Nvidia Won’t Bring Down the U.S. Economy or the AI Boom The Nvidia panic has hit warp speed. When Jensen Huang’s adventures with an alleged Chinese spy lady were first revealed, we were told that the removal of Nvidia’s chief executive would bring down the stock market. Now the alarm bells are ringing over proposals to simply further restrict sales of advanced superintelligence hardware to China. This will bring down the entire U.S. economy, the apocalyptic Branch Nvidians have declared. We’re not making that up. They really said it. “China Hawks are going to get a chip ban, and Americans are going to get a recession,” Jason Calacanis—whose insights into technology and its financing we’ve often admired—warned in an X post. We suspect the American economy can survive making it harder or even impossible for China to buy Nvidia chips. Listening to the tech bros, however, you might think the country’s prosperity depends on keeping Beijing supplied. Calacanis’s All-In cohost Chamath Palihapitiya warns that disruption to the “Bank of Nvidia” could subtract 50 to 150 basis points from economic growth. Add political obstacles to artificial intelligence development, and he sees a recession by 2028. Nvidia is important, but it just isn’t that important. The VC tech bros crowd is giving Nvidia far too much credit for the economy around it. The Bank of Nvidia Is Necessary The AI investment boom is a big deal. J.P. Morgan Asset Management estimated in September that AI-related investment accounted for roughly one-fifth of economic growth over the preceding year, contributing 0.47 percentage points after subtracting imported hardware. Breckinridge Capital Advisors reviewed estimates in November ranging from 0.3 to 1.2 percentage points of growth this year, with the higher figures including more software and related spending. But it takes a considerable leap to put all that growth on Nvidia’s tab. The idea of a “Bank of Nvidia” refers to the company’s practice of helping finance its customers. Nvidia buys stakes in businesses that buy its chips, backs certain obligations, and agrees to purchase computing capacity that cloud operators cannot sell. A lender considering a data-center loan has less to worry about when Nvidia promises to buy otherwise unsold capacity. Look at CoreWeave as an example. It is a cloud computing company that specializes in renting access to powerful computing arrays for artificial intelligence. In the jargon of the sector, it is a “neocloud” company. And it is a customer of the “Bank of Nvidia.” It has a $6.3 billion arrangement in which Nvidia agrees to buy its unsold capacity through 2032. That guaranteed revenue means that other customers and sources of financing are comfortable doing business with CoreWeave. Yes, that sounds scarily circular. The Bank of Nvidia uses its capital to fund its customers who then buy hardware designed by Nvidia. But it can make sense for a capital-rich supplier to finance its customers, so long as it is confident those customers really will have the profits to earn it a return on that financing. What this means, however, is that some of those customers are partially dependent on Nvidia’s revenues to finance their own growth. A decline in Nvidia sales or profits due to trade restrictions might force Nvidia to scale down its customer financing. And the customers most exposed to a retreat are the heavily indebted “neoclouds,” businesses that buy chips and rent computing capacity to others. But even with these, Nvidia is more of a financer of the residual funding needs. Customer contracts with Microsoft, Meta, and OpenAI are far more important. And those neocloud companies are not the biggest drivers of AI investment in the United States. The companies that are—Microsoft, Amazon, Google, and Meta—have cash flows and access to credit that do not depend on Nvidia underwriting their expansion. If the neocloud companies cannot finance expansion to meet the demands of the hyperscalers, the hyperscalers will wind up building more themselves or providing the financing to the neocloudians. The demand for more compute power does not disappear because a particular financing deal falls through. Nvidia: The Offshoring Powerhouse The Branch Nvidians also overestimate the direct role of Nvidia in the U.S. economy. Nvidia outsources the manufacturing of its chips, much of which takes place overseas. Its American design and research contribute domestic value, but its sales cannot be counted dollar for dollar as American production. Purchases of imported servers add to investment spending that gets included in GDP, but these are offset through imports in the GDP accounts. Domestic construction, design, research, and services contribute far more to American GDP than chip manufacturing or design, especially chip manufacturing for export to China. Nvidia helps finance some of this activity, but it is hardly central to it. Losing that support could complicate particular projects. But the hyperscalers’ investment programs draw on much broader sources of funding. More expensive financing for some developers would not shut down the American buildout. What’s more, additional export restrictions would likely mean Nvidia’s hardware would become less expensive for American buyers. Cheaper equipment or shorter waits would help offset more expensive financing. The cloud operator paying a higher interest rate might need a smaller loan. And then there’s the possibility that slowing down AI investment might actually help other parts of the economy. Just as we put too much capital into homebuilding during the housing bubble of 20 years ago, starving other parts of the economy for investment, there’s a strong risk that we’re now over-committed to AI. If AI investment is crowding out investment in other parts of the economy, a slowdown would release capital for factories, housing, and other businesses. Fed officials, meanwhile, have been worrying that the AI boom is feeding inflation. Governor Christopher Waller has cited AI-related price increases while advocating further rate hikes. A little cooling could do some of the Fed’s work, sparing other borrowers some of the squeeze. A trade off of lower mortgage rates for a bit less AI investment might be something the American public would applaud. China Won’t Be an Nvidia Customer for Long One weird thing about the panic over export restrictions is that direct sales to China make up about eight percent of Nvidia’s sales. Even if we account for actual smuggling to China and neo-smuggling by renting China access to Nvidia chips, the company’s sales will still be the envy of the world. When the Branch Nvidians get this concerned about export restrictions, it raises the question of whether they suspect Nvidia’s dependence on revenue from China is actually much higher than is disclosed or even speculated about. One theory is that the China anxiety is really about future sales. That would explain why such a modest current revenue share supposedly carries such enormous consequences. They aren’t worried about ending the sales today but about pre-empting the sales they thought they’d get when Nvidia would inevitably dominate the Chinese market the way it does the U.S. market. But that’s a fantasy rather than a realistic growth forecast. Beijing has been very clear that it wants China to have its own semiconductor industry. Xi Jinping has called self-reliance in science and technology “the foundation of national prosperity and the key to national security.” He has also declared that “China cannot ask for, buy, or beg for core technologies in key fields.” It can buy Nvidia chips while financing their eventual replacement, but it absolutely will not tolerate long-term dependence on U.S. technology. The theory that selling to China will keep it dependent assumes Beijing would allow superior American products to starve Chinese competitors of customers. This is a failure to properly model the mind of China’s leaders. There’s no world in which they decide to remain dependent on U.S. technology or to allow U.S. firms to dominate its market for AI hardware. Nvidia has helped make our current economic boom possible and profits handsomely from it. But American companies are building because they expect to make money supplying and using AI. They have customers, cash, lenders, and investors beyond Nvidia. More expensive financing for some cloud operators will change some deals. It will not extinguish those opportunities. The AI boom is bigger than the Bank of Nvidia. America’s prosperity does not require keeping that bank’s China business open.
Breitbart Business Digest: Nvidia Is Less Important to the U.S. Economy Than the Tech Bros Think
Full Article
Original Source
Read the full article at Breitbart →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.