Oprisa's Journal

In today's weekly journal

IN A WORLD THAT’S CHANGING SO QUICKLY THE BIGGEST RISK YOU CAN TAKE IS NOT TAKING ANY RISK. Peter Thiel

Table of Contents

 

GEOPOLITICS

Trump confirms strikes on Iran’s Nuclear sites

President Trump addressed the nation at 2200ET and said that with “massive precision strikes”, Iran’s nuclear facilities have “been completely and totally obliterated” in a “spectacular military success.”

He warned Iran(“the bully of the Middle East”) that if they do not come to peace then there are many other targets:

  • “There will be either peace; or tragedy for Iran worse than the likes of which we have seen over the last eight days.”

  • “If peace does not come quickly,” we will go after the rest of the nation with “speed and skill.”

He ended the speech by saying “God bless the Middle East, God bless Israel, and God bless the United States.”

Science & Technology

Engineers at Italian Institute of Technology just debuted iRonCub3 — the world’s first humanoid robot capable of hovering and balancing midair using jet engines and AI-powered controls system.

Core details:

  • iRonCub3 is equipped with four jet engines on its back and legs, a titanium spine, and heat shields to withstand temperatures up to 800°C.

  • In its first flight, the robot successfully hovered 50cm off the ground, maintaining balance and stability using real-time AI posture control.

  • Advanced algorithms dynamically manage thrust and body orientation to prevent midair spin or instability.

  • Though still a proof-of-concept, iRonCub3 marks a breakthrough in creating humanoids that can operate both on the ground and in the air.

Why it matters: After two years of development, this flying humanoid opens the door to next-gen hybrid robots with the potential to revolutionize search-and-rescue, disaster relief, and even off-world exploration — navigating environments that grounded robots or drones alone simply can’t handle.

The WSJ reports tensions between OpenAI and Microsoft have hit a “boiling point,” with OpenAI now weighing antitrust action as disputes over compute access, IP ownership, and corporate control escalate.

Core details:

  • The latest argument comes over OpenAI’s $3B acquisition of Windsurf, with the company wanting to withhold the IP due to Microsoft’s rival GitHub Copilot.

  • Executives at OpenAI are now said to be contemplating the “nuclear option”: filing an antitrust complaint and triggering a federal review of their partnership.

  • Microsoft had also resisted elements of OpenAI’s recent restructuring as a Public Benefit Corporation, with negotiations between the two sides reportedly resuming last month.

  • In a clear bid to diversify, OpenAI inked a cloud deal with Google last week — a notable step away from Microsoft’s Azure.

Why it matters: What was once AI’s most iconic alliance is unraveling fast. A full-scale legal move from OpenAI against its biggest investor would mark a dramatic turn — but also signal the company’s urgency to gain independence and control its future as AGI nears. What began as collaboration is quickly morphing into rivalry.

Meta just unveiled V-JEPA 2 — a 1.2B-parameter “world model” built to give robots and autonomous system real-world intuition and physical reasoning.

Core details:

  • Unlike language-based models, V-JEPA 2 learns by simulating the physical world, using over a million hours of unlabeled video and images.

  • These internal simulation let it predict motion, interactions, and outcomes with human-like logic — in real time.

  • It’s then fine-tuned on 62 hours of robot interaction data, enabling action-based predictions and decision-making.

  • The result: robots that can plan and execute tasks with unfamiliar object in new environments — no retraining or demos needed.

Why it matters: V-JEPA 2 brings robots a step closer to generalized intelligence, capable of adapting in dynamic settings. With Meta, DeepMind, Odyssey, and World Labs all racing to master world modeling, this may be the key to unlocking robots that actually understand the world they move through.

Robotics startup Skild AI just secured $35M from Samsung and Nvidia — $10M and $25M respectively — as part of a Series B round that now values the company at $4.5B.

Core details:

  • The round is anchored by a $100M commitment from SoftBank, with major South Korean players like LG, Hanwha, and Mirae Asset also participating.

  • Skild’s core product, “Skild Brain,” is a foundational model built to power robotic manipulation, navigation, and locomotion.

  • The company’s big bet: creating adaptable robotic intelligence for everything from factory work to home assistance.

  • Skild’s platform takes a horizontal approach — letting robots handle diverse tasks across industries with minimal retraining.

Why it matters: With giants like Tesla, Amazon, Figure, and Google ramping up in robotics, Skild’s push for scalable, general-purpose intelligence could be a game-changer — laying the foundation for robots that seamlessly move between sectors like healthcare, logistic, and construction.

