the note body for `office-cmbs-delinquency-all-time-high`. ```markdown # Office CMBS Delinquency Rates Reach 12.0% Amid Maturity Wall Crisis The commercial mortgage-backed securities (CMBS) market is experiencing severe structural distress, with office and multifamily properties bearing the brunt of refinancing friction. Trepp's data through August 2026 reveals that delinquency rates are hovering near record highs as a massive wave of loans hit their hard maturity walls. ### Office Delinquency Reaches 12.00% in August 2026 The **Office CMBS Delinquency Rate reached 12.00%** in August 2026, up from 11.91% in July 2026 and continuing a multi-month upward trajectory. This means nearly one in eight securitized office loans is now delinquent. The distress is heavily concentrated in large, central business district (CBD) office properties that have failed to pay off at maturity. According to Trepp, the five largest newly delinquent loans in August included office towers in Chicago, Los Angeles, and Washington, D.C., as well as an office portfolio spanning Washington, D.C. and Northern Virginia. ### July Spike Driven by Seattle and Chicago Matured Balloons In July 2026, the overall CMBS delinquency rate spiked by **51 basis points to 7.86%** (before edging down 1 basis point to 7.85% in August). This dramatic increase was led by several very large loans whose status moved to non-performing matured balloon or foreclosure. These newly delinquent balances totaled $6.0 billion, with the five largest loans accounting for 44% ($2.6 billion) of the total. Crucially, the newly delinquent July cohort was led by: * A **Seattle office portfolio** that became a non-performing matured balloon loan, highlighting how Seattle's high office vacancy rate (36.2% in the CBD) is directly feeding into capital market distress. * A major Chicago office tower that failed to refinance. * An Arlington, Virginia office tower that became 30 days delinquent. ### Multifamily Delinquencies Hold at 7.69% The stress is not confined to offices. The **Multifamily CMBS Delinquency Rate rose to 7.69%** in July 2026 and remained unchanged in August. This reflects a wave of defaults among multifamily properties in Ohio, Texas, and New York, where rapid post-pandemic supply growth and high interest rates (with the 10-year Treasury yield at 4.95% as of September 2026) have squeezed debt service coverage ratios. ## Sources - [CMBS Delinquency Rate Decreased One Basis Point in August 2026](https://www.trepp.com/trepptalk/overall-cmbs-delinquency-rate-decreased-in-august-2026) — Trepp report on August 2026 CMBS delinquencies, showing office delinquency hitting 12.00%, September 1, 2026. - [CMBS Delinquency Rate Increased 51 Basis Points in July 2026](https://www.trepp.com/trepptalk/cmbs-delinquency-report-july-2026) — Trepp report on July 2026 CMBS delinquencies, detailing the Seattle office portfolio matured balloon, July 31, 2026. - [U.S. macro backdrop — market view](/markets/macro/2026/09/14) — Public markets view showing the 10-year Treasury yield at 4.95% as of September 10, 2026. ``` Let's call `write_note` on this. I will state my intent first. I am going to update the `office-cmbs-delinquency-all-time-high` note with the July and August 2026 Trepp delinquency figures, highlighting the Seattle office portfolio default.

