Meta’s AI bill changed its balance sheet.

The Frontier

Science & tech breakthroughs in AI models, hardware, and computing that define what's possible

Meta’s AI spending strains its cash flow. Meta reported increased artificial-intelligence spending in its earnings results as its cash flow turned negative. The company also said it would issue a new class of stock to raise additional capital, tying its financing plans directly to the cost of its AI buildout.

Meta’s negative cash flow means its AI buildout is no longer fully supported by operating cash: the announced new class of stock shifts part of the funding burden to equity. The test is whether Meta reports positive cash flow in its next earnings results.

New York Times

The U.S. Army makes a $700 million AI bet. The Army has entered a $700 million agreement with Striveworks to expand an AI platform designed to help soldiers process battlefield intelligence and make decisions faster. The contract is described as a tenfold increase from the initial enterprise agreement for the company’s Chariot program, which develops AI agents and models for operational support.

Striveworks’ $700M, 10x Army expansion signals Chariot has moved beyond an initial enterprise deal into operational support. The test is whether soldiers use it to shorten intelligence-processing and decision cycles.

Businessinsider

OpenAI models are coming to Google Cloud. OpenAI and Google Cloud announced a partnership to make OpenAI frontier models and Codex agentic workflows available through Google Cloud’s Model Garden. Enterprise customers will be able to access the offerings within Google Cloud infrastructure and governance controls, including for regulated-industry deployments.

OpenAI is using Google Cloud’s Model Garden to sell its frontier models and Codex workflows inside Google Cloud’s governance controls, targeting regulated enterprises rather than a simple infrastructure arrangement. The test is whether those customers adopt through Model Garden.

Openai

Model Developments

Latest AI model releases, capabilities, and technical breakthroughs shaping the competitive landscape

DeepSeek reportedly raises $15 billion days after U.S. restrictions. Bloomberg reported that Chinese AI lab DeepSeek raised about $15 billion, or roughly 100 billion yuan, days after the U.S. ordered federal civilian agencies to remove its technology over security concerns. The reported round, potentially involving CATL and Tencent, follows DeepSeek’s more than $7 billion June financing and is intended to support AI infrastructure and open-model development.

DeepSeek’s reported $15 billion raise suggests U.S. agency restrictions have not constrained its access to major Chinese capital. The test is whether it announces AI infrastructure spending or releases an open model by June.

TechCrunch

Claude Duet is Anthropic’s bet on agents. Anthropic launched Claude Duet, an agent-oriented application for professionals and small businesses. The product can generate daily briefings, build dashboards, suggest actions from user data, and execute tasks in a background virtual machine while seeking approval when needed.

Anthropic is betting Claude Duet’s approval-gated background VM—not briefings or dashboards—drives adoption. The test is whether professionals and small businesses repeatedly approve delegated tasks from their data.

TechCrunch

Capital & Control

Tracking the flow of capital and influence that shapes the tech landscape

Meta is ordered to pay $25 billion over Cambridge Analytica privacy violations. A New Mexico lawsuit over Cambridge Analytica-era privacy violations resulted in Meta being ordered to pay $25 billion. The penalty is among the largest imposed on a technology company and centers on allegations that Meta misled users about the privacy of their data.

Meta’s $25B order makes privacy enforcement financially material: it is 5× its 2019 FTC fine, though under 3% of its $900B cap. The test is whether the ruling resets privacy-penalty benchmarks.

New York Times

Infrastructure & Power

The physical backbone of AI: compute, data centers, supply chains, and the geopolitics of energy

Crusoe and Araico plan an AI cloud deployment at Paraguay’s Itaipú Dam. Crusoe and Paraguayan developer Araico announced plans to deploy an AI cloud facility beside the Itaipú Hydroelectric Dam substation, with initial operations targeted for mid-2027. The companies said the project has a path to 106 MW of IT capacity and would be Crusoe’s first dedicated AI cloud deployment in Latin America.

Itaipú gives Crusoe 106 MW. That is about 2 % of Paraguay’s entire load, enough for 70 000 homes. Every megawatt pulled for AI cuts lucrative exports to Brazil, so watch the 2027 Itaipú treaty talks get tougher on price.

Unite.ai

Digital Defense

Threats and defenses shaping the global security landscape

Patch the Planet coordinates a response to critical vulnerabilities. OpenAI said a researcher using GPT-5.6 Sol identified a critical PHP vulnerability during its Patch the Planet initiative, while Wiz independently reported both disclosed flaws. PHP maintainers issued fixes after a coordinated response involving OpenAI, Wiz, Cloudflare, and other security providers, with protections prepared before disclosure.

OpenAI’s Patch the Planet case suggests AI’s value is in speeding coordinated defense, not just finding flaws: Wiz independently confirmed both PHP issues, while Cloudflare and others prepared protections. The test is whether protections are ready before the next PHP disclosure.

Openai

The Performance Edge

Peak performance science for sustained execution in high-stakes environments

Government commits $41 million to dementia-prevention trials. The Advanced Research Projects Agency for Health is committing $41 million to clinical trials assessing supplements and approved medicines intended to prevent dementia in people identified as at risk. The funding comes as blood-based tests are expected to expand identification of people with Alzheimer’s-related risk markers before clinical symptoms become apparent.

$41M from ARPA-H bankrolls multiple trials testing already-approved drugs and supplements to keep blood-test-flagged adults from developing Alzheimer’s. Repurposed status means positive two- to three-year data could pivot guidelines by 2031. Watch for first readouts around 2029.

Science.org

Crypto & Digital Assets

Bitcoin, Ethereum, DeFi protocols, and digital asset market dynamics

Toronto’s mayoral race exposes prediction-market loopholes. More than $600,000 has been wagered on Toronto’s mayoral race through U.S.-based prediction-market platforms Polymarket and Kalshi, despite Canadian restrictions on political-event betting. Authorized Canadian dealers may offer markets tied to economic indicators, financial markets, and climate trends, but not election outcomes.

Polymarket and Kalshi expose a jurisdictional gap: Canadian rules bar authorized dealers from election markets, yet more than $600,000 has flowed to U.S.-based platforms. By the race’s end, watch whether that total rises—a measure of whether access remains open.

Betakit

Market Pulse

Daily market data: indices, sectors, top movers, and volatility

Markets retreat as technology leads the sell-off. The S&P 500 fell 1.0%, the Nasdaq 100 dropped 1.8%, the Dow Jones slipped 0.1%, and the Russell 2000 declined 0.2%. Consumer Staples led sectors with a 1.6% gain while Technology fell 2.3%; META dropped 6.1%, AMD lost 6.0%, and SMCI declined 4.9%.

META’s 6.1% drop and AMD’s 6.0% loss signal a concentrated Technology unwind, not a broad-market break: the Dow fell 0.1% while Consumer Staples gained 1.6%. The test next session is whether Technology again underperforms the S&P 500.

Yahoo Finance