Research & Insights | Read Case Studies | Work With Us

Hey all,

Three presentations crossed my desk last week that tell a similar story. Blackstone, Bond, and Apollo all putting out research on where capital is moving and why. Surprise, surprise… everything tracks infrastructure and AI adoption.

See below for links to the full presentations:

How the world's largest alt manager spots trends before they hit headlines

Blackstone manages $1.1T across 12,500 real estate assets, 250 companies, and 4,800 credit issuers. That portfolio size gives them pattern recognition at scale. They spot patterns early, build conviction through research, then deploy capital before trends become obvious.

The logistics case study: They went from under 1% allocation in 2007 to 40% today as e-commerce grew 10x over that period. They were buying warehouses when everyone else was building malls. At the same time, they reduced office exposure from 61% to less than 1.5% years before COVID forced the issue. The shift happened gradually but decisively. Logistics, rental housing, and digital infrastructure now represent 75% of their real estate book.

Current conviction areas: Four themes are driving their deployment today. AI infrastructure, where data volumes are up 101x and driving 16x increase in data center demand. Power generation, where they expect a 40% demand surge over the next decade. Digital economy plays around e-commerce and last-mile logistics. And life sciences, where there's a $172B annual R&D funding gap. Their track record speaks to the approach. QTS data centers grew 9x capacity under their ownership while outperforming initial projections by 15x. Jersey Mike's expanded from 324 stores in 2006 to 3,002 today, compounding at 13% annually. They've deployed $55B in India over 20 years across real estate, credit, and private equity.

View Full Presentation →

340 slides on adoption speed and the USA-China AI race

Bond has deployed $5B in technology investments and just released their analysis on AI adoption patterns. The report covers compute infrastructure, geopolitical competition, and labor market shifts.

Adoption velocity: ChatGPT reached 1 million users in 5 days versus 74 days for iPhone and years for the internet. It's now at 90% global adoption in 3 years versus 23 years for internet adoption. The comparison to historical technology adoption is striking. Ford Model T took 2,500 days to reach 1 million users, TiVo took 1,680 days, iPhone took 74 days, ChatGPT took 5. Data center construction is up 16x since 2020 to support the infrastructure requirements. AI surpassed human-level performance in 2024, scoring 92.3% versus 89.8% baseline.

The geopolitical landscape: The race is consolidating around USA and China. USA leads in large language model performance and semiconductor technology. China leads in open-source development and has installed more industrial robots than the rest of the world combined. Both governments are treating this as strategic priority with winner-take-all implications. The employment shift is becoming visible in the data. AI tech jobs increased 448% while traditional IT jobs declined 9%. Autonomous vehicles have captured 27% market share in San Francisco. OpenAI is building toward an everything app. ChatGPT's top user countries are India at 13.5%, USA at 8.9%, and Indonesia at 5.7%.

View Full Presentation →

Market concentration reaches levels not seen since the dot-com bubble

Apollo's Chief Economist published research on S&P 500 concentration with a direct assessment: "The AI bubble today is bigger than the IT bubble in the 1990s." The report quantifies concentration across market cap, returns, earnings, and capital expenditure.

Concentration metrics: Magnificent 7 stocks now represent 32% of S&P 500 market capitalization, up from roughly half that just a few years ago. They've generated 55% of all index returns since January 2021. The top 10 companies account for 31% of total S&P 500 capital expenditures, more than double their 2019 share. Technology, media, and telecom sector concentration hit 45% of the index, matching the 2000 dot-com peak. The average P/E ratio of the top 10 companies sits around 50x trailing earnings.

Capital deployment intensity: These companies are reinvesting 60% of operating cash flow into capex, which is twice the rate telecom companies deployed during their infrastructure buildout. S&P 500 price-to-sales ratio reached 3.3x, an all-time high surpassing dot-com bubble levels. Hyperscalers' capital expenditure as a share of US private domestic investment has doubled since 2023 and is significantly higher than what telecom companies were spending during the late 1990s buildout. The Magnificent 7 now account for roughly one-third of S&P 500 capex versus one-sixth just five years ago.

View Full Presentation →

Three different firms reaching the same conclusion. Massive infrastructure deployment happening now, adoption curves steeper than any historical precedent, capital allocation accelerating rather than decelerating. The question is whether productivity gains will justify current spending levels.

"Collateral Partners' research capabilities set them apart. They aggregated bespoke market data and industry intelligence that perfectly supported our investment thesis. Their ability to synthesize complex real estate fundamentals into a compelling fundraising narrative was instrumental in our capital raising success."

— Chris Powers, Founder & CEO, Fort Capital ($2.1B Transacted)

Best,

Founder, Collateral Partners

P.S. Want to see how we apply these insights? Check out our latest work.

_{Disclaimer: The companies referenced in this email are not clients. All research highlighted is publicly available material published by the respective firms. Links direct to the original publisher's content. We do not claim ownership of any materials referenced herein. All trademarks, logos, and company names belong to their respective owners. This email is for informational purposes only and should not be construed as investment advice. Any summaries or interpretations are our own and may not reflect the views of the original publishers.}