What public companies just told the SEC about AI, read straight from the filings. Every quote links the original document.
Real Estate Firm La Rosa Acquires NVIDIA B300 GPUs
La Rosa Holdings Corp. (LRHC) filed an 8-K on October 1 disclosing it had acquired NVIDIA B300 GPUs and is evaluating strategic alternatives for its real estate operations, potentially including a full divestiture. The purchase was structured as a sale-leaseback, with the GPU assets expected to generate recurring, long-term cash flows. The filing describes the transaction as advancing the company's previously announced transition into AI infrastructure.
...acquired latest-generation NVIDIA B300 GPUs, the high-performance processor at the core of modern AI data centers, and leased those assets back under a long-term lease, which is expected to generate recurring, long-term cash flows for the Company.
The filing does not disclose the number of units acquired or the purchase price. A company with real estate operations executing a GPU sale-leaseback and signaling a possible exit from its core business is a structural pivot, not a side bet.
TotalEnergies Partners on Frontier Models for Oil Reservoirs
TotalEnergies SE (TTE) disclosed in a 6-K on October 1 that it has entered a partnership to develop frontier AI models for its geosciences division. The effort draws on agentic AI to process reservoir data at scale for exploration scenario modeling.
...to combine next to 10 petaflops of data with nearly a century of knowledge, know-how and expertise accumulated by TotalEnergies in geosciences and reservoir engineering.
The filing does not name the technology partner. The objective is generating multiple scenarios for the development of exploration targets. TotalEnergies filed this on a standalone 6-K basis, indicating the company considers the partnership material to investors.
Agentic AI Reaches Board Appointments, Fintech Pivots, and ETF Names
Vertex, Inc. (VERX), the tax technology company, disclosed on October 1 that it appointed the CEO of DataRobot to its board of directors. The 8-K is the first time Vertex has used the phrase "agentic AI" in a filing, and it characterizes the incoming director's tenure at DataRobot this way:
DataRobot has transformed from a pioneer in machine learning to a leader in agentic AI, serving enterprise customers worldwide.
BNB Plus Corp. (BNBX) filed an 8-K on October 2 announcing it is abandoning its digital asset treasury strategy and pivoting to infrastructure for financial institutions deploying blockchain and agentic AI. The filing's thesis is that as these technologies mature, value shifts from holding assets to operating the infrastructure that puts them to work. On the asset management side, ETF Opportunities Trust filed an amendment registering a "Tuttle Capital Agentic AI Income Blast ETF," and Tidal Trust I filed an amendment listing a "SoFi Agentic AI ETF." Neither fund is yet active. The phrase "agentic AI" appearing across a board appointment, a corporate pivot, and two ETF registrations in a single filing window marks a shift from niche vocabulary to mainstream disclosure language.
Three Large Filers Add AI to Their Safe Harbor Risk Lists
Cigna Group (CI), Conduent Inc. (CNDT), and Iperionx Ltd. (IPX) each added a standalone AI risk item to their SEC filings on September 29 and 30. Cigna, the health insurer, listed risks related to its use of AI and machine learning alongside cybersecurity and third-party dependency in its 8-K safe harbor language. Conduent, the business process outsourcer, used nearly identical phrasing. Iperionx, a titanium materials company filing a 20-F, framed its AI risk as covering operational, cybersecurity, legal, and reputational exposure. None of the three quantifies the risk or references a specific incident. Adding language to a safe-harbor list is a low-cost disclosure move, but the pattern across industries signals that general counsel now treats AI risk as a line item investors expect to see.
Natural Gas S-1 Cites AI Power Demand as Core Investment Thesis
WhiteHawk Minerals Corp. (WHK) filed an S-1 on October 1 positioning its natural gas royalty portfolio as a direct beneficiary of AI-driven electricity demand. The filing notes that natural gas supplied roughly 41% of U.S. electricity in 2025 and names AI data center construction as one of two core demand catalysts for the company's assets. PowerCompute, Inc. (PWCM) made a related argument in an 8-K filed September 30: its Iowa and Mississippi data centers currently run on Bitcoin mining, but the company views HPC and AI workloads as higher-revenue alternatives that could potentially generate more revenue per megawatt. Both filings treat AI infrastructure demand as a durable input cost for the power and compute stack, not a cyclical spike.