YouTube Insights in 5 Minutes Instead of 3 Hours

The Lelit Steve is a dedicated espresso grinder with 64 mm burrs. It is compact (23 cm deep, 14 cm wide, 39 cm high) and weighs 6.5 kg. Its design is distinctive and fits well with white Lelit machines.

🔧 Features & Operation

• Grind adjustment: Stepped (not stepless). Steps are smaller than the predecessor (approx. 5 s brew time change per click), but in testing a single click shifted brew time from 27 s to 18 s – too coarse for precise dialing.
• Display: New, with two programmable doses plus continuous mode. Easy programming via the plus button.
• Portafilter holder: Works with all common portafilters (ECM, Rocket, Arkel, Ascaso, La Marzocco, Lelit Mara), but the holder feels outdated and poorly designed.
• Bean hopper: Holds approx. 300–350 g, not removable – makes cleaning and bean changes very difficult. Four screws must be removed to access the burrs, and the plastic threads can easily strip.

⚙️ Grinding Performance & Precision

• Retention: Temporary 5.2 g, permanent 0.3 g → total 5.5 g. High; single-dosing is nearly impossible (2–2.5 g remain).
• Particle distribution: Main peak at approx. 240 µm (espresso setting), broad distribution with significant fines. Delivers full-bodied, sweet espresso, especially with medium to dark roasts. Light specialty coffees lose some complexity.
• Reproducibility: After returning to the same grind click, the main peak shifted to 267 µm and fines increased by 5 % – the grinder does not return exactly to the same point. Cause: play in the gear transmission and manufacturing tolerances.
• Dosing accuracy: Time-based, but output varies by up to 1 g (e.g., set to 18 g, delivered 18.5–19 g). A scale and manual adjustment are required.

📊 Other Measurements

• Grind speed: Approx. 30 g in about 20 seconds – fast.
• Noise: 82 dB, about 6–7 dB quieter than the predecessor. Pleasant grinding sound.
• Heat: Grounds remain cool enough; no negative taste effects.

🎯 Conclusion & Recommendation

The Lelit Steve can produce solid espresso in everyday use, but has clear weaknesses:

• Pros: Fluffy grounds, good particle distribution for medium/dark roasts, quieter grinding.
• Cons: Coarse grind steps, poor reproducibility, high retention, non-removable hopper, inaccurate dosing.

👉 For €400–500 there are better alternatives (especially single-dosing grinders with higher precision and lower retention). The Steve is best suited for Lelit enthusiasts who are comfortable with manual corrections and primarily use darker roasts. A clear „past wasn‘t all better“ – Lelit‘s grinder development has fallen short of expectations.

„When buying Lelit products, look for a female name – the espresso machines are more convincing than the male-named grinders (Fred, Steve, William).“

Summary follows...

The new washed espresso "La Konserva" comes from the highlands of Western Guatemala (Antigua region) and is produced by several smallholder families. Its name pays homage to a traditional Guatemalan dessert, perfectly reflecting the coffee's sweet, sticky fruitiness. As part of the Orange Line, it stands for clarity, vibrant acidity, and perfect balance.

• Complex & zesty: Reminiscent of a citrus-mandarin lemonade with a sparkling acidity.
• Tasting Notes: Juicy mandarin, grapefruit, tonic water, mango, and a hint of honey.
• Transformative: The espresso reveals impressive complexity as it cools – almost like a chameleon.

The beans are surprisingly homogeneous, making grinding easy to dial in. However, the sweet spot is very precise:

• Dose: 20 g
• Yield: 46 g (1:2.3 ratio)
• Extraction Time (Standard Profile):
  • Vibe Pump: 26–30 seconds
  • Rotary Pump: 31–32 seconds
• Why not a Pressure/Flow Profile? The team tested a pressure profile (pre-infusion at 3 bar) and a flow profile. While the flow profile made the coffee too round and "boring", the pressure profile failed to integrate the bright complexity. The standard profile is the clear recommendation.

This coffee also shines as a filter brew, revealing an intense, surprising sweetness. Don't forget to enjoy it as it cools down!

• Every crypto wallet is secured by a mathematical problem – no password, no bank. Security relies on the difficulty of deriving the private key from the public address.
• This system has worked for decades, but the rapid progress of AI in mathematics is raising new concerns.

• AI systems like OpenAI and DeepMind achieved gold-medal level at the International Mathematical Olympiad in 2025.
• DeepMind's Alpha Evolve finds improved algorithms, and AI is even helping with open research problems.
• An AI model already improved an attack on a new signature design – not yet on Bitcoin or Ethereum, but a clear warning sign.

• Ethereum researcher Justin Drake calls for a “bunker mode”: gradually move funds to new addresses and switch to more robust locks.
• Vitalik Buterin agrees but warns against panic.
• Two camps: The alarm camp sees AI as a superintelligence, the skeptic camp points to 30 years of unchanged attack methods and considers the warning exaggerated.

• At risk are addresses that have already made transactions – here the “lock” is publicly visible.
• About one-third of all bitcoins sit on such addresses, including Satoshi's coins.
• Mining and hashing are hardly affected, as they offer no mathematical structure for shortcuts.

• Ethereum plans faster upgrades, including hash-based signatures and a dedicated team for the transition by the end of the decade.
• Bitcoin is deliberately slow: A proposal suggests banning old address types after ~3 years and shutting them down after ~5 years – that would freeze inactive coins.
• There are proposals for optional upgrades (Adam Back) and zero-knowledge proofs to identify holders without moving coins.

