Original Title: "Q2 Revenue Surges 92%, Still Below IPO Price, Should You Buy SpaceX or Wait?" Original Author: Golem, Odaily Planet Daily On August 4, SpaceX released its Q2 2026 financial report, with overall results exceeding expectations. The report shows that SpaceX's Q2 2026 revenue reached $7.814 billion, a year-over-year increase of 92%, surpassing the market expectation of about $6.8 billion. The connectivity business, mainly Starlink, still accounts for the majority of revenue, generating $4.291 billion, up 66% year-over-year and 32% quarter-over-quarter; AI business revenue grew rapidly this quarter, reaching $2.561 billion, a 247% year-over-year surge and a 213% quarter-over-quarter increase; the space business also recorded a 29% year-over-year growth, with revenue of $962 million. SpaceX still posted a net loss of $541 million this quarter, but this was within market expectations and significantly narrowed compared to a $1.008 billion loss in the same period last year. The loss per share was about $0.09, better than the market expectation of a $0.23–0.24 loss per share. Although the AI and space businesses remain unprofitable, losing $1.257 billion and $542 million respectively, which consumed $1.656 billion of Starlink's profit, losses have narrowed compared to market expectations and the previous quarter. As SpaceX's first financial report since going public, the revenue and expectations indicate that SpaceX deserves praise. Dramatically, after the earnings release, SpaceX's stock price experienced rollercoaster trading. On August 4, after the U.S. market opened, SpaceX's stock rose steadily, closing at $125.33, up 9.43%; however, in after-hours trading, the stock price turned down, currently down 7.5%, hovering around $116, essentially erasing all intraday gains. The market's harsh reaction to SpaceX is understandable, mainly due to two reasons. First, on the macro level, the market's focus on AI companies has shifted from revenue growth to concerns about when the massive AI capital expenditures will translate into actual returns; second, on the micro level, the first batch of about 910 million restricted shares will be unlocked on August 6, bringing selling pressure to the market. Elon Musk cannot calm market worries over massive AI capital expenditures. From the report, AI has become SpaceX's second-largest revenue source, second only to the connectivity business led by Starlink. For SpaceX, which has always regarded AI business as the core of its valuation, this is great news, but behind the rapid growth lies massive capital expenditure. SpaceX's total capital expenditure in Q2 2026 was about $18.369 billion, with AI business capital expenditure alone accounting for $15.828 billion, over 86%. This AI capital expenditure not only increased 104% from the previous quarter but is also six times the AI business revenue this quarter and twice the total revenue ($7.814 billion). Where did the massive AI capital expenditure go? According to SpaceX CFO Bret Johnsen during the earnings call, over $15.8 billion of AI spending was mainly used to support AI computing infrastructure construction. By the end of Q2, SpaceX's total computing power power load had expanded to 1.4 GW, up from 1 GW in Q1 and 0.4 GW in the same period last year. SpaceX expects to reach 2 GW by the end of this year. Musk also added that by the end of 2027, SpaceX's actual computing capacity will approach 10 GW instead of the previously expected 5 GW. Johnsen also stated that AI capital expenditure will not significantly decrease in the next two quarters but emphasized that current economic benefits allow SpaceX to recoup its AI computing capital investment in less than a year. To further ease investors' concerns about the return cycle of SpaceX's AI capital expenditure, Johnsen cited a recently signed $6.7 billion cloud service contract during the earnings call, stating that SpaceX will achieve $100 billion in annual recurring revenue (ARR) by the end of this year. Bold Musk added, "This goal is not out of reach; even if we do nothing, we can achieve it, and I think the actual number might be higher." Musk also revealed that SpaceX internally advanced its revenue forecast of $1 trillion (note: revenue, not ARR) from 2031 to 2030. To give readers a sense of $1 trillion revenue, no company worldwide has yet reached $1 trillion in annual revenue. Even "GPU Godfather" Jensen Huang only expressed confidence at the March 2026 GTC conference that Nvidia would achieve $1 trillion in cumulative