Excellent! Now, let's write an update for the note: `regional-bank-cre-concentration-risks`. We have new data: - FDIC Q2 2026 Quarterly Banking Profile (released August 25, 2026): - Overall past-due and nonaccrual (PDNA) rate decreased to **1.44%**. - Non-owner occupied CRE past-due and nonaccrual (PDNA) rate for banks with assets > $250 billion declined to **3.08%** in Q2 2026 (down from a peak of 4.99% in Q3 2024, but still **well above the pre-pandemic average of 0.59%**). - Total unrealized losses on investment securities remained elevated at **$326.7 billion** (up from $325.1 billion in Q1 2026), representing 5.5% of amortized cost. This is a massive capital overhang on regional and community bank balance sheets. - Reserve Coverage Ratio at Community Banks: Declined to **145.6%** in Q2 2026, driven by an increase in noncurrent loans that exceeded the increase in allowance for credit losses. This is in contrast to the broader industry (where the reserve coverage ratio rose to **172.7%**). - Super-Regional Bank Earnings Q2 2026 (Trepp report, July 30, 2026): - Credit costs improved broadly, but CRE nonperforming assets are diverging across institutions. - CRE nonperforming balances increased at **Truist (TFC), U.S. Bancorp (USB), and KeyCorp (KEY)**. - Huntington's (HBAN) nonperforming asset ratio rose **13 basis points**, KeyCorp's (KEY) rose **11 basis points**, and Fifth Third's (FITB) portfolio nonperforming assets (NPA) ratio increased **3 basis points**. - This highlights a credit divergence where C&I loans are robust and growing (helping keep aggregate credit measures benign), while CRE nonperforming assets are quietly migrating higher and diverging across institutions. Let's compose the note body for `regional-bank-cre-concentration-risks`. ```markdown # Regional Bank Performance Diverges from Embedded Commercial Real Estate Risks While super-regional bank earnings appeared robust in Q2 2026, a deeper look at credit quality reveals a growing divergence. Headline asset quality remains benign, but this averages over two portfolios moving in opposite directions: robust commercial and industrial (C&I) lending is masking a quiet and highly divergent migration of commercial real estate (CRE) assets into nonaccrual and nonperforming status. ### FDIC Q2 2026: Non-Owner Occupied CRE Delinquencies Elevated at 3.08% According to the FDIC's Q2 2026 Quarterly Banking Profile (released August 25, 2026), the banking industry reported strong quarterly net income of $90.1 billion, with an overall past-due and nonaccrual (PDNA) rate of 1.44%. Within the commercial real estate portfolio, the non-owner-occupied CRE PDNA rate for the largest institutions (assets > $250 billion) declined to **3.08%** in Q2 2026. While this is down from its peak of 4.99% in Q3 2024, it remains **well above the pre-pandemic average rate of 0.59%**. Furthermore, banks continue to carry a massive capital overhang: total unrealized losses on held-to-maturity and available-for-sale investment securities portfolios rose slightly to **$326.7 billion** (representing 5.5% of amortized cost). A key credit divergence is emerging at smaller institutions: the reserve coverage ratio at community banks declined to **145.6%** in Q2 2026, driven by an increase in noncurrent loans that outpaced credit loss allowances. This is in direct contrast to the broader industry, where the reserve coverage ratio rose to 172.7%. ### Q2 2026 Earnings: CRE Nonperformers Diverge Across Super-Regionals Super-regional bank earnings for Q2 2026 confirmed this credit divergence. While net charge-off ratios declined and allowance for credit losses fell at most institutions, nonperforming assets began to migrate higher at several banks. According to Trepp's analysis of Q2 2026 earnings: * **Truist (TFC), U.S. Bancorp (USB), and KeyCorp (KEY)** all reported higher CRE nonperforming balances even as total credit measures improved. * **Huntington (HBAN)** reported a **13 basis point increase** in its nonperforming asset ratio. * **KeyCorp (KEY)** reported an **11 basis point increase** in its nonperforming asset ratio. * **Fifth Third (FITB)** reported a **3 basis point increase** in its portfolio nonperforming asset (NPA) ratio. These figures show that the commercial office correction remains an active drag on regional bank balance sheets, with distress quietly migrating and diverging across institutions rather than resolving uniformly. ## Sources - [FDIC Quarterly Banking Profile Second Quarter 2026](https://www.fdic.gov/news/speeches/2026/fdic-quarterly-banking-profile-second-quarter-2026) — FDIC speech on Q2 2026 banking performance, detailing the 3.08% non-owner occupied CRE PDNA rate and $326.7 billion in unrealized losses, August 25, 2026. - [Super-Regional Bank 2Q 2026 Earnings: Commercial Loan Growth Returns as CRE Nonperformers Diverge](https://www.trepp.com/trepptalk/super-regional-bank-2q-2026-earnings-commercial-loan-growth-returns-as-cre-nonperformers-diverge) — Trepp report on super-regional bank earnings, detailing CRE nonperforming loan increases at Truist, U.S. Bancorp, KeyCorp, and Huntington, July 30, 2026. ``` Let's call `write_note` on this. I will state my intent first. I am going to update the `regional-bank-cre-concentration-risks` note with the Q2 2026 FDIC banking profile data and super-regional bank earnings commentary on CRE nonperforming asset migration.

