New name (The Future Today). New look (bright & bold). Same obsession with making the future legible, so without further ado...

Good morning. It's Friday, March 6, and we're covering Anduril’s defense funding push, OpenAI’s $25B revenue surge, GPT-5.4’s launch with stronger AI agents, and more.

OpenAI introduced GPT-5.4 for ChatGPT, the API, and Codex, aimed at professional knowledge work. The model combines stronger reasoning, coding, and tool use, enabling it to operate computers and complete multi-step workflows autonomously. It supports up to a 1M-token context window and is more token-efficient than GPT-5.2. Benchmarks show gains in coding, research, and document-heavy tasks while reducing errors.

The U.S. Department of Defense has formally notified Anthropic that it considers the company and its products a “supply-chain risk,” a rare designation typically used for foreign adversaries. The decision escalates a standoff between the Pentagon and the AI firm over limits on how the military can use its models. Anthropic has resisted government requests to remove safeguards that restrict uses such as domestic surveillance or fully autonomous weapons.

Cursor introduced Automations, a feature that lets developers create always-on AI agents triggered by events like GitHub pull requests, Slack messages, or scheduled tasks. The agents run in cloud sandboxes, review code, investigate incidents, and handle routine engineering work such as bug triage and test generation. Cursor says the agents can learn from previous runs through memory tools and integrate with platforms like Datadog.

Major tech companies agreed to cover the cost of new power generation for their data centers under a “Ratepayer Protection Pledge.” The commitment aims to ensure electricity costs tied to AI infrastructure aren’t passed on to residential consumers. Companies involved include Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI. The plan also suggests excess power from new generation could be sold back to the grid to lower prices.

FRIDAY FACTS

Stick around to find out!

Most AI tools can’t access your real workspace. SciSpace Agent can. Connect GitHub, Notion, Zotero, Mendeley, Drive, and more — and work directly on your files. And activate Skills, our built-in expert modes, to automate grant discovery, generate presentations, and create websites — all inside a single AI conversation.

Try Skills in Agent for free.

Claim limited time discount: Use FTFF40 for 40% off (Annual plans) and FTFF20 for 20% off (Monthly plans)

Watch Forward Future’s weekly live show on X or YouTube.

Defense startup Anduril told prospective investors it expects revenue to reach about $4.3 billion in 2026, nearly double the previous year, while operating losses widen to roughly $1.2 billion as it scales manufacturing of autonomous weapons systems. The Costa Mesa–based company is simultaneously raising $4 billion at a $60 billion valuation, led by existing investors Andreessen Horowitz and Thrive Capital. Investors are betting the company can evolve from counter-drone and border systems into a major supplier of autonomous fighter jets, submarines, and missiles, though forecasts show profitability may not arrive until 2030.

Anduril represents a new model of defense contractor: venture-backed, software-driven, and willing to absorb startup-style losses to build large hardware programs. If its autonomous aircraft and submarine programs secure major government contracts, the company could reshape how defense technology is developed—moving faster than traditional primes. → Read the full article here. (Paywall)

As of late February 2026, OpenAI reached more than $25 billion in annualized revenue, up from $21.4 billion at the end of 2025, according to people familiar with the figures. Rival Anthropic has rapidly closed the gap, with annualized revenue recently surpassing $19 billion, driven in part by strong demand for its coding-focused AI models. Both companies are growing faster than expected as businesses adopt AI agents for software development and other white-collar tasks.

The surge in revenue at OpenAI and Anthropic shows how quickly AI coding tools and agents are becoming core enterprise infrastructure, not just experimental products. At the same time, both companies are spending heavily on compute and cloud capacity, highlighting the capital-intensive nature of scaling frontier AI models. With revenue climbing into the tens of billions, the race between the two firms is increasingly shaping the future structure of the AI industry—and potentially the timing of major AI IPOs. → Read the full article here. (Paywall)

Cybertruck Gets Active Noise Canceling: Tesla update 2026.2.6.1 uses seat microphones and speakers to counter road noise on rough surfaces, creating quieter cabins; feature limited to Premium and Cyberbeast trims.

Meta Smart Glasses Face Privacy Lawsuit: U.S. plaintiffs allege Meta misled buyers about privacy after reports contractors reviewed sensitive footage from AI glasses; regulators are probing as over 7 million units circulate.

Anthropic Reopens Pentagon AI Talks: After the Pentagon threatened to cancel Anthropic's $200M Defense Department deal, CEO Dario Amodei is reportedly renegotiating access terms

Oracle Plans Major Job Cuts: Oracle may lay off thousands as soaring AI data-center spending strains finances; the cloud push—tied to OpenAI, xAI, and Meta demand—could require $45–50B in funding.

Netflix Buys Ben Affleck’s AI Startup: Netflix acquired InterPositive, an AI tools firm for film production founded by Ben Affleck in 2022; its 16-person team joins Netflix, with Affleck becoming a senior adviser.

FRIDAY FACTS

Every prompt, whether it's debugging code or asking a philosophical question about raccoons, fires a full neural-network inference across massive GPU clusters. This requires thousands to millions of mathematical operations. A brand new answer generated every time.

This isn't a search engine returning a cached result. It's billions of live reasoning sessions happening weekly — on technology most people hadn't heard of just a few years ago.

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team