revenue by 2027. Compared to that, Musk's boast is even bigger. In SpaceX's AI business revenue of $2.561 billion, "AI solutions and infrastructure" revenue reached $2.194 billion, while advertising revenue was only $367 million. Undoubtedly, this revenue growth was driven by massive AI capital expenditure, but the AI revenue explosion has not covered the capital speed of SpaceX's AI infrastructure expansion, nor changed the pattern that the AI business still relies on Starlink for funding. Market concerns over massive AI capital expenditure have dragged down SpaceX. For SpaceX itself, this is not a big problem. As a long-cycle, high-growth tech company, high upfront capital investment is necessary. Moreover, newly public SpaceX holds $100 billion in cash equivalents and $47.5 billion in unfulfilled orders, giving it strong financial power to "afford" this. But SpaceX's mistake is that it coincided with a period of poor market sentiment. On the macro level, almost all AI tech companies are expanding AI infrastructure investment, but the market's patience for when this massive spending will yield actual returns is rapidly wearing thin. Even "Silicon Valley Iron Man" Musk cannot calm investors' anxiety. Last week, as the U.S. entered Q2 earnings season, Google, Intel, Meta all saw revenue surge but their stock prices plummeted; their commonality was increased AI capital expenditure. Meanwhile, Microsoft cut its full-year capital expenditure, and its stock price hit the best single-day performance in 18 years. When the whole market doubts AI capital expenditure, how can SpaceX, with AI capital expenditure accounting for 86% and AI business still losing money, have an independent rally? The AI capital expenditure black hole is an objective fact for SpaceX; the key is how investors view it. Some worry, others are optimistic. Morgan Stanley maintains a $300 target price for SpaceX, mainly betting on the future value of its AI business. The $300 target price breakdown is: space business $8, Starlink business $128, X and Grok business $12, enterprise AI business $152. Excluding AI value, SpaceX's rocket launch and Starlink businesses alone can support the current stock price, so with AI business included, SpaceX is currently trading at a discount. August 6 Stock Unlocking On the macro level, investors have mixed views on SpaceX's AI capital expenditure, but on the micro level, investors are generally bearish on the first early stock unlocking of about 910 million internal shares on August 6, which is one reason SpaceX's stock price is under pressure after the earnings report. On August 6 (this Thursday), about 910 million internal shares of SpaceX will be unlocked and enter the market, far exceeding the current public float of only 640 million shares. After this unlocking, the supply of shares in the market will increase significantly, and the market generally believes that current buying demand is insufficient to absorb such a large sell-off. According to OptionCharts data, the current SPCX put-call ratio is 0.97, indicating neutral investor sentiment. But looking at contracts expiring on August 7, SPCX's psychological and technical support level has dropped to $100, with 24,947 puts accumulated at this strike price, the largest open interest contract currently. If this support breaks, a downward breakout could trigger a negative Gamma effect, and SPCX falling below $90 is also very likely. Currently, the battle between bulls and bears centers around the $110–115 strike price range. Considering two scenarios: if the market has already priced in the negative impact of the stock unlocking, SPCX may dip briefly on August 6 but then rebound, staying in this price range; if the market has not fully priced in the unlocking downside, SPCX is likely to continue breaking down as described above. After the earnings beat was quickly digested during yesterday's U.S. trading session, it is difficult to foresee any short-term positive factors that could boost SpaceX's stock price. This does not mean SpaceX's stock is expensive in the long term, as SPCX has already dropped 17% from the $135 IPO price, but if you want to buy, it's best to wait at least until after the first wave of stock unlocking ends. Original Link

