InvestAnswers host and Jo Bhakdi model a Tesla-SpaceX merger near 30-40% premium

A Tesla-SpaceX merger is coming, and the premium is now priced into the thesis

Discussing Tesla's and SpaceX's back-to-back earnings calls, Jo Bhakdi said the real signal from the Tesla call was not the operational commentary but Elon Musk's confirmation that "there are committees on both sides now discussing the details of the merger." Bhakdi argued this was perceived positively by markets because most investors were not yet pricing in the near-certainty of a merger that he and the host consider close to inevitable.

Bhakdi estimated the merger would require at least a 30% premium on whatever Tesla's share price is at the time, calling it "the most litigated merger in history" that cannot include any pricing tricks. The host matched