BTC is treated as structurally supported above the 200-week moving average, with long-term holders still absorbing supply and ETF flows mostly green, while a power-law floor model maps a path from a ~$60k 2026 floor toward ~$580k by 2034/2035. The bigger thesis is that AI value is concentrating in the cheapest, fastest, most integrated infrastructure: SOL dominates tokenized equities volume, Chinese models are said to be undercutting OpenAI and Anthropic on price, and X is said to be far ahead of banks for AI-agent settlement. On the hardware side, TSLA’s FSD mileage, imminent robotaxi rollout, and SpaceX’s compute deal rumors are used to argue for enormous, underpriced optionality, while the piece also leans hard into a China-AI takeover narrative and a bullish MU setup.
BTC bounced off its 200-week moving average and is back above it, framed as the bear-market support level.
Joe Consorti-style power-law floor numbers put BTC at $2 in 2011, $62 in 2013, $154 in 2015, $475 in 2016, $3,000 in 2018, $3,800 then $3,600 in the March 2020 dip, $16k in 2022, about $60k in 2026, $188k in 2030, and $582k in 2034/2035; the path from 2026 to 2035 is described as $60k to $580k, or 10x in 10 years, as a floor path rather than a top.
BTC ETFs were green 5 of the last 6 days, after two green days in a row and then a sell-off day.
CryptoQuant/Axel data says long-term holders added 371,000 BTC over just over 30 days, bringing long-term holder holdings to 16.34 million BTC and leaving only about 3.5 million BTC available for sale if big buyers show up.
SOL is said to process 96% of tokenized equities spot volume, with $3.5 billion handled in June 2026 versus $141 million across every other chain, which is why it is called a quasi-monopoly.
Tom Lee is cited predicting ETH at $250,000 from about 1,800 across three conferences, implying more than a 50x move and a roughly $30 trillion market cap.
Bully’s case is that ETH is money and the base layer for agentic AI, but the counterpoint is that AI agents will choose the best, quickest, easiest chain with best finality, since ETH is described as running only 15 transactions per second.
TSLA FSD just broke 12 billion miles, with 30–35 million miles added per day; that is compared with driving Earth-to-Moon 51,000 times using a 230,000–238,000-mile Earth-to-Moon distance.
Cybercabs are described as imminent, with robotaxis already all over the country and an “awakening” framed as 4 weeks or 8 weeks away.
Mike Moritz is quoted saying nobody understands what TSLA and SpaceX do because there is no precedent, and that you cannot bet against a level of vertical integration that nobody has.
Stripe is said to be looking to buy PYPL for $53 billion, versus PYPL’s roughly $50 billion market cap, after PYPL fell from north of 3,320 to $50 and bounced on the news.
The AI compute-war pricing spread is laid out as a ‘hot dog’ equal to 1 million tokens: Anthropic at $56, OpenAI at $26, Meta/Google/xAI around $1–$2 with xAI the cheapest, and Chinese models around 50 cents.
The China AI argument is that Kimi 3 is number one in front-end code arena ahead of Claude 3.5, GPT 5.6, JLM 5.2 and Opus 4.8, and David Sacks is cited saying China is also at or near the frontier on other benchmarks while US regulation blocks new data centers and frontier-model rollout.
SpaceX compute is described as having current annual customers paying $28–$29 billion combined from Reflection AI, Google and Anthropic, with a Pentagon customer adding another $10–$20 billion a year; the stock moved about 122 to 130 on the rumor and then back to around 124.
Lummis is said to have posted 91 times about the Clarity Act being ready, but the passage odds are put at 40% by year-end; China is also described as having more electricity, power, natural resources, rare earths and about 130 humanoid companies, making it a force to be reckoned with.
MU is framed as a buy-the-dip setup: $100 of capital should either be split into a tokenized version or about $1 into MU because a real share needs roughly $800; the call is for MU to reach $1,600 within about a year, with the suggested tactic to layer in, then triple down if it falls.