The Need For a Third Agricultural Revolution
Research Rundown #157, plus new memos on Ramp, Perplexity and more.
Research Rundown
Agriculture in the 21st century is already struggling to meet current global demands and may fail entirely based on future demands. Global food demand is expected to rise 51% by 2050 to feed a projected population of 9.7 billion.
The biggest breaking point in agriculture is the people upon whom the industry depends. In 2025, the median farmer was 58, and one-third of farmers were over the age of 65. Agriculture also has the highest rate of fatal occupational injuries in the US, adding to the difficulty of attracting new workers.
We need a Third Agricultural Revolution to have any hope of feeding the planet in the future. Farmers and technologists alike are struggling to find innovations in agricultural operations that will make a dent. Automation is likely the key. Farming is a function of inputs, tools, and labor. The Industrial Revolution provided better tools; the Green Revolution provided better inputs. The Third Agricultural Revolution will have to tackle labor.
Read our full deep dive here on how automation in agriculture stands to reduce reliance on strenuous labor, increase output, and lower costs.
What started as a smarter corporate card, Ramp has now evolved into a command center for business finance. Check out the full memo here to learn more.
Perplexity is quietly leading the shift from searching for answers to collaborating with them. Find out more in our memo here.
Scale is building the data infrastructure that every AI company quietly depends on. Explore our memo here for all the details.
In a landscape of short-term bets, Kraken is the rare crypto company thinking in systems, not seasons. Get the details in our memo here.
Meta announced a partnership this week with semiconductor designer Arm that will move Meta’s recommendation and feed ranking algorithms to the Arm Neoverse platform for cloud AI systems. For Arm, known for its mobile CPUs, this will be the first GPU partnership with a hyperscaler.
OpenAI also announced a partnership with Arm week to develop a CPU specifically for OpenAI’s AI Accelerator project with Broadcomm. The project is expected to deliver an AI accelerator with 10GW of compute capacity to be completed by 2029. Softbank owns 90% of Arm, and has invested a share of the $500 billion commitment to OpenAI’s Stargate project along with Oracle and other investors.
Paxos, a stablecoin issuer, erroneously issued $300 trillion of PayPal’s stablecoin PYUSD on Wednesday. The company described the incident in a tweet as an “internal technical error” which analysts have speculated may have been a human fat-finger error. The global crypto market is valued at $3.8 trillion at present.
Anthropic shared with investment bankers its intentions to acquire additional startups this week, while also revising its annualized revenue to $7 billion. To date Anthropic has invested in five companies but acquired only one, Humanloop, an LLM evaluation platform purchased for an undisclosed amount in August 2025.
Media outlets reported on Wednesday that Apple’s newly instated head of AI-driven web search, a project speculated to be Apple’s version of ChatGPT, left the company for Meta. This has further increased speculation that Apple may be looking to acquire an existing AI company. It was reported last week that the company is in late stage discussions about acquiring Prompt AI, a computer vision startup.
Palmer Luckey, cofounder of autonomous defense company Anduril, and Joe Lonsdale, co-founder of Palantir, are funding the creation of a new bank called Erebor designed to serve crypto and AI startups, among others. Erebor was conditionally approved by the Treasury Department for a national bank charter on Wednesday, and requires final approval from the FDIC before it can begin operating. The bank has drawn comparisons to Silicon Valley Bank given its promise to cater to startups with non-traditional assets.
Earlier this month, xAI announced its intentions to pursue $20 billion in financing via both debt and equity through a special purpose vehicle (SPV) that would lease Nvidia GPUs to xAI over five years for the Colossus 2 project. The funds will be raised by Valor Equity Partners, an investor in SpaceX, and $2 billion in equity funding will come from Nvidia directly.
Last week we highlighted investor concerns about Oracle in reaction to reporting that Oracle’s AI Cloud Compute for-rent had generated only a 14% margin in Q3 2025. Oracle shared this week that it projects this figure will grow to 30-40% in the next few years, and described itself as a “hypergrowth” company, projecting annual revenue to grow to $225 billion by 2030, compared to annual revenue of $53 billion in 2024.
At Contrary Research, we’ve built the best starting place to understand private tech companies. If you're interested in researching and writing about tech, apply here for our Research Fellowship!