• Stay informed through official channels, not DMs.
• Use current wallets, keep your seed phrase offline, and avoid address reuse.
• Don't panic: Both camps advise against rushed sales.

AI hasn't cracked the locks yet, but it is forcing the industry to act sooner than planned – because if the threat materializes, the window to respond might already be closed.

The channel analyzes the crash of CoreWeave (CRWV) and the subsequent 'divine intervention' by investment banks (Goldman, JPM) and Nvidia CEO Jensen Huang. The central thesis argues that banks and Nvidia protect each other to sustain the profitable AI narrative.

• CoreWeave stock dropped 21% last week.
• The yield on CoreWeave's 2030 bond spiked to 14-15% – a level that fundamentally alters the investment thesis.

• Goldman Sachs upgraded the stock to "Buy" despite fears, claiming mixed debt costs only rose 100-200 basis points.
• Deutsche Bank and JP Morgan echoed the positive sentiment ("The business is solid").
• The Conflict of Interest: These banks were the lead underwriters for CoreWeave's IPO. The host points out the irony of research departments recommending to buy what the investment banking arms just issued (comparing it to SPACs and the Dot-com era).

• Jensen Huang (Nvidia CEO) publicly reassures the market: "No reason for concern."
• The Catch: Nvidia holds 10-11% of CoreWeave (one of the largest shareholders) and invested $2 billion, which CoreWeave exclusively uses to purchase Nvidia chips – a closed loop they must defend.

• The entire AI boom rests on a narrative. Without rising stock prices and optimism (fueled by bond sales and favorable analyst reports), the house of cards crumbles.
• "Look at the incentive, and you understand the narrative." (Charlie Munger): Wall Street collects its hefty fees at issuance. Pension funds and retail investors are left holding the risk.
• Warning to Investors: Projections assume 7-10 years of linear growth, but AI will drastically lower prices. What looks like a value investment today could become worthless once the narrative flips.

The host strongly advises against such narrative-driven investments and instead recommends tangible assets. The "divine intervention" is, in his view, nothing more than a desperate support operation for an overheated house of cards.

This video analyzes SpaceX's explosive growth potential, especially in compute. Elon Musk has never been more bullish, tweeting that SpaceX could be worth over $100 trillion in less than 10 years – a 50x from today's valuation.

🔥 Why SpaceX Has an Unfair Advantage

• Fastest data center build: 20 weeks to completion, 100,000 GPUs operational in 90 days. Competitors take 2–4 years.
• Coherent compute: SpaceX is the only company able to cluster more than 30,000 GPUs – Elon runs 220,000 GPUs repeatedly (e.g., Colossus: 150k H100s, 50k H200s, 30k GB200s).
• Power on site: 400 MW wind turbines in Brownsville, Texas. Up to 456 Starship launches/year under FAA review.

📈 Starlink & Starship as Accelerators

• Starlink revenue could grow 925% in 5 years (10x). Ron Baron values Starlink alone at $14–15 trillion.
• Starship lifts 100–150 ton payloads; competitors (China, Europe, Russia) are far behind.

🤝 SpaceX–Tesla Merger?

• Rumors of an all-stock acquisition of Tesla by SpaceX are heating up. Prediction markets show 60% chance by spring 2027.
• Tesla already holds billions in xAI shares converted to SpaceX stock.

💰 Price Targets – Compute Only (no Launch, Starlink, etc.)

• Base case (66.5 GW online by 2031): SpaceX share price $825 (5.65x). A 655-share „retiree bag“ = $540,000.
• Bull case (double the GW): Share price ~$4,000 (21.7x) → 655 shares = $2.4 million.
• Musk says even 66 GW is too low – actual numbers could be much higher.

⚡ Scarcity Drives Prices Higher

• Compute demand is parabolic, but supply is constrained. Power shortages (13–15 GW gap in North America) mean only SpaceX can deliver fast – and charge a premium.

🗳️ Community Poll

• 74 % believe 100 SpaceX shares will outperform 0.18 Bitcoin (each about $15,000) over 4 years.

Takeaway: SpaceX’s compute business alone could make investors wealthy – before factoring in Starlink, Starship, or a Tesla merger.

Analysis: What the Federal Reserve Will Do Next

This video analyzes the likely course of US monetary policy in the coming months, with a special focus on parallels to 2018.

• The federal funds rate is currently at 4% – another hike before year-end is expected.
• For the upcoming FOMC meeting (in about 9 days, October), the market implies an 80% probability of a pause – not a hike.
• Recent inflation data (CPI, PPI) were elevated, but the market still does not expect a move in October.

• In 2018, the Fed raised rates in September – right in the middle of a midterm election year.
• It paused in November (the meeting took place right after the midterms) and then hiked again in December.
• The projections at the time (SEP) called for further hikes in 2019 – but they never happened. Instead, the Fed started cutting rates in 2019.

• The Fed closely follows the 2-year Treasury yield – currently at 4.76%, above the federal funds rate.
• In late 2018, the 2-year yield was around 2.9%, matching market expectations at the time. When it later fell, further hikes became unnecessary.
• Potential scenario: The 2-year yield rises into the midterms, falls afterward, and the Fed hikes in December. Then the funds rate and the 2-year yield would be nearly identical – making further hikes unnecessary.

• Long-term yields in midterm election years often peak in October, rise again shortly before the elections, and then fall after the midterms.
• This dynamic could repeat in 2026.

• No rate hike in October (only about 16% probability).
• Yields rise into the midterms and fall afterward.
• Likely one more rate hike in December.
• After that, the Fed could keep rates stable through 2027 – similar to 2018, when planned hikes never materialized.