At the beginning of August, the market's first turnaround was not gold, but crude oil. Expectations of the Strait of Hormuz reopening for navigation pushed oil prices down. According to the usual script, safe-haven assets should have cooled off as well. However, gold moved upward. According to Reuters on August 6, spot gold was quoted at $4,285.84 per ounce, rising for the fourth consecutive day to the highest level since mid-June. This is easily framed as "geopolitical risks still persist." But another clue from Reuters is faster: after oil prices fell, the US dollar and US Treasury yields weakened simultaneously, and the market began to recalculate how high the Federal Reserve would raise interest rates. Gold's rise first occurred at the moment when the opportunity cost of holding it was repriced. Short-term changes first affect rate hike pricing. There is no direct supply-demand transmission between crude oil and gold. Oil influences market inflation expectations. When energy prices stop pushing higher, the urgency to continue raising rates diminishes. According to Reuters on August 6, market expectations for further rate hikes in September dropped from 67% to 55% within two days. The same report mentioned that US Treasury yields fell and the US dollar index came under pressure. This explains why a seemingly risk-reducing geopolitical message can instead provide short-term support for gold. Chart data comes from Reuters intraday quotes on August 4, 5, and 6. The three quotes in the chart are not daily or closing prices. They are like snapshots taken by the market at different times. As prices rose, rate hike pricing retreated; these are just two sides of the same macroeconomic revaluation. Data from the St. Louis Fed FRED database shows the 10-year Treasury Inflation-Protected Securities (TIPS) yield dropped from 2.47% to 2.40%. For gold, this is not an abstract macro term. It means the risk-free return on money not invested in gold has decreased slightly. The threshold for zero-yield assets lowers, and a weaker dollar reduces the pricing cost for overseas buyers, providing a foothold for short-term buying. Price revaluation did not bring more tonnage. Rapid price increases can easily mislead people into thinking the whole world is rushing for gold. The World Gold Council's Q2 data paints a quieter picture: the London Bullion Market Association's midday gold price average rose 37% year-on-year, but total gold demand including OTC remained roughly flat at 1,269 tons. The World Gold Council places these two facts in the same summary. The key point of this chart is not that "demand did not grow," but that the definition of demand has changed. Total demand includes OTC and other balancing items; it is not the total amount of gold bars moved by retail consumers. The nearly unchanged tonnage in the chart indicates that price revaluation happened first, rather than every buyer category suddenly increasing purchases. The World Gold Council's first half statistics also show demand value hit a record $380 billion, while demand volume only increased 2% year-on-year. This data indicates that the increase in demand value did not translate into a proportional increase in tonnage. Prices reflect changes in the weighting of different demand segments, not a uniform increase in purchases across all buyer types. ETFs are not the total ledger of gold. The most noticeable sellers came from gold ETFs. In Q2, ETFs and similar products turned to net outflows. Meanwhile, the World Gold Council recorded a rebound in net gold purchases by central banks and other official institutions, and OTC and other segments also expanded. Viewing these segments side by side shows that the gold market does not breathe solely through a single public holdings curve. However, this chart should not be read as a "delivery note" showing who took over the ETF sell-off. The World Gold Council clarifies in its methodology that OTC and others also include exchange inventory changes, unobserved manufacturing inventory changes, and statistical residuals. It shows that public ETF outflows do not mean the entire market lacks absorption, nor can it be traced to a specific country or type of capital. Central bank data should also not be drawn as an always upward straight line. The World Gold Council has revised down its official gold purchase estimates for Q1 due to reporting and statistical lags. Treating ETFs as the only thermometer and central banks as the only buyers compresses a multi-layered market into a single story. The oil price decline brought a short-term drop in opportunity cost. The decoupling of price and tonnage, and the differentiation between public and non-public flows, indicate that when gold rises, the market's real change is often in the holder structure.