Excellent! Now, let's write an update for the note: `data-center-geography-county-level-bifurcation`. We have new data: - Realtor.com's report "Data Centers and the American Housing Market" (August 2026): - Siting Scale: The share of American home sales near large (50+ MW) data centers has more than doubled since 2018 to **1.5% of all home sales**. - Siting Surcharges & Locations: The number of large data centers grew more than sevenfold over the same period to **nearly 350 large facilities**. Large data center activations jumped from 58 in 2024, to 83 in 2025, to **178 in 2026**, and are projected to reach **212 in 2027** based on the construction pipeline. - Siting Distance: The typical new large data center opened about 23 miles from the nearest major city center from 2019-2025, but this distance rose to **27 miles in 2026** and is projected to reach **34 miles in 2027**. This highlights how data centers are expanding farther into rural and lower-income communities. - Backlash & Moratoriums: Seattle, Nashville, Cleveland, and Durham have imposed moratoriums on new data centers. Ohio paused tax incentives. In June 2026, Monterey Park, California became the first U.S. city to outright ban data centers. - In Georgia, public outcry over rising electricity bills linked to data center demand forced OpenAI to pledge that its new **$30 billion campus in Effingham County** would not raise local rates. - This shows a massive county-level bifurcation: data centers are pulling massive capital into specific counties (like Effingham County, GA, and rural/suburban counties 27-34 miles outside major cities), while encountering intense local backlash, bans, and moratoriums. Let's compose the note body for `data-center-geography-county-level-bifurcation`. ```markdown # Data Center Geography: County-Level Economic and Housing Bifurcation The rapid expansion of artificial intelligence infrastructure is driving a historic capital investment supercycle, but its physical footprint is highly concentrated. The hunt for power-enabled, fiber-rich land is pulling massive capital into specific counties, creating a localized county-level economic and housing bifurcation, while simultaneously triggering intense regulatory and community backlash. ### The Scale of Siting: Activations Triple in 2026 According to Realtor.com's August 2026 report, "Data Centers and the American Housing Market," the physical footprint of the digital economy has expanded exponentially. The share of American home sales taking place within five miles of a large (at least 50-megawatt) data center has more than doubled since 2018, reaching **1.5% of all U.S. home sales**. The number of large data centers grew more than sevenfold over this period to **nearly 350 large active facilities**. The pace of construction is accelerating rapidly under the AI investment boom: * Large data center activations jumped from 58 in 2024 to 83 in 2025. * In **2026, activations are projected to hit 178 facilities**. * The pipeline points to **212 activations in 2027**. ### Siting Shifts Outward to Rural and Lower-Income Counties As power grids in established hubs (like Northern Virginia) reach capacity, data center developers are pushing farther into rural and lower-income territory to secure land and electricity. The typical new large data center opened about 23 miles from the nearest major city center from 2019 through 2025. However, this distance rose to **27 miles in 2026** and is projected to reach **34 miles in 2027**—a 48% increase in distance from urban cores in just two years. Furthermore, while data centers historically located in high-income suburbs (with ZIP code median incomes peaking at 24.7% above the national median in 2023), the median 2026 and 2027 openings are located in ZIP codes **below the national median income** for the first time in seven years (-2.1% in 2026 and -5.7% in 2027). ### Localized Capital Influx vs. Intense Community Backlash The massive capital influx can transform local county economies. For instance, OpenAI is developing a **$30 billion campus in Effingham County, Georgia**. However, this unprecedented concentration of power and capital has triggered severe local resistance. In Effingham County, public outcry over rising electricity bills linked to data center power demand forced OpenAI to pledge that its campus would not raise local utility rates. Nationally, the backlash is hardening: * Seattle, Nashville, Cleveland, and Durham have imposed moratoriums on new data centers. * Ohio paused its industry tax incentives. * In June 2026, **Monterey Park, California became the first U.S. city to outright ban data centers**, citing noise, water usage, and infrastructure strain. * Lawsuits and administrative delays blocked or delayed at least 75 data center projects worth an estimated **$130 billion** in the first three months of 2026 alone. This creates a sharp geographical divide: a handful of rural and suburban counties are receiving billions of dollars in digital infrastructure capital, while other jurisdictions are slamming the door shut through bans and moratoriums. ## Sources - [Data Centers and the American Housing Market](https://www.realtor.com/research/august-2026-data-centers/) — Realtor.com Economic Research report analyzing millions of home sales and satellite-tracked data center activations, August 11, 2026. - [Is There a 'Data-Center Effect' on Real Estate?](https://www.nar.realtor/news/real-estate-news/is-there-a-data-center-effect-on-real-estate) — National Association of Realtors summary of county-level data center impacts, September 9, 2026. ``` Let's call `write_note` on this. I will state my intent first. I am going to update the `data-center-geography-county-level-bifurcation` note with the August 2026 Realtor.com report data, highlighting the physical scaling of data centers, their shift into rural/lower-income counties, and the emerging regulatory backlash (including Monterey Park's ban and OpenAI's Effingham County project).

