Good morning. It's Monday, March 2, and we're covering rising tensions over AI safety and defense, Meta’s chip strategy reset, OpenAI’s record-breaking funding push, and more.
New here? Keep up with the future of tech, sign up here. Have feedback? Send us a note: astro@forwardfuture.ai. If you liked this email, share it with a friend.
🗞️ THE WEEKEND RECAP
Top Stories You Might Have Missed
💰 OpenAI Lands $110B Mega-Round: OpenAI secured $110 billion from Amazon, NVIDIA, and SoftBank at a $730 billion valuation, launching massive AWS and infrastructure deals.
📈 Claude Tops App Store After Pentagon Clash: Anthropic’s Claude surged to No. 1 on Apple’s U.S. App Store, overtaking ChatGPT amid a public Pentagon dispute.
🛑 Citadel Pushes Back on AI Doom: Citadel argues 2026 AI investment—$650 billion, 2% of GDP—looks inflationary short term, but labor data shows limited displacement and rising tech hiring.
💥 Military Laser Downs Border Drone: The U.S. military used an anti-drone laser to disable a Customs and Border Protection drone near El Paso prompting temporary Federal Aviation Administration airspace restrictions.
🎉 Paramount Wins $111B Warner Deal: Paramount Skydance agreed to acquire Warner Bros. Discovery in a $110 billion deal, outbidding Netflix’s $82.7 billion offer after a five-month battle. The deal unites Warner, HBO, CNN and CBS under CEO David Ellison.
🤝 OpenAI Eyes Pentagon AI Deal: CEO Sam Altman told staff that OpenAI is working with the Defense Department to deploy models in classified settings, aiming to preserve safety guardrails amid Anthropic’s battlefield-use standoff.
📽️ VIDEO
The Government Just Blacklisted Anthropic...
Anthropic says it will not remove AI guardrails at the Pentagon’s request, escalating tensions over surveillance, autonomous weapons, and AI control.
📺️ FROM THE LIVE SHOW
If we get world models right, what does that mean for society?@runwayml CTO @agermanidis on what world models unlock:“There are two buckets: what they enable for humans, and what they enable for autonomous systems.”“For humans, it’s a new kind of software. Once you have a… pic.twitter.com/18uXi0eg4e
@ForwardFutureView post on X⚖️ GEOPOLITICS
Anthropic Resists Pentagon Demand to Drop AI Safeguards
Anthropic CEO Dario Amodei said the company would not remove two key safeguards from its AI contracts with the U.S. Department of Defense—referred to in the statement as the “Department of War”—despite threats of contract termination and other penalties. Anthropic says it will not support mass domestic surveillance or deploy its models in fully autonomous weapons systems, citing democratic values and reliability concerns.
The Defense Department has reportedly required contractors to agree to “any lawful use” and remove such restrictions, and has threatened to label Anthropic a “supply chain risk” or invoke the Defense Production Act. The standoff underscores growing tensions between AI firms and the U.S. military over how far AI should be used in national security operations. → Read the full article here.
🔲 SEMICONDUCTORS
Meta Scraps ‘Olympus’ AI Training Chip After Design Struggles
The Recap: The Information reported that Meta Platforms scrapped its most advanced in-house AI training chip, code-named Olympus, after running into design and manufacturing challenges, according to six people with direct knowledge. The move comes as Meta strikes major supply deals with Advanced Micro Devices and NVIDIA, underscoring how difficult it is for tech giants to rival NVIDIA’s dominance in AI hardware. Meta is now shifting focus to a less complex internal chip while continuing to rely on external suppliers to power its AI model training.
Highlights:
Meta scrapped Olympus—its most advanced AI training chip—after design and manufacturing challenges, following the earlier cancellation of a version of its Iris training chip.
On February 25, 2026, Meta and AMD announced a 6-gigawatt capacity agreement for chips, while Meta also committed earlier in February to deploy millions of NVIDIA chips under a multigenerational partnership.
Meta expects $115 billion to $135 billion in 2026 capital expenditures, with most spending earmarked for chips and servers to support AI infrastructure.
Olympus used a single instruction, multiple threads (SIMT) architecture similar to NVIDIA’s, but internal sources cited risks around software stability, power consumption, and large-scale manufacturing compared with NVIDIA’s mature CUDA ecosystem.
Forward Future Takeaways:
Meta’s retreat from Olympus highlights how AI chip design has become one of the most technically and operationally complex challenges in modern computing—where hardware, software ecosystems, and manufacturing scale must align. Even companies with Meta’s resources are finding that competing with NVIDIA means replicating not just silicon performance but a mature software stack and supply chain. The bigger question: Will Big Tech’s custom silicon ambitions ultimately reduce NVIDIA’s grip—or reinforce it as the default AI infrastructure provider? → Read the full article here. (Paywall)
💼 JOBS
AI Threatens India’s $300B Outsourcing Engine
Artificial intelligence is beginning to undercut India’s nearly $300 billion outsourcing industry, which employs more than six million people and accounts for over 7% of GDP. Start-ups like Hunar.AI are automating white-collar tasks such as recruitment, while giants including Tata Consultancy Services and Infosys have slowed hiring or reduced headcount.
At Hunar.AI, revenue reached $3 million from 70 clients, the company says, after it replaced dozens of recruiters with AI voice agents. The shift is squeezing new graduates and raising political and economic stakes for Prime Minister Narendra Modi, who has pledged to make India an AI leader even as the country remains reliant on U.S. chips and foundational models. → Read the full article here. (Paywall)
🔭 ASTRONOMY
Rubin Observatory Sends 800,000 Real-Time Sky Alerts
On February 25, 2026, the Vera C. Rubin Observatory issued its first live batch of 800,000 astronomical alerts, flagging newly detected asteroids, supernovae, variable stars, and active galactic nuclei. The notifications came from its Alert Production Pipeline, developed at the University of Washington, which processes roughly 10 terabytes of images per night and delivers alerts within two minutes of observation.
The system is designed to scale to as many as 7 million alerts nightly once fully operational. The milestone clears the way for the observatory’s 10-year Legacy Survey of Space and Time, expected to catalog more objects in its first year than all other optical observatories combined. → Read the full article here.
𝕏 TWEETS
Tech Workers Back Anthropic Guardrails
200+ Google and OpenAI staff have signed this petition to share Anthropic's red lines for the Pentagon's use of AIlet's find out if this is a race to the top or the bottomhttps://t.co/3qgmaLfM0i pic.twitter.com/gSHMxRUvCR
@jasminewsunView post on X🧰 TOOLBOX
Trending AI Tools
🗒️ FEEDBACK
Help Us Get Better
What did you think of today's newsletter? |
That’s a Wrap!
❤️ Love Forward Future? Share your referral link with friends and colleagues to unlock exclusive perks! 👉 Get your link here.
📢 Want to advertise with us? Reach 1M+ AI enthusiasts, tech leaders, and decision-makers. Just reply to this email.
Thanks for reading today’s newsletter. See you next time!
Matthew Berman & The Forward Future Team
🧑🚀 🧑🚀 🧑🚀 🧑🚀 🧑🚀
