Cisco Launches $1B AI Fund Backing Scale AI and Cohere

Cisco Systems has officially launched a disclosed $1 billion (https://newsroom.cisco.com/c/r/newsroom/en/us/a/press-releases/2024/06/cisco-launches-1b-global-ai-investment-fund.html) global investment fund to bolster the development of secure and reliable AI solutions, marking a significant strategic pivot toward the data-heavy infrastructure layer. The networking titan has already committed a disclosed $200 million (https://www.reuters.com/technology/cisco-launches-1-billion-ai-investment-fund-2024-06-04/) of this capital to industry leaders including Scale AI (https://scale.com/blog/series-f), which recently reached a disclosed $13.8 billion (https://scale.com/blog/series-f) valuation, as well as foundational model developers Cohere and Mistral AI. This move signals a shift in corporate venture capital (CVC) priorities, where the focus is no longer just on software applications, but on the underlying data pipelines and compute efficiency required to sustain enterprise AI.

The Data-Centric Investment Thesis

Cisco’s investment strategy focuses on the "connective tissue" of the AI stack. By backing Scale AI, Cisco is securing a stake in the world's most prominent data-labeling and curation engine, which is essential for transforming raw enterprise data into high-quality training sets. This comes as the demand for specialized datasets grows, evidenced by SAP’s disclosed $1.5 billion (https://news.sap.com/2024/06/sap-to-acquire-walkme/) acquisition of WalkMe, a deal designed to capture user-interaction data to feed SAP’s "Joule" AI assistant. The Cisco fund aims to bridge the gap between networking hardware and the data-intensive workloads of LLMs, ensuring that the flow of information across enterprise networks is optimized for model training and inference.

Infrastructure and Sovereign Data Demands

The capital influx into the AI data ecosystem is being met by a massive expansion in physical infrastructure. Specialized cloud provider CoreWeave has announced a disclosed $2.2 billion (https://www.bloomberg.com/news/articles/2024-06-03/coreweave-to-invest-2-2-billion-in-european-data-centers) investment to build three new data centers in Europe, specifically targeting Norway, Sweden, and Spain. This expansion is driven by the increasing demand for sovereign data solutions, where enterprises and governments require AI training to happen within specific legal jurisdictions. Similarly, Intel has secured a disclosed $11 billion (https://www.intel.com/content/www/us/en/newsroom/news/intel-announces-11-billion-investment-from-brookfield.html) joint venture with Brookfield to fund its Fab 34 facility in Ireland, reinforcing the compute backbone necessary to process the world's burgeoning AI data assets.

Regulation and the Licensing Frontier

As investment pours into the sector, regulators are tightening the rules on how data is harvested and utilized. In a landmark move, the New York State Legislature passed the SAFE Kids Act (https://www.governor.ny.gov/news/governor-hochul-majority-leader-stewart-cousins-and-speaker-heastie-announce-agreement-landmark), which aims to restrict the use of addictive algorithmic feeds and the unauthorized collection of minor data. This regulatory pressure is forcing AI companies to seek legitimate data sources, leading to a surge in licensing deals. Perplexity AI is reportedly in talks (https://www.reuters.com/technology/perplexity-ai-plans-revenue-sharing-deal-with-publishers-2024-06-03/) with publishers to establish a revenue-sharing model, following the path of The Atlantic and Vox Media (https://www.theatlantic.com/press-releases/archive/2024/05/the-atlantic-and-openai-partnership/678531/), who recently signed multi-year licensing agreements with OpenAI to provide high-quality editorial data for model training.

Market Valuation and Benchmarking

The market's appetite for AI-related assets remains insatiable. Nvidia’s market capitalization recently hit a record $2.8 trillion (https://www.cnbc.com/2024/06/04/nvidia-shares-hit-record-high-as-market-cap-nears-apple.html), fueled by the demand for H100 and the newly announced "Rubin" chips. To bring transparency to this rapidly evolving market, Hugging Face has launched an Open Medical LLM Leaderboard (https://huggingface.co/blog/leaderboard-medicalllm), providing a standardized benchmark for data performance in the healthcare sector. This benchmarking is critical for data owners to value their assets accurately as they enter the marketplace.

Why it matters for data owners

For data owners, the Cisco fund and the SAP acquisition underscore a fundamental shift: the enterprise is now the primary customer for high-fidelity data. As infrastructure providers like CoreWeave and Intel scale to meet compute demands, the bottleneck remains the availability of structured, legally compliant datasets. The emergence of revenue-sharing models from players like Perplexity AI suggests that the window for "free" data scraping is closing, replaced by a formal marketplace where data assets are treated as high-yield financial instruments. Owners of proprietary enterprise data, particularly in regulated industries, are now sitting on the most valuable fuel in the AI economy.