From blockbuster AI deals and hyperscaler land grabs to mounting political scrutiny over power costs, it’s been an eventful end to 2025 for the data center industry. Plus, which operator is pushing AI training compute capacity to a staggering 2GW? Here’s a roundup of the latest news shaping the data center landscape.
AI Mega Deals and Strategic Shifts
The AI infrastructure race is entering a new phase of consolidation. Nvidia announced that it’s acquiring the assets and leadership of AI chip startup Groq in a roughly $20 billion deal – its largest ever – bringing in technology designed to accelerate real-time AI inference, not just training. While structured as a “non-exclusive licensing agreement,” the deal effectively deepens Nvidia’s control over a critical choke point in AI workloads. Analysts have cautioned that the deal’s non-exclusive licensing structure appears aimed at sidestepping antitrust scrutiny while preserving the appearance of competition.
Meanwhile, Japan’s SoftBank is acquiring data center investment firm DigitalBridge for $4 billion. SoftBank CEO Masayoshi Son is framing the deal as foundational to next-generation AI infrastructure. That urgency is underscored by SoftBank’s parallel scramble to raise $22.5 billion to pay OpenAI – a funding commitment that has already required the company to sell stakes in Nvidia and T-Mobile, according to Reuters.
And after years of regulatory limbo, ByteDance has agreed to cede TikTok US under the control of an Oracle-led consortium – cementing US-based operational control as the price of continued access to the market.
Political Heat Fueling Renewed AI Backlash
The rapid growth of data centers is now drawing attention from US politicians at every level. In a rare bipartisan alignment, Senator Bernie Sanders and Florida Governor Ron DeSantis have both criticized the AI data center boom, citing concerns over electricity prices, grid stability, and labor impacts. That convergence suggests opposition to large-scale AI infrastructure could soon extend beyond activist groups and local-level pushback into mainstream policy debates ahead of the 2026 midterms.
At the federal level, three Democratic senators – Elizabeth Warren, Chris Van Hollen, and Richard Blumenthal – have launched an inquiry into whether hyperscalers are driving up consumer electricity costs, demanding transparency from companies including Google, Microsoft, Amazon, and Equinix. The Energy Information Administration shows prices have gone up by 25% nationwide from 2020 to 2024, but that the increases are driven by multiple factors.
Amazon, meanwhile, is pushing back on claims that data centers are driving up utility bills. A report commissioned by AWS argues that some facilities can actually reduce costs for other ratepayers through thoughtful rate design and infrastructure contributions. Whether regulators will accept that framing remains to be seen.
At the same time, the Trump administration announced plans for a 1,000-person “US Tech Force” to accelerate AI deployment across federal agencies, backed by nearly every major hyperscaler and AI vendor. Engineers, data scientists, and AI specialists are expected to be brought on for a period of two years starting March 31, 2026.
Energy, Sustainability, and Next-Gen Power
Faced with mounting scrutiny and constrained grids, hyperscalers are increasingly moving upstream. Alphabet’s $4.75 billion acquisition of data center and energy infrastructure company Intersect Power gives Google access to several gigawatts of energy and data center projects.
Hydrogen producer Verne has signed a long-term hydrogen supply deal with on-site power and cooling provider Vema to support on-site, low-emissions power generation – another signal that alternatives to grid dependence are gaining traction. Under the terms of the ten-year agreement, Vema intends to expand its annual hydrogen output to more than 36,000 tons.
In other next-gen energy news, Google-backed nuclear fusion firm TAE Technologies has merged with Trump Media & Technology Group in a $6 billion deal. The company plans to build the world’s first utility-scale fusion plant at roughly 50MW, with larger follow-on facilities envisioned.
Lastly, Adani Group – one of India’s largest companies – is exploring nuclear power to support its AI data center ambitions, acknowledging that traditional grids may not scale fast enough to meet future demand.
Major US Expansions
On the development front, scale remains staggering. Elon Musk’s xAI is expanding its Memphis campus toward nearly 2GW of AI training capacity, while Vantage has broken ground on its $15 billion, 902MW Stargate campus in Wisconsin as part of Oracle and OpenAI’s initiative.
Microsoft also announced multiple projects: A new data center in Grand Rapids, Michigan and a $400 million colocation facility outside San Antonio. AWS, meanwhile, is planning a data center in Covington, Georgia, alongside a $100 million water and cooling facility.
Elsewhere, UK-based neocloud firm Nscale is expanding aggressively in the US, signing an $865 million, 40MW deal at WhiteFiber’s North Carolina campus with room to grow.
But not every project is moving forward. A proposed Tier III AI data center in Lewiston, Maine was voted down after strong public opposition, underscoring how community pushback is becoming a real constraint on development timelines.
Lastly, a Georgia-based film studio, facing declining film production, is exploring converting the space into a data center – proof that even small operators are eyeing AI infrastructure opportunities.
Cooling and Satellite Expansion
Finally, innovation continues quietly in the background. A UK university spin-out claims its new indirect evaporative cooling system can reduce cooling energy consumption by up to 90% – an eye-catching figure as operators hunt for efficiency gains. In emerging markets, Airtel Africa’s partnership with Starlink points to a future where direct-to-cell satellite connectivity plays a larger role in edge and regional infrastructure.