Blackstone is deepening its bet on the artificial intelligence infrastructure boom with a new $5 billion partnership with Google, marking another episode of how the race for AI dominance is rapidly shifting beyond software models into ownership of the physical computing backbone powering the industry.

The new U.S.-based infrastructure company, announced Monday, will deploy Google’s proprietary tensor processing units, or TPUs, to build large-scale AI compute capacity, with the first 500 megawatts expected online by 2027 and plans for substantial future expansion.

The venture places Blackstone at the center of one of the fastest-growing segments of the global technology industry: AI data infrastructure.

As demand for generative AI systems accelerates, technology companies, cloud providers, and private equity firms are now scrambling to secure access to power, chips, networking capacity, and data centers capable of handling increasingly complex workloads.

“This new company has enormous potential as it helps to meet the unprecedented demand for compute,” Blackstone President and Chief Operating Officer Jon Gray said in a statement.

The deal also represents a major escalation in Google’s long-running effort to reduce dependence on Nvidia, whose graphics processing units have become the dominant hardware powering the global AI boom. While Google still uses Nvidia chips extensively across its cloud infrastructure, the company has spent years building its own semiconductor ecosystem around TPUs, chips specifically designed for artificial intelligence computations.

Unlike Nvidia’s GPUs, which are general-purpose accelerators originally developed for gaming and graphics rendering, Google markets TPUs as specialized processors optimized for machine learning and agentic AI workloads.

The infrastructure partnership, therefore, is seen as a reflection of more than a financing arrangement, as it signals Google’s attempt to establish its hardware architecture as a viable alternative to Nvidia’s near-monopoly in AI computing.

The rivalry has intensified as hyperscalers increasingly seek vertical integration to control both the software and hardware layers of AI systems. Amazon Web Services, for example, has developed its own Trainium and Inferentia chips, while Microsoft has expanded investment in proprietary AI processors through its Maia programme. Google was among the earliest major technology companies to pursue in-house AI chips, developing its first TPU as far back as 2015, years before the generative AI frenzy transformed semiconductor markets.

Now, the economics of AI are making that strategy increasingly important. The explosive growth of large language models has triggered an unprecedented surge in demand for computing infrastructure, driving shortages in high-performance chips, soaring data-center construction, and escalating electricity consumption globally.

Industry analysts increasingly describe compute capacity as the new bottleneck in AI. That has created massive opportunities for infrastructure-focused investors such as Blackstone, which has aggressively expanded across digital infrastructure, energy, and data centers in recent years.

The asset manager, which oversees more than $1.3 trillion in assets, is already the world’s largest private owner of data centers. Earlier this month, Blackstone launched another AI-focused infrastructure venture with Anthropic, highlighting how private capital is rapidly moving deeper into AI’s foundational layers rather than limiting exposure to software applications alone.

The latest partnership also underpins how Wall Street and Silicon Valley are teaming up in financing AI expansion. Building hyperscale AI infrastructure now requires enormous capital commitments that increasingly resemble utility or energy projects rather than traditional software businesses.

Training and operating advanced AI systems require massive amounts of electricity, cooling systems, networking hardware, and specialized semiconductor supply chains. The planned 500 megawatts of compute capacity in the Blackstone-Google venture would rank among the larger AI infrastructure deployments globally and reflects expectations that AI demand will continue growing sharply over the next decade.

The Wall Street Journal reported that the venture has already identified likely data-center sites, some of which are reportedly under construction. The company will be led by Benjamin Treynor Sloss, a longtime Google executive with deep operational experience in infrastructure scaling.

Neither Blackstone nor Google disclosed the ownership structure of the venture, although reports indicate Blackstone will hold a majority stake.

After years of dominating the AI hardware market, Nvidia is facing growing competitive pressure from customers determined to reduce dependence on a single supplier. This makes the partnership pivotal.

Nvidia’s extraordinary rise following the launch of OpenAI’s ChatGPT in 2022 transformed the chipmaker into the world’s most valuable company last year. But the company’s dominance has also created strategic concerns among hyperscalers wary of supply constraints, pricing power, and overreliance on one hardware ecosystem.

Google’s TPU strategy is partly aimed at solving those concerns internally while positioning its cloud business as an alternative AI infrastructure provider. The company already uses TPUs to run its Gemini AI models, while clients including Anthropic and Citadel Securities also use the technology.

