SF1 General Partner Sudarshan Calls for Halt on Recommendations for Power-Ready Land
Investor Sudarshan: Stop Promoting Power-Ready Land
Sudarshan, General Partner of the San Francisco fund SF1, posted a request to stop recommending power-ready land to him and instead introduce him to companies under construction that he can invest in. The preferred location is the Bay Area, followed by El Segundo, New York, or Austin.
On the same day, he also posted a procurement request: looking for at least 250 megawatts, preferably 500 megawatts, and a maximum of 2 gigawatts of power-ready land, ideally with the power shell already built. He stated that the company is price insensitive and has billions of dollars behind it. Another invested company needs 5 megawatts to 75 megawatts, preferably located in California, Nevada, or West Texas. The call to stop recommendations came after inquiries about this batch of land.
On September 7, he wrote about four directions he is exploring. The first is treating capital itself as a moat: in the era of artificial intelligence, venture capital and founders are rethinking heavy capital expenditures as a defense. Industry consolidation remains hot, with an increase in hardware and companies starting with over $100 million. The second is the shift of American dynamism from defense to re-industrialization and self-sufficiency, prioritizing the manufacturing of automated, modular factories and new energy.
The third is robotics and biotechnology, which he refers to as the new frontier. The market is waiting for the ChatGPT moment in these two industries, and he anticipates substantial progress in the next 12 to 24 months. The overarching theme is the law of scaling entering the physical world, where software code will be able to manipulate atoms. The fourth is developer infrastructure, which he believes still has over $1 trillion in value to capture. The market share for agent tools is still below 10%, and non-technical people are starting to write code, creating new infrastructure demands that did not exist during the previous linear growth by headcount.
His public resume lists him as a General Partner at SF1, having previously worked with wildfire-related Fion and AI sales company Pipeline. The power-ready land post separates investment preferences from the power procurement of invested companies: the land is production material being purchased by portfolio companies, while he personally seeks builders at any stage in the Bay Area and three alternative cities.
This is project screening, not a fund closure. The sellers are energy intermediaries promoting land parcels and power-ready landholders, while the buyers are his invested companies. Funds from his inbox are split into two paths: power capacity flows to portfolio companies with billions of dollars in backing, while equity checks flow to robotics, biotechnology, factory automation, new energy, and agent development tools. Beneficiaries are early builders who can meet in the Bay Area or El Segundo, New York, or Austin; those under pressure are intermediaries who only acquire land parcels without investable companies.
Source: Public Information
ABAB AI Insight
Heavy capital is becoming hot again, not just a new cycle slogan. In the 2010s, software funds avoided factories and inventory, while a16z later used American dynamism to group defense, aerospace, and manufacturing under the same label. After SpaceX turned El Segundo into a hardware hub, small funds began to recognize this city as an investable location. Sudarshan has disclosed rooftop integration: buying in with about $200,000 in equity plus seller notes and small business loans, then using AI to improve operations. This is his "capital as a moat" example that has already occurred, not just a pitch deck.
He has separated power procurement from equity investment. The 500 megawatts, with a cap of 2 gigawatts of power-ready land, is for capacity with existing shells and grid connection conditions, backed by portfolio companies with billions of dollars; the 5 to 75 megawatts is another batch of invested company demand. By asking outsiders to stop promoting land, he acknowledges that the land channel is already noisy, and what is lacking are builders who can spend capital. The location is limited to the Bay Area, with El Segundo, New York, and Austin as fallback options, to maintain visibility post-investment, not to sweep projects nationwide.
Comparable examples include Andreessen Horowitz's shift from military demonstrations to factories and energy, as well as Coatue and Thrive moving to power and data centers after exhausting model layers. The El Segundo City Council has also imposed a 45-day pause on new data center approvals until September 2026, indicating that the local supply of power-ready land is being stalled by municipal processes. The current phase is expanding into physical assets, not just adding another software argument.
Structurally, this is a reconstruction of the industrial chain. The mechanism is that as the marginal cost of model capabilities decreases, defenses shift to expenditures that others cannot replicate in the short term: grid-connected power, factory buildings, robots, and experimental equipment. Code can manipulate atoms, provided that there is electricity and production lines on the atomic side first. The market share for agent tools is still below 10%, indicating that the software side has not yet consumed its budget; the power-ready land has been cleared from his inbox because that order has already been taken on by the capital expenditures of portfolio companies, no longer a source for his projects.
ABAB News · Law of Cognition
As software becomes cheaper, moats return to electricity and factories.
Land parcels can be purchased on behalf, but builders cannot mass mail.
The law of scaling enters the physical world, first getting stuck on grid connection.