Musk throws out another astonishing target: SpaceX annualized revenue of $500 billion next year, Starlink computers going to space simultaneously, Terafab chip capacity aiming for one terawatt.

Musk has painted another astonishing growth blueprint for the AI industry. On Friday local time, the 33rd Baron Investment Conference held a virtual fireside chat lasting about seventy minutes. The host was Ron Baron, founder of Baron Capital, and Elon Musk attended remotely. During the interview, Musk gave a series of blockbuster predictions: SpaceX's annualized revenue is expected to exceed US$100 billion this December, and annualized recurring revenue is expected to surpass US$500 billion next December; Terafab aims to produce one terawatt of chips annually, equivalent to 30 times next year's global AI chip output; the Starlink computers jointly designed with Nvidia are expected to begin entering orbit in Q3 2027; agent internet traffic has already surpassed human traffic and could reach 1,000 times human traffic within five years. SpaceX revenue: from US$100 billion to US$500 billion in just one year SpaceX was the core subject of greatest market attention in this interview. Musk explicitly gave two revenue targets: "We are on track to achieve annualized revenue exceeding US$100 billion in December. Obviously, December is very close, so I am quite confident about this. Unless some kind of natural disaster occurs, we will reach that target." Going further, he was equally optimistic about next year: "Then according to our plan for next year, annualized recurring revenue exceeding US$500 billion next December looks very likely. This is not a promise, it is just what I think will happen." As for US$1 trillion in annualized revenue, Musk said: "US$1 trillion is not impossible. It was impossible before, but now it is no longer completely impossible, perhaps around a 2% probability. But what is interesting is that it is not 0%. That is the key. And that probability is increasing over time." SpaceX's revenue last year was about US$15 billion to US$16 billion. If the above path comes true, that means annualized revenue will expand more than 30-fold in less than two years. Terafab: targeting annual output of 30 times the world's AI chips Musk elaborated in the interview on the Terafab plan led by xAI. He explained the naming logic: "When I say terafab, sometimes people think terafab has three possible meanings: one is terrifying - TERROR, then TERRA, which means Earth, but it is actually TERA, meaning terawatt. But you could say all three meanings apply." In terms of scale, Terafab's goal is to produce chips corresponding to one terawatt per year. According to Musk's forecast, next year's total global AI chip output will be about 35 to 45 gigawatts - meaning Terafab's single-factory capacity would be about 30 times the global annual output by then. Musk pointed out that there is huge room for improvement in existing chip manufacturing efficiency, illustrating with one detail: "Right now, in a chip fab, more than half of a wafer's time is spent just on transport and waiting... sometimes the time in which atoms actually undergo useful changes is only about 20%." His solution is to build a research fab in Austin and try to concentrate design, manufacturing, and testing under one roof, "I think we may be able to shorten what usually takes a year to one month." The underlying logic is: "For any given technology endeavor, the rate of progress depends on how many iterations you can do... as long as you have the ability to iterate an order of magnitude faster, roughly speaking, your technological progress will also be an order of magnitude faster." Space computing: first batch of Starlink computers to launch next year Musk repeatedly positioned space computing as a strategic-level asset in the interview, with a clear chain of logic. In his view, the expansion of ground-based computing is constrained by power: average US electricity consumption has only just exceeded 500 gigawatts, "if you build 200 gigawatts of AI per year, that is equivalent to adding 40% to the entire US power grid every year. That is simply not feasible." The advantage of space lies in continuous sunlight and no atmospheric attenuation - CECEP Solar Energy utilization can be improved by an additional roughly 30%, and there is almost no need for large-capacity energy storage. "Once you have a very low cost per ton to orbit, the cost of intelligence in space will be much lower than on the ground." Musk disclosed that satellites currently being launched have already carried Nvidia GPUs and Google TPUs for testing, "and as far as we know, they are operating well in space." The first batch of Starlink computers jointly designed with Nvidia is expected to begin launching in Q3 