The broader significance of the Blackstone-Google venture lies in how it highlights the next phase of the AI race. Competition is no longer centered solely on chatbots or consumer applications. Increasingly, the battle is about who controls the underlying infrastructure powering artificial intelligence itself. That includes semiconductors, energy access, cloud architecture, networking systems, and data-center capacity. In that environment, firms capable of controlling multiple layers of the AI stack may hold the strongest long-term advantage.

The market appeared to respond positively to the announcement, with shares of both Alphabet and Blackstone rising in premarket trading.

In a candid revelation about the rapidly evolving AI-driven hiring landscape, Nvidia’s Chief Software Architect, Jonathan Ross, has advised job seekers to strategically game the system by tailoring their resumes to the specific large language models (LLMs) that recruiters are using.

Speaking at the Sohn Investment Conference 2026, Ross, a veteran AI hardware architect who previously helped design Google’s Tensor Processing Unit (TPU), highlighted a growing phenomenon: AI screening tools exhibit strong self-preferencing, favoring resumes generated by the same underlying model.

“Someone did a study and showed that resumes generated from one LLM are preferred by that same LLM over the resumes from the other,” Ross told John Yetimoglu, CIO of Infinitum.

He continued: “The recruiters are now using LLM to determine who to interview, but you got to figure out which LLM the recruiter’s using.”

Ross recommended a pragmatic, multi-model approach.

“So, you should build one resume with Claude or Opus 4.7 and one with ChatGPT, and you’ll have the highest probability of being selected, basically,” he said.

The Science of AI Self-Preferencing

According to Business Insider, Ross was referencing the 2025 academic paper “AI Self-preferencing in Algorithmic Hiring”, published in the Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society. Researchers Jiannan Xu, Gujie Li, and Jane Yi Jiang analyzed over 2,200 resumes across 24 occupations and found that candidates using the same AI model as the evaluator were 23% to 60% more likely to be shortlisted than those submitting human-written resumes with equivalent qualifications.

This bias appears to stem from stylistic alignment: AI models favor language patterns, phrasing, and structures that mirror their own training data and generation style.

The advice arrives as AI adoption in recruitment has reached critical mass. A 2025 Resume.org survey of nearly 1,400 U.S. workers found that 57% of companies are already using AI in hiring processes. Among those: 79% use AI to screen resumes. 74% allow AI systems to automatically reject candidates without human intervention.

What started as a tool to reduce recruiter workload has evolved into a powerful gatekeeper. Many companies now run initial resume screens entirely through AI, meaning millions of candidates are being evaluated by algorithms before any human eyes see their applications.

A Perfect Storm of Risks and Challenges

Feross Aboukhadijeh, CEO of code security startup Socket, and Isaac Evans, CEO of Semgrep, have both warned about the broader implications. The combination of AI-generated code flooding systems and AI-powered hiring tools creates what Aboukhadijeh called a “perfect storm” — where the volume of code (and therefore potential vulnerabilities) explodes while human oversight shrinks.

This shift raises serious concerns about fairness, diversity, and innovation in talent acquisition. AI systems may inadvertently penalize unconventional career paths, non-traditional education, or candidates who don’t optimize for machine readability. There is also growing evidence of false negatives, where strong candidates are rejected prematurely by overly rigid algorithms.

Business Insider recently highlighted the case of an IT professional rejected just six minutes after applying, strongly suspecting an AI system had instantly screened him out.

For ambitious professionals, Ross’s comments reveal a new competitive reality: understanding and adapting to recruitment technology is becoming as important as core qualifications. Job seekers who treat resume creation as a strategic exercise, testing multiple LLMs, refining prompts, and maintaining different versions, may gain a significant edge in an increasingly automated process.

It is believed that the findings highlight the urgent need for better governance for employers and HR teams. Experts have warned that over-reliance on single-model AI screening risks creating echo chambers that reduce diversity of thought and background. Against that backdrop, forward-thinking companies are beginning to implement hybrid approaches: using AI for efficiency while ensuring meaningful human review for final shortlists, along with regular audits for bias.

Manoj Nair of security startup Snyk described the current environment for Chief Information Security Officers and HR leaders alike as living in “AI fog” — a period of uncertainty where powerful new tools create both opportunities and hidden dangers.

As AI capabilities continue to advance, the hiring process is likely to become even more sophisticated — and potentially more opaque. We may soon see the emergence of “AI detection” tools designed to identify machine-generated resumes, as well as countermeasures from candidates. Regulatory scrutiny around algorithmic fairness in hiring is also expected to intensify.