BlockBeats news, on August 6, according to TradingBeats (formerly Hyperinsight) monitoring, an address starting with 0xd9a5 initiated a 30-minute ETH sell TWAP this morning, planning to short 52,956 ETH, with a nominal amount of about 100.4 million USD. If fully executed, it would directly become the largest ETH short on Hyperliquid. However, this order was terminated after running for only about 45 seconds, with only 1779.67 ETH sold, approximately 3.373 million USD, completing 3.36%. An address with only 335,000 USD in equity almost placed a short order equivalent to 299 times its own funds, then withdrew about 97% within a minute, which looks very much like a "slip of the hand." This does not seem to be the first "slip." Yesterday, this address initiated 39 ETH TWAP orders, of which 31 were for shorting, and all were terminated early; the median running time was only 16.5 seconds, with a median completion rate of about 3.42%. In the previous two rounds, this address actually sold about 1779.67 ETH, allowing it to observe whether a single round of about 880 ETH on Hyperliquid can be quickly executed, the average slippage size, and whether the price would significantly drop after continuous selling. It is judged that this abnormal execution may involve two behaviors: 1. Using exaggerated parent tasks to quickly generate two market price slices close to the account limit, which is more convenient than manually placing continuous orders. 2. Through the first and second executions, judging how much active sell orders the current ETH order book can bear and the real execution slippage. Although these huge orders are repeatedly canceled and replaced, the actual executed volume is not small. Yesterday, this address cumulatively opened shorts of 58,200 ETH, then bought them all back, with a bilateral transaction amount of about 218 million USD; the gross profit for this round was about 398,000 USD, and after deducting about 62,000 USD in fees, the net profit was about 336,000 USD. As of press time, it only has 1779.67 ETH short positions left, with a position value of about 3.374 million USD, using 10x full margin leverage, with an average entry price of 1895.08 USD, currently floating a loss of about 1265 USD, and the account has no other running TWAP or pending orders. Whether this 100 million USD short order was a "slip of the hand" or its fixed strategy remains uncertain; but from the 39 similar operations yesterday, it at least does not look like a one-time accidental trigger.

BlockBeats news, on August 6, according to Farside data, the total net inflow of US Bitcoin spot ETFs yesterday was $244.4 million. Among them, BlackRock IBIT had a net inflow of $196.8 million, ARKB a net inflow of $37.6 million, FBTC a net inflow of $11.3 million, BITB a net inflow of $10.6 million, MSBT a net inflow of $2.8 million; HODL had a net outflow of $14.7 million, and the remaining ETF fund flows were zero. The total net inflow of Ethereum spot ETFs was $60.8 million. Among them, BlackRock ETHA had a net inflow of $50.3 million, ETHB a net inflow of $4.9 million, FETH a net inflow of $2.9 million, ETHW a net inflow of $1.4 million, TETH a net inflow of $1.3 million, and the remaining ETF fund flows were zero.

BlockBeats news, on August 6, according to Lookonchain monitoring, Arthur Hayes purchased another 10.9 million Ethena, worth about $985,000. In the past 5 days, he has accumulated a total purchase of 22.64 million Ethena, with a total value of about $2 million.

