KKR-Led Consortium Inks $1.3B Deal for STT GDC Data Assets

A KKR-led consortium, including Singtel, has finalized a disclosed $1.3 billion investment (https://www.reuters.com/business/kkr-led-consortium-invest-13-bln-singtels-st-telemedia-gdc-2024-06-18/) in ST Telemedia Global Data Centres (STT GDC), marking the largest digital infrastructure transaction in Southeast Asia to date. The deal, structured as an initial $1.3 billion (S$1.75 billion) (https://www.bloomberg.com/news/articles/2024-06-18/kkr-consortium-to-invest-1-3-billion-in-singtel-backed-stt-gdc) injection with options for further expansion, values the data center operator as one of the most significant AI-ready asset clusters in the Asia-Pacific region. The capital is earmarked for rapid capacity scaling across its portfolio of more than 95 data centers (https://www.sttgdc.com/our-presence), as the global race for GPU-optimized floor space reaches a fever pitch.

The Infrastructure Scramble: Data Centers as AI Factories

The investment by KKR and Singtel highlights a fundamental shift in how institutional investors view data infrastructure. No longer seen as mere real estate, facilities like those operated by STT GDC are being revalued as "AI factories" capable of hosting the massive compute loads required for Large Language Model (LLM) training and inference. The consortium’s entry follows a period of intense competition for high-density power sites. By securing a minority stake in STT GDC, KKR is positioning itself at the bedrock of the AI supply chain, where the physical scarcity of power and cooling—rather than software—has become the primary bottleneck for data-asset monetization.

Monetizing the Bio-Data Stack: SoftBank and Tempus AI

Parallel to the infrastructure boom, the market for specialized vertical data is seeing massive capital commitments. SoftBank Group has announced a disclosed $190 million (30 billion yen) (https://www.bloomberg.com/news/articles/2024-06-18/softbank-tempus-ai-to-form-30-billion-yen-joint-venture) joint venture with Tempus AI to bring advanced medical data analytics to the Japanese market. This partnership aims to leverage Tempus AI's massive library of clinical and molecular data to provide personalized medicine and diagnostic tools. For data investors, this move signals that the next wave of value lies in "high-fidelity" datasets—proprietary, regulated information that cannot be easily scraped from the public web.

The appetite for biological data assets was further confirmed today as EvolutionaryScale, an AI-for-biology startup, closed a disclosed $142 million seed round (https://www.reuters.com/technology/ai-biotech-startup-evolutionaryscale-raises-142-million-2024-06-18/). Led by Lux Capital and Nat Friedman, the funding will support the development of models capable of designing new proteins, a process entirely dependent on the acquisition and processing of vast genomic and proteomic datasets.

Strategic Acquisitions and the Rise of Sovereign AI

As the cost of data acquisition rises, hyperscalers are moving to acquire the software layers that manage data reliability. Nvidia has reportedly moved to acquire Shoreline.io, a software firm specializing in automated incident response for cloud infrastructure, for an estimated $100 million (https://techcrunch.com/2024/06/18/nvidia-to-acquire-shoreline/). This acquisition suggests that Nvidia is no longer content selling chips; it is building a comprehensive data-management ecosystem to ensure that the "data pipelines" feeding its H100 and Blackwell GPUs never experience downtime.

Simultaneously, the concept of "Sovereign AI" is driving new partnership models. Hewlett Packard Enterprise (HPE) and Nvidia have launched a joint "Private Cloud AI" solution (https://www.hpe.com/us/en/newsroom/press-release/2024/06/hpe-and-nvidia-announce-nvidia-ai-computing-by-hpe.html) designed to allow enterprises and nations to keep their data assets within their own borders. This move directly addresses the tightening global data regulation landscape, where the ability to train models on localized, secure data is becoming a competitive necessity.

Why it matters for data owners

The $1.3 billion STT GDC deal and the SoftBank-Tempus venture demonstrate that the valuation of data assets is bifurcating. While generic web data is commoditizing, the value of physical data infrastructure and specialized vertical datasets (like medical or industrial telemetry) is skyrocketing. For data owners, the message is clear: the most lucrative path to monetization lies in creating "moats" around proprietary data stacks and securing the high-density infrastructure required to process them. As capital flows toward the physical and specialized layers of the AI stack, those who control the storage and the specific context of data will hold the ultimate leverage in licensing negotiations.