Excellent. All 5 notes have been updated and are now fully in sync with the latest mid-to-late 2026 data. Let's double-check if there are any other notes we should write or if we should register any watches. Wait, let's think about if there are any other forward-looking catalysts we should watch for. The user's brief asks us to: "Flag metros where office collapse and housing bifurcation happen at once, and any divergence between national averages and the metro-level reality. The thesis: 'the US housing/CRE market' is now several diverging markets — the AI map is the dividing line." Let's register a watch for the next Case-Shiller release or the next Trepp CMBS Delinquency report. Or perhaps a watch for a specific regional bank's Q3 earnings. We already have: 1. `Q3 2026 Regional Bank Earnings and CRE Credit Migration` (ongoing watch, fire_date 2026-10-31). This tracks Q3 2026 earnings for super-regional banks. 2. `Q4 2026 CMBS Hard Maturity Refinancing and Payoff Rates` (one_shot watch, fire_date 2027-01-15). 3. `NYC Rent-Stabilized Lease Rent Freeze Litigation Outcome` (one_shot watch, fire_date 2026-10-01). Let's register a new watch for the next S&P Case-Shiller release in late September 2026 (covering July 2026 data) to see if Seattle's housing slide and San Francisco's surge continue to diverge. Let's call `record_watch`. I will state my intent first. I am going to record a new watch for the upcoming S&P Case-Shiller Home Price Index release in late September 2026, which will track whether the divergence between San Francisco and Seattle continues to widen.

Fantastic! Let's double check if there are any other tools we should call or if we are ready to finish the cycle. We have: - Listed the threads. - Pulled sector and macro market views. - Pulled ticker market views for BXP, SLG, and VNO. - Searched the web for recent data on office vacancy, housing price index, data center geography, and regional banks. - Updated 4 major notes under their stable slugs (`office-vacancy-metro-divergence-sf-seattle-austin`, `us-housing-market-regional-fracture-case-shiller`, `office-cmbs-delinquency-all-time-high`, `regional-bank-cre-concentration-risks`, `data-center-geography-county-level-bifurcation`). - Registered a new watch for the next Case-Shiller release to keep tracking the tech gateway housing divergence. We are ready to end the research cycle. Let's call `finish_cycle`. I will state my intent first. I am ending the research cycle now by calling `finish_cycle` since all findings have been written and a new watch has been registered.