OpenAI just launched “OpenAI for Government,” a new initiative consolidating its federal partnerships — and secured a $200M DoD contract aimed at applying frontier AI to national security across “warfighting and enterprise” areas.

Core details:

  • The one-year Pentagon deal marks OpenAI’s official entry as a U.S. military contractor, with work focused in the D.C. region.

  • ChatGPT Enterprise will assist service members with tasks like navigation benefits, while custom models will address areas like proactive cyber defense.

  • Existing collaborations with NASA, NIH, the Air Force Research Lab, and the Treasury are now unified under the “OpenAI for Government” umbrella.

  • The DoD’s contract listed the role as developing “prototype frontier AI” for “warfighting and enterprise,” though OAI said it would follow usage policies.

Why it matters: As AI leaders deepen ties with governments and defence agencies, the role of advanced models is shaping modern warfare — from drones to cyber — is quickly expanding. OpenAI’s shift into defense signal a future where military advantage may hinge on who controls the smartest machines.

>Octopuses use their arms to sample the microbial population on different surfaces, distinguishing between food, eggs and regular object (More)

>Researcher develop technique to target specific cells affected by ALS, a key step in delivering effective treatments; neurodegenerative disease affects around 30,000people in the US, has no current cure (More)

>Astronomers discover roughly three-quarters of the universe’s unaccounted-for ordinary matter — which makes up 5% of the universe’s matter and energy — is in thin gases located between galaxies(More)

Market view

Friday Charts:

Markets are forward-looking — that’s their job — even if we humans tend to obsess over the present. Legendary investor Bill Miller once said, “100% of the information you have about a business reflect the past, and 100% of its value depend on the future.” That’s what makes investing hard.

This week offered a perfect case study: war broke out in the Middle East, and yet markets barely flinched. It’s natural to think “the market doesn’t get it,” but the truth is more likely that it does — and just doesn’t think it matters(yet).

Our instincts scream for action in crisis, a holdover from ancient survival wiring. In investing, that turns into impulsive trades. But usually, staying still is the better move.

As Deutsche Bank’s Henry Allen reminded clients this week, “Geopolitics doesn’t normally matter much for long-run market performance.” LPL Financial data shows most market dips from geopolitical shocks recover within weeks or months.

Global prosperity continues its long climb, despite wars and poor policymaking. So when markets stay calm, maybe it’s not ignorance — but perspective.

Let’s look at the charts. Sometimes, nothing happening is the signal.

It’s been a chaotic month globally — but you wouldn’t know it by looking at the markets: Bitcoin, Gold, stocks and bonds have all barely moved, trading sideways despite the headlines.

Expectations re-coupling with reality?

The “soft” economic data — like sentiment survey — has bounced back to align with hard data, which measures actual activity. While the hard data is trending lower, the immediate impact of tariffs and policy uncertainty hasn’t been as severe as many feared.

The news aside, things are unusually good right now:

Ed Yardeni describes the current economic environment as a rare moment of “nirvana”: “The history of economic nirvanas with both inflation subdued at around 2.0% year over year and the economy at full employment (i.e., with the jobless rate at 4.0%) shows eight such transcendental experiences since the 1950s, including the current one.”

The class off 2025 is not feeling it, however:

2025 may not be the worst year to graduate college, but it’s likely the toughest compared to how strong the overall economy appears. The reason? AI could be replacing many entry-level roles, or-as Paul Krugman suggests-political uncertainty might be making employers hesitant to hire. Or maybe it’s something else entirely. Either way, if you’re still in school, it might be smart to stay a bit longer.

Off to a slow start:

Recent grads are currently three times more likely to be unemployed than those aged 35-44 — and the long-term implications are grim. As Paul Krugman points out, early career unemployment can have lasting effects. An NBER study found that entering the job market during a downturn leads to persistently lower earnings, higher disability rates, fewer marriages, and even fewer children. In short: the scars can last a lifetime.

Ethereum has cycled through various narratives — from “world computer” to “ultra-sound money” to “crypto app store” — but none have truly stuck with mainstream investors.

Now, the team at Etherealize is giving it another spin: positioning ETH as “a productive reserve asset — digital oil powering the digital economy.”

They argue this framing supports a short-term price target of $8,000, a long-term goal of $80,000, and a more speculative “thought-experiment” valuation of $740,000 — which would peg Ethereum’s total value at $89 trillion.

Lord, grant me the confidence of someone who believes a $300B asset is undervalued by 30,000%.