Trump Rally or Bessent Put? Elon Back at Tesla, Google's Gemini Problem, China's Thorium Discovery
Speakers
David Friedberg
— CEO, The Production Board
Chamath Palihapitiya
— CEO, Social Capital
David Sacks
— General Partner, Craft Ventures
Andrew Ross Sorkin
— Co-Anchor, Squawk Box
Summary
Chamath Palihapitiya, David Sacks and David Friedberg are joined by guest moderator Andrew Ross Sorkin to debate whether the week's 3 to 7 percent equity rally is a Bessent put or simply the market pricing out the worst-case tariff outcome, and what China's rare earth chokehold means for American supply-chain dependency. They then discuss Apple shifting US-bound iPhone production to India and whether India is actually investable, before breaking down Alphabet's earnings, the hidden value of Waymo and Google's Gemini distribution dilemma against ChatGPT. The show closes with Tesla's jump on Elon Musk's partial return from DOGE, a debate over camera-only FSD, and a science segment on China's thorium reserve and working molten salt reactor.
- Hosts debate whether the rally reflects a Bessent put or the market pricing out a global trade standstill.
- Chamath argues tariffs flipped US daily trade cash flow positive and that the US remains the only investable market for scaled capital, with Europe roughly uninvestable.
- Sacks details China's dominance of rare earth processing and magnets as a critical dependency exposed by the trade war.
- Apple's plan to move US-bound iPhone manufacturing to India prompts a debate on the 18-month timeline and on India's investability.
- Alphabet earnings reviewed: about 90 billion in revenue, a 4 to 5 percent dividend and buyback yield, cloud growing 30 percent, and Waymo valued at zero.
- Google's innovator's dilemma: Gemini benchmarks well but lags ChatGPT on usage, with calls to put it in front of 270 million paid subscribers.
- Chamath says Nvidia's roughly 47 percent Asia revenue is a billing-address artifact and flags cloud KYC risk for Google Cloud, Azure and AWS.
- Tesla rallies on Musk's return, with FSD quality praised and camera-only sun-glare disconnects flagged as a robotaxi risk; science corner covers China's thorium and fusion lead.
Ideas
Tail risk priced out, deals coming.
The rally is not a rescue bid. What got priced out this week was the probability the market was assigning to this administration being crazy enough to leave tariffs very high and let the global economy grind to a halt. Everyone now realises Bessent and Trump are not going to let trade come to a standstill, and the indications are that they intend to get deals done, which means their intention is to make sure the market does not tank. The flip side is that with that tail risk out of the market there is less leverage in the negotiation than there was three weeks ago.
China's rare earth chokehold must break.
China used its WTO developing nation status to subsidise and take over deliberate choke points in the global supply chain, and rare earths are the clearest example. The ore itself is distributed all over the world, but the processing is expensive and complicated and China now accounts for over 90 percent of it, plus over 90 percent of the cast rare earth magnets that are a critical component of essentially every electric motor, in cars and in many other products. China has now cut the United States off as leverage in this trade fight, which exposes a critical dependency on a country of concern that has to be corrected very rapidly. Security has to be worked out first, then trade.
No investable alternative to the US.
Investing is always a relative exercise. A very successful Brazilian billionaire he met this week said there is still no other country than America where he can put money to work in any size, and pushing him on China and India did not change that: China has its vagaries, India has different problems, and Europe is roughly uninvestable. On a relative basis the United States is still the shining beacon on a hill and there is no investable alternative for scaled capital. That is also why Ken Griffin, for all his criticism of the tariff policy, is massively long Pax Americana with a 42 billion dollar firm and a billion dollars of American real estate rather than investing in Ecuador.
No investable alternative to the US.
Investing is always a relative exercise. A very successful Brazilian billionaire he met this week said there is still no other country than America where he can put money to work in any size, and pushing him on China and India did not change that: China has its vagaries, India has different problems, and Europe is roughly uninvestable. On a relative basis the United States is still the shining beacon on a hill and there is no investable alternative for scaled capital. That is also why Ken Griffin, for all his criticism of the tariff policy, is massively long Pax Americana with a 42 billion dollar firm and a billion dollars of American real estate rather than investing in Ecuador.
US-India security alignment makes India investable.
If you build supply chains and trading relationships with an adversary you should expect them to be disrupted, but India is fundamentally aligned with the United States on security because it views China as its biggest potential threat, exactly the way Washington sees the world, and that will not change for probably the entire century. Therefore the investments that get made in India and the trade relationships that get forged there should be very stable over the next couple of decades. The structural realities simply make India a good place to invest.
Apple's India shift is further along.
The reported move of US-bound iPhone manufacturing to India is not starting from a standstill. Foxconn and others have been standing up Indian capacity for some time and probably already have one or two model runs active in the country, so this is a replication and scale-up problem rather than recruiting people, designing processes and building facilities from scratch. That is why Apple has confidence in the stated timeline, and Tim Cook does not stand up and commit to something in 18 months and then miss it by years the way others do.
Own Indian infrastructure and resources, not tech.