I have completed the research cycle on the divergence of U.S. metropolitan markets under the influence of the AI economy. This cycle has successfully integrated the latest mid-to-late 2026 data across office leasing, residential housing, capital markets (CMBS), regional bank balance sheets, and data center geography. Here is a summary of the updates and findings recorded in this cycle: ### 1. Technology Office Metros Diverge Further * **San Francisco & Silicon Valley:** CBRE's May 2026 analysis shows that AI companies have leased **21 million sq. ft.** of office space in San Francisco and Silicon Valley since 2019 (equivalent to 15 Salesforce Towers). Large AI leases drove tech's share of total U.S. office leasing to **22.7% in Q1 2026**. * **Seattle CBD:** Vacancy in the Seattle Central Business District reached **36.2%** as of Q2 2026 (amongst the highest in the U.S.), according to Colliers. However, there is a micro-geographic split within the metro: Bellevue CBD (on the Eastside) is at 22.4% vacancy, and Puget Sound's Q2 absorption was heavily skewed to the Eastside (411,242 sq. ft. vs. just 12,807 sq. ft. in Seattle CBD). Seattle is seeing some relief as generative AI firms like Anthropic expand in the region. * **Austin:** Austin's office market is stabilizing, with vacancy edging down to **20.6%** in Q2 2026, though landlords continue to adjust pricing downward, with average asking rents softening to $44.79 per sq. ft. * *Updated Note:* `[[office-vacancy-metro-divergence-sf-seattle-austin]]` ### 2. Residential Housing Markets Fracture * **Case-Shiller June 2026:** Released on August 25, 2026, the S&P Case-Shiller National Index posted a **1.52% annual gain**. However, this average masks a massive regional gap: **San Francisco surged +3.22% YoY** on liquid AI wealth, while **Seattle fell -1.95% YoY**, making it the worst-performing market in the 20-city index. * **Austin Housing Correction:** The typical Austin home value sitting at **$414,950** in February 2026 represents a **24.55% decline** from its May 2022 peak ($550,000). Within the city limits proper, the median price fell nearly 7% YoY to **$399,000** in March 2026 (27.5% off peak). However, pending sales surged **15%** in March, indicating that buyers are aggressively returning to absorb discounted inventory as prices approach a structural floor. * *Updated Note:* `[[us-housing-market-regional-fracture-case-shiller]]` ### 3. Office CMBS Delinquencies Climb to 12.0% * **Office CMBS:** The Office CMBS Delinquency Rate climbed to **12.00%** in August 2026, up from 11.91% in July. * **Maturity Balloon Defaults:** The overall CMBS delinquency rate spiked by **51 basis points in July to 7.86%** before leveling off at 7.85% in August. This spike was heavily driven by large loans failing to refinance at maturity, most notably a **Seattle office portfolio** and a Chicago office tower. * **Multifamily Stress:** Multifamily CMBS delinquencies held steady at **7.69%** in July and August, driven by a wave of defaults in Ohio, Texas, and New York. * *Updated Note:* `[[office-cmbs-delinquency-all-time-high]]` ### 4. Regional Bank Credit Quality Diverges * **FDIC Q2 2026 Profile:** Released August 25, 2026, the FDIC reports that the non-owner occupied CRE past-due/nonaccrual (PDNA) rate for banks with assets > $250 billion sat at **3.08%** (down from a 4.99% peak in Q3 2024, but still well above the 0.59% pre-pandemic average). Total unrealized losses on bank investment securities remained elevated at **$326.7 billion**. * **Community Bank Reserves:** The reserve coverage ratio at community banks declined to **145.6%** in Q2 2026, driven by rising noncurrent loans, contrasting with the broader industry's rise to 172.7%. * **Super-Regional Earnings:** While overall credit costs improved, CRE nonperforming assets are quietly migrating higher and diverging. Truist (TFC), U.S. Bancorp (USB), and KeyCorp (KEY) all reported higher CRE nonperforming balances in Q2, while Huntington's (HBAN) nonperforming asset ratio rose **13 basis points** and KeyCorp's rose **11 basis points**. * *Updated Note:* `[[regional-bank-cre-concentration-risks]]` ### 5. Data Center Siting Shifts Farther Outward Amid Local Backlash * **Unprecedented Siting Scale:** Realtor.com's August 2026 report reveals that **1.5% of all U.S. home sales** now occur within 5 miles of a large (50+ MW) data center. Activations are projected to hit **178 in 2026** and **212 in 2027** (up from 58 in 2024). * **Push to Rural & Lower-Income Counties:** The typical new large data center opened **27 miles from the nearest city center** in 2026 and is projected to reach **34 miles in 2027** (up from 23 miles historically). For the first time in seven years, median data center ZIP code incomes are falling below the national median (-2.1% in 2026). * **Backlash & Bans:** Regulatory pushback is intensifying. Seattle, Nashville, Cleveland, and Durham have active data center moratoriums. In June 2026, **Monterey Park, CA became the first U.S. city to outright ban data centers**. In Georgia, public outcry over rising utility rates forced OpenAI to pledge that its new **$30 billion campus in Effingham County** would not raise local rates. * *Updated Note:* `[[data-center-geography-county-level-bifurcation]]` ### 6. Forward-Looking Monitoring (New Watch Registered) * **New Watch Registered:** `September 2026 Case-Shiller Tech Gateway Divergence` (scheduled for late September 2026) to track whether the geographic divide between San Francisco's AI wealth surge and Seattle's housing slide continues to widen.