2027, scale to a considerable level in 2028, and reach "a very large scale" in 2029. He further gave a revenue estimate: "If we deploy 100 gigawatts of AI computing to space each year, even if the value per gigawatt falls to US$10, that is US$1 trillion a year. And it is cumulative; another US$1 trillion will be added every year, because satellites do not evaporate." If the unit price is US$50, that would be US$5 trillion per year. Musk predicted that if one terawatt of computing is sent to space each year, that would be equivalent to sending digital superintelligence into space each year at twice the scale of current average US electricity consumption. Starlink: 100-fold bandwidth growth, agent traffic to be 1,000 times human traffic Musk disclosed that about 11,000 Starlink satellites are currently in orbit, mainly V2 versions, with V3 satellites just beginning to launch. V3 has more than 10 times the capability of V2, and the number launched will also be 10 times, "the bandwidth Starlink generates in the next two to three years will be 100 times what it currently provides." The key variable driving demand is AI agents. Musk said: "My understanding is that, in fact, agent internet traffic has already surpassed human traffic, if I am reading Cloudflare's data correctly. And I believe Cloudflare's forecast that within five years, agent AI traffic will be 1,000 times human traffic. And I think that even underestimates the scale of agent AI traffic." Data centers: US$35 billion investment, US$50 billion recovered in one year Musk introduced the business logic of xAI's data centers: self-building takes about 8 to 9 months, far faster than the two and a half years of external contractors; investment is about US$35 billion, below market levels. "We will rent out the computing power before our own Grok needs it, collecting about US$50 billion a year. A US$35 billion investment, recovering US$50 billion in one year." He also disclosed that xAI built an internal power plant of more than one gigawatt during the construction of the Memphis data center, "taking about one-fifth of the time normally required to build such facilities - about one year to build, whereas it usually takes about five years," and revealed that he personally participated in cable laying, fiber wiring, and electrical installation. Revenue growth from Q2 to Q3 this year exceeded 80%, and from Q3 to Q4 is expected to be about 70%. Humanoid Siasun Robot&Automation and AI's impact on the economy Musk estimated that digital superintelligence will add about 20% to 30% to the global economy, equivalent to about US$25 trillion, "and this will happen very quickly," and may double white-collar productivity. On physical AI, he gave a preliminary estimate of Siasun Robot&Automation replacing labor: an efficient humanoid Siasun Robot&Automation can be approximated as working 168 hours per week, with minimum productivity expected to be about five times that of a human; "rough estimate, about 20 million to 30 million Siasun Robot&Automation units can match US output of goods and services." In terms of business model, he prefers leasing rather than selling. On OpenAI Musk once again stated his position on OpenAI, in direct terms: "I did found OpenAI as an open-source nonprofit organization. Now it is a closed, maximum-profit entity, which is the complete opposite of what I founded." He said he did not expect at the time that OpenAI would one day convert into a for-profit company, and he was confused by the structure of holding commercial interests within a nonprofit organization: "I was confused at the time about how you could have a commercial interest inside a nonprofit organization. That did not make sense to me." Regarding Anthropic, he pointed out that it was founded by seven former OpenAI employees who left, "in effect, Anthropic is a branch of OpenAI." Starship and ultra-fast reuse Musk disclosed that about a week after the interview, they would attempt to catch the returning Starship spacecraft with the tower mechanical arm "chopsticks," achieving ultra-fast reuse in the full sense. He explained the business logic of this system: "In principle, it can allow us to relaunch the rocket once per hour, that is, that giant booster, the Super Heavy booster." Starship is a fully reusable orbital-class launch system, fundamentally different from Falcon 9 - Falcon 9's upper stage cannot be recovered, "equivalent to throwing away a medium-sized jet aircraft on every flight." He expects that in the extreme case, the cost per ton to orbit "may be lower than the shipping cost from California to Australia by air," a reduction of more than 100 times compared with the historical peak. This article is reproduced from the "Wallstreetcn" APP, author: Bu Shuqing; GMTEight editor: Song Zhiying.