BlockBeats news, on August 6, according to TradingBeats (formerly Hyperinsight) monitoring, Sandisk (SNDK) reported a price of $1245.2 after the earnings report, down 12.6% in 24 hours, with trading volume expanding to about $818 million, ranking first among Hyperliquid traditional market targets. The price drop was not accompanied by capital outflow. SNDK open interest increased from about 104,800 contracts 24 hours ago to 131,500 contracts, a rise of about 25.4%; the corresponding position value rose from about $149 million to $164 million. Price decline and inverse expansion of OI indicate that longs are retreating after the earnings report, while a new round of long and short positions are still entering. The funding rate has also turned from negative to positive, currently about +0.0016% per hour, accumulating about +0.0164% over the past 24 hours, showing that the market overall still has capital counter-trend buying; however, more significant top addresses are executing the opposite operation. In the past 24 hours, million-dollar-level addresses reduced long positions by about $3.945 million; during the same period, short positions increased by about $33.585 million, a net increase of about $21.924 million. The value of long positions decreased by 8.6%; the value of short positions increased by 44.9%. The long-to-short position value ratio dropped further from about 0.94:1 to 0.59:1. The number of long addresses did not increase, while short addresses increased by 5. Currently, 3 whales collectively hold about 14,800 short SNDK contracts, with a position value of about $18.412 million, with average entry prices ranging from $1261.6 to $1279.4: - Address starting with 0xfc27 flipped from long to short after the earnings report, currently holding about $8.163 million in shorts, with an unrealized profit of about $161,000; - Address starting with 0xfc66 also switched from long to short, with short positions valued at about $3.463 million, unrealized profit about $45,000; - Address starting with 0x49e9 continues rolling short positions, currently holding about $6.787 million in shorts, unrealized profit about $186,000. Meanwhile, counter-trend bottom-fishing capital has begun to face pressure. Whale address starting with 0x7fda added about 1,716.5 longs today, currently holding 2,734.8 SNDK long contracts, with a position value of about $3.406 million, average entry price $1342.9, unrealized loss about $267,000. The above shorts have not yet shown clear exit actions. No "reduce only" take-profit buy orders or stop-loss orders have been placed; a small number of two-way ordinary limit orders are closer to rolling adjustments. Subsequent monitoring will continue to track whether the $18.41 million short position begins large-scale covering; whether the order book shows concentrated "reduce only" buy orders.

BlockBeats news, on August 6, according to previously exclusive Pro data combined with TradingBeats (formerly Hyperinsight) monitoring review: on August 4, when SNDK rose 8.4%, 4 early bulls collectively reduced their positions by about $11.111 million, with the number of million-dollar addresses and the long-short ratio of amounts dropping to 0.75:1 and 0.72:1 respectively; meanwhile, the largest short continued to increase positions, with the 24-hour cumulative funding rate rapidly turning negative. On the eve of the earnings report, this signal was further confirmed by address behavior: 0x0ad, after taking profits, was only willing to buy back at a price 12.2% lower than the current price, while 0x0c4 sold long positions and reversed to short $4.79 million. In the 24 hours before the earnings release, million-dollar long positions net decreased by $3.945 million, and short positions net increased by $21.923 million. One tracked "storage all-winner" who took profits on SNDK before the earnings and planned to wait for a lower price to buy back, saw the price drop as expected. Yesterday, this address sold 3,890.8 SNDK at an average price of about $1,390, realizing a profit of approximately $401,000, capturing about 8% of the price increase. It then placed a buy order of $7.5 million below. This operation likely reflects the judgment that the market had already priced in considerable positive news when the price rose to $1,443.6 before the earnings. This judgment was validated after the earnings release. SanDisk's revenue this quarter reached $8.965 billion, a 51% quarter-over-quarter increase, but SNDK still plunged about 11.8% within hours before and after the earnings, currently trading at $1,249.5. From last night to early this morning, the address repurchased in four batches: about $1 million at $1,401; about $3 million at $1,301; about $1.007 million at $1,308.9; and about $1 million at $1,251. As of press time, it holds 4,588.3 SNDK long with full position leverage, valued at about $5.733 million, with a weighted average entry price of $1,309.2, currently floating a loss of about $274,000, a return rate of approximately -4.6%. The address still has a new buy order of $5 million reserved at $1,201. If fully executed, its holdings will increase to 8,751.5 SNDK, with an overall buyback scale of about $11.007 million, expanding the plan by 46.8% compared to yesterday's $7.5 million, and the weighted average entry price is expected to drop to about $1,257.7. Completed transaction records show that since June 25, this address has completed 11 major trades in MU and SNDK, all profitable, with a cumulative net profit of about $2.83 million; out of 16 total trades, 14 were profitable, with a cumulative net profit of about $5.634 million. Previous news: Before the earnings, 2 major whales have repositioned, and profitable bulls are unwilling to buy at the current price.

BlockBeats news, on August 6, according to OnchainLens monitoring, a newly created wallet withdrew 540.17 BTC from BitGo about 3 hours ago, valued at approximately 34.9 million USD.