Alphabet’s Google has announced a sweeping $25 billion investment plan in artificial intelligence infrastructure and data centers across the United States—focusing heavily on states connected to PJM Interconnection, the country’s largest electric grid.

The tech giant’s initiative was announced Tuesday at a high-profile summit in Pittsburgh, Pennsylvania, hosted by Senator Dave McCormick and attended by President Donald Trump, members of his cabinet, and top executives across the energy and tech industries.

The investment comes as Google races to meet ballooning electricity demand driven by the rapid adoption of AI technologies, with data centers becoming some of the largest consumers of power on the grid.

At the heart of Google’s commitment is a landmark $3 billion deal with Brookfield Asset Management to modernize two hydropower plants on Pennsylvania’s Susquehanna River. The deal, part of a broader 20-year agreement, will provide Google with 3,000 megawatts of renewable hydroelectric power—marking the largest corporate clean energy agreement of its kind globally, according to Brookfield.

The upgrade will transform aging infrastructure into a reliable power source capable of supporting next-generation AI data centers. It also represents a strategic pivot by Google toward achieving its 24/7 carbon-free energy target, ensuring that every operation across its facilities runs on clean power at all hours.

Pennsylvania Summit Showcases Public–Private Tech Alliance

The announcement was made during the Pennsylvania Energy & Innovation Summit at Carnegie Mellon University, where more than $90 billion in combined private sector investments were announced. Organized by Sen. McCormick, the summit hosted corporate leaders from Google, Amazon, Blackstone, CoreWeave, and ExxonMobil, alongside officials from the Trump administration.

President Trump, who attended the summit in person, hailed the announcements as proof that the United States is entering a “true golden age of AI and energy,” adding that “America will win the AI race not just with innovation, but with American power.” The administration is pushing for a revival of all energy sources—coal, gas, nuclear, solar, and hydro—to ensure grid reliability as AI technologies strain national energy systems.

“There’s no better place than Pennsylvania to lead the next revolution in energy technology and artificial intelligence,” said McCormick during his welcoming remarks at the Pennsylvania Energy and Innovation Summit. “Our commonwealth is poised to lead the next era of growth and opportunity.”

PJM Grid Under Pressure

Google’s decision to prioritize investments across the PJM Interconnection—which spans 13 states across the Mid-Atlantic, Midwest, and parts of the South—underscores the growing pressure on U.S. grid operators. PJM is home to the world’s largest data center market in Northern Virginia, and recent studies have warned that rising electricity demand from AI and industrial usage could exceed available generation in the coming years.

By targeting hydroelectric power and investing in modernized infrastructure, Google aims to contribute not only to grid reliability but also to long-term climate goals. In many ways, the deal also serves as a response to public concerns over AI’s environmental footprint.

AI Boom Reshaping U.S. Economic Geography

Google’s announcement adds to a wave of corporate spending transforming Pennsylvania into a national tech-energy hub. Blackstone announced $25 billion in energy infrastructure investments; CoreWeave committed to $6 billion for new data centers; and Amazon said it would spend $20 billion to scale its cloud and AI systems across the region.

The convergence of energy, data, and AI at this scale is reshaping not just the technology sector, but the geography of American economic power. Pennsylvania, once defined by steel and coal, now finds itself at the epicenter of a 21st-century transformation driven by digital infrastructure.

The broader strategy fits into the Trump administration’s geopolitical stance on AI. Officials at the summit emphasized that ensuring dominance in AI is not only about algorithms and chips but about energy access. The U.S. aims to curb reliance on foreign supply chains and solidify its edge over rivals like China by tying AI expansion to domestic energy investment.

“The AI revolution has potential and promise to transform our nation’s economic outlook … but also risks for national security,” McCormick said. “If the United States does not lead this revolution on our own terms, we will hand control of our infrastructure, our data, our leadership, and our way of life to the Chinese Communist Party.”

Anthropic has unveiled a specialized AI solution tailored for the financial services industry, signaling a major expansion of its enterprise ambitions and intensifying competition in the AI-for-finance space.

The new product, called the Financial Analysis Solution, is built on Anthropic’s Claude 4 model family and is aimed squarely at helping analysts and financial institutions conduct research, evaluate deals, and model investments with speed and precision.