India has one fifth of the labour cost of China, which itself has one fifth of the labour cost of America, plus an educated and young workforce, so it is the natural home for the next generation of manufacturing labour. It looks like China circa 2003, and China grew GDP at about 10 percent a year from 2006 to 2012, so India should have that moment over the next 20 or 30 years. The hard part is expression: a decade of Indian tech investing returned him roughly 50 cents on the dollar because he imposed an American mental model and only copies of American businesses worked. After meetings with the Ambanis and Adanis he pivoted to infrastructure and hard assets, including a large rare earth mine and specialty chemicals processing negotiated with the prime minister's office to supply the United States, and that has worked.
Cheap, resilient, non-search growth supports Alphabet.
Alphabet is far more resilient than the narrative that AI kills search implies. About 10 billion dollars a year of dividends plus a newly announced 70 billion dollar buyback against a two trillion dollar market cap is roughly a 4 to 5 percent yield just for owning the stock. Of the 90 billion dollar quarter, only about half is search advertising: YouTube is 9 billion, non-Google ads 7.5 billion, subscriptions 10 billion across 270 million paying subscribers, and cloud 12 billion growing 30 percent year over year. The downside case is losing half of search to ChatGPT, and with the rest of the business compounding you get back to parity fairly quickly; the upside case is reinventing search around chat and keeping the audience. At 18 times free cash flow, with that yield and that gap between downside and upside, it is a powerful business at a good price, not one in decline.
Market gives Alphabet zero for Waymo.
Google's problem is distribution and taste rather than model quality, because Gemini is a meaningfully better model than anything else on multiple dimensions. With 270 million paid subscriptions across YouTube and Google One, plus Gmail and Workspace, Google should put Gemini at the front door of exactly those surfaces that do not cannibalise the blue links, habituate 300 to 500 million people a week, and work out google.com and the derivative revenue later. Instead it is shipping junior, tasteless implementations such as Gemini pop-ups scattered through Gmail. Sundar, Larry and Sergey have the equity and political capital to impose one person's taste and force the decision through; if they keep hemming and hawing while OpenAI compounds its usage lead, they will look back in four years and regret it.
Nvidia's China revenue is billing-address artifact.
After last week's segment the Nvidia team walked him through the filings, and the roughly 47 percent of revenue attributed to China, Singapore and Taiwan is reported on the billing or invoicing address, not on where the silicon is shipped. Those invoices are largely major US companies, OEMs and clouds that route billing through a Singapore office, while the product is almost always shipped elsewhere, to the US, Mexico, Taiwan and so on. So the headline geographic mix badly overstates Nvidia's real China exposure.
Cloud KYC rules threaten hyperscaler revenue.
A study by Artificial Analysis showed the cloud providers are not doing any real form of KYC on who is actually using their compute, which is how Chinese entities can reach restricted capacity through a cloud loophole. If western governments react by requiring the clouds to verify customers so that nobody is distilling the best western models into places we do not agree with, there could be blowback on that revenue. This is a Google Cloud problem, an Azure problem and an AWS problem, and it is going to have to get sorted out over the next couple of quarters.
Hyperscaler AI capex is inelastic.
AI capex holds up even if there is a tariff-driven hiccup between now and whenever the trade fight is settled. The data centre buildout is not a discretionary growth investment for the hyperscalers; it is good for their businesses but they see it as strategic to their survival, so they simply have to do it no matter what the macro does. In other words they are totally inelastic on this spend, which keeps the money flowing through the whole ecosystem around them.
Gemini usage gap leaves Google exposed.
Gemini scores well on the benchmarks and has substantially caught up on capability, but it is not getting the usage while ChatGPT keeps growing like crazy and users learn a new habit of going there for their questions. Making Gemini the default interface in search is a true innovator's dilemma because doing so could give up more than half of search revenue, and bolting AI summaries onto the twenty blue links is clunky and does not compare with a product that is all in on the AI experience. The longer Google waits to put Gemini front and centre, the worse the usage problem gets, so Google is in a really tough spot.
FSD step-change makes Tesla robotaxis credible.
Full Self Driving took a step change in roughly the last month or two, going from something like a 90.8 to a 99.5 in his experience, with very few disengagements. It is the reason he bought the car, he uses it all the time, and functionally his Tesla is already a robotaxi except that the interior camera forces him to keep his hands on the wheel and look ahead. On the strength of that product experience he believes Tesla really is doing real robotaxis within two years.
Camera-only FSD risks robotaxi disengagements.
There is a real technical risk in Tesla's camera-only approach to autonomy. Videos are circulating of FSD driving into the sun where, at a certain angle, the light hitting the cameras forces an emergency disconnect, and he has had that happen himself, so he knows it is real. Waymo runs a never-disengage service with lidar on board; without lidar it is questionable whether Tesla can operate a full driverless robotaxi service to that standard.
Up Next
This All-In Podcast video, published April 26, 2025, features David Friedberg, David Sacks, Chamath Palihapitiya discussing SPY, REMX, VTI, VGK, INDA, AAPL, GOOG, NVDA, MSFT, AMZN, AI Data Center Infrastructure, TSLA. 15 trade ideas extracted by AI with direction and confidence scoring.
Speakers: David Friedberg, David Sacks, Chamath Palihapitiya · Tickers: SPY, REMX, VTI, VGK, INDA, AAPL, GOOG, NVDA, MSFT, AMZN, AI Data Center Infrastructure, TSLA