Unveiled at an event in New York on Tuesday, the suite includes Claude Opus 4, Claude Code, and Claude for Enterprise. It’s available immediately on the AWS Marketplace, with Google Cloud Marketplace support expected soon. The move comes amid growing enterprise interest in tailored AI tools that can handle sector-specific complexity, and it positions Anthropic as a serious player in a space that’s been dominated by legacy systems and a few tech incumbents.

“This is a tailored version of Claude for Enterprise,” said Kate Jensen, Anthropic’s head of revenue, at the launch event. “It’s specifically built for financial analysts, and it’s equipped for the nuance, accuracy and reasoning that you need to handle the complexity of your work.”

Real-Time Data Access Sets Claude Apart

One of the Financial Analysis Solution’s key differentiators is direct integration with financial data sources, allowing Claude to access and analyze live financial information from platforms like PitchBook, S&P Global, Box, Databricks, Snowflake, and Daloopa in real time. These integrations are made possible via Claude’s Model Context Protocol (MCP), enabling dynamic access to proprietary and third-party data sets during chats.

Unlike typical LLMs, Claude doesn’t merely hallucinate data—it references live numbers and includes links to source materials, ensuring transparency and compliance for users in tightly regulated industries. According to Jensen, this allows analysts to go from “hypothesis to insight in a fraction of the time,” by bypassing the usual toggling between datasets, Excel models, and search tools.

Financial professionals using Claude can now ask complex questions like “What’s the latest earnings forecast for Tesla according to S&P Global?” or “Compare recent private equity deals in biotech over $100M from PitchBook” and receive instant, verifiable responses.

Enterprise-Grade Features and Use Cases

Beyond data access, the new solution includes enterprise security features such as single sign-on (SSO), audit logging, SCIM provisioning, user permissions, and end-to-end encryption. Claude also offers 200,000-token context windows, allowing it to process entire financial documents or pitch decks in one go.

The package comes with implementation support and a financial prompt library, with templates for tasks such as equity research, due diligence, valuation modeling, and risk assessment. Consulting firms like KPMG, Deloitte, Slalom, and PwC are working with Anthropic to help deploy the solution inside financial firms.

According to Anthropic, this level of support is intended to shorten time-to-value for companies and reduce the burden on in-house IT and analyst teams. The AI can handle research, summarize quarterly earnings, or even write draft investment memos in minutes—freeing up analysts for higher-order thinking.

“We are excited to partner with Anthropic to deliver S&P Global’s trusted data and insights wherever our customers need them,” said Bhavesh Dayalji, CEO of S&P Global Market Intelligence.

Positioning Against Competitors

Anthropic’s move into finance mirrors similar efforts by rivals like OpenAI, which has quietly worked with hedge funds and investment banks using GPT-4, and Google DeepMind, which has promoted agentic AI for technical users. But Anthropic’s vertical focus—starting with finance—and native integration with live data streams gives it a potential edge.

This is Anthropic’s first big disruption to the finance industry, Axios reported, citing the unique blend of real-time access, domain-specific tuning, and compliance-grade enterprise features.

Claude’s emergence as a high-trust AI assistant is timely, as financial institutions grapple with the balance between productivity gains and risks posed by AI-generated content. Anthropic says Claude’s outputs can be linked back to primary sources, minimizing the chances of misinformation—a key requirement for institutional clients.

Economic Potential and AWS Partnership

The launch also strengthens Anthropic’s deepening ties with Amazon’s cloud division. Amazon is Anthropic’s largest investor and strategic partner. Earlier this month, AWS revealed it would host Anthropic’s AI agents in its upcoming AWS Agent Marketplace, with Claude tools among the first listings. Morgan Stanley estimates that AWS could earn up to $3 billion in cloud revenue by 2026 from Anthropic-powered deployments.

This growing partnership could accelerate Anthropic’s market share, particularly among AWS’s existing base of financial services clients. The new Claude deployment appears designed to tap into that demand.

Broad Implications for Financial Workflows

The Claude Financial Analysis Solution highlights a shift in how finance teams engage with technology. While previous AI efforts focused on cognitive assistants or chat summarizers, Anthropic is introducing agentic AI capable of executing complex workflows, modeling decisions, and generating high-confidence analysis.

Though some financial institutions are likely to adopt the tool cautiously—especially given the risks around AI hallucinations—early feedback on Claude’s financial-specific version has been largely positive, including among hedge funds and buy-side firms seeking faster deal analysis.

With the Claude Financial Analysis Solution, Anthropic is signaling that it’s not content with being just another chatbot. And if it succeeds, Anthropic could become the first AI firm to fully institutionalize large language models into the heart of global financial decision-making.

When Aravind Srinivas took the stage at YCombinator’s AI Startup School, the Perplexity AI chief executive had a bracing message for the roomful of would be founders: assume the giants will clone your best ideas, and build as if they already have.

Srinivas knows the playbook firsthand. Perplexity launched in December 2022 as an “answer engine” whose chatbot could crawl the live web—a capability missing from ChatGPT and Google’s Bard at the time. Within months, Google added real time search to Bard (now Gemini), OpenAI rolled out browsing for ChatGPT, and Anthropic’s Claude followed suit in 2025.

“They raise tens of billions, they need fresh revenue, and they will copy anything that’s good,” Srinivas told the audience. “You’ve got to live with that fear.”

From Answer Engine to Browser Challenger

Rather than retreat, Perplexity has doubled down on speed and differentiation. On 9 July the company unveiled Comet, an AI centric browser that bakes Perplexity’s answer engine directly into the navigation bar, promising instant summaries, live web citations, and personalized context. Srinivas claims the integration barrier is high enough that “Big Tech cannot copy Comet” as quickly as it mimicked Perplexity’s earlier chat features.

Hours after Comet’s debut, Reuters reported that OpenAI is quietly developing its own Chrome style browser, signaling that a new front—what Perplexity’s communications chief Jesse Dwyer calls “Browser War III”—is about to erupt. Both OpenAI and Google, he warned, could lean on their incumbent platforms to “drown out” smaller rivals by bundling browsers with other must have services.

The Copycat Economy

Srinivas’s cautionary lesson taps into a broader reality of the AI boom. Deep pocketed firms under shareholder pressure to justify multibillion dollar capital expenditure lines now view nimble startups as live idea labs. Meta recently snapped up voice tech outfit PlayAI, while Google lured Windsurf’s CEO and R&D team for a reported $2.4 billion licensing pact.

In the same week, Cognition Labs acquired Windsurf’s remaining assets to fortify its own autonomous coding tool, Devin—a flurry of split deals that showed founders can be dismembered as readily as they’re acquired.

Perplexity, backed by Jeff Bezos and Peter Thiel’s Founders Fund, has itself fielded overtures; Srinivas told Business Insider he has no intention of selling. Instead, he is betting on pace: Perplexity mandates internal use of AI coding agents to cut prototyping from days to hours, and it ships upgrades to its engine weekly. The goal, he says, is to iterate faster than a giant can integrate.

Stakes for Users—and the Web

Dwyer argues that if dominance in AI browsers coalesces around one or two “everything companies,” consumers could face the same choice limiting bundling that defined earlier browser battles. “Browser wars should be won by users,” he wrote in a company blog, “not by monopolistic tactics that force a product onto the market.”

Perplexity’s strategy is to keep its engine open to the public web while layering premium features—such as larger context windows and image analysis—behind a subscription tier. Comet extends that posture: live citations next to every answer, side by side comparisons of sources, and a data privacy pledge not to track users across sites.

Whether that will be enough against incumbents with vast distribution remains to be seen. Google commands roughly two thirds of global browser share via Chrome, while Microsoft pushes Edge through Windows defaults. If OpenAI enters the fray backed by Microsoft, competition could hinge not just on innovation, but on regulatory scrutiny of bundling practices—a replay of antitrust debates from the late 1990s browser wars.

Lessons for Founders

For the students at YC’s event, Srinivas distilled his advice to two imperatives: work incredibly hard and ship relentlessly, assuming knock offs will appear. “Speed is your moat,” he said. Differentiation must come from product depth—like Comet’s integrated answer layer—rather than from any single feature that a trillion dollar rival can replicate overnight.

Yet he closed on a note of guarded optimism. Innovation cycles move faster than ever, he said, and a nimble startup can still outrun the giants by pivoting before the copy lands. In a market where the cost of experimentation is collapsing, the next breakout advantage is likely to be cultural—an ability to live with the fear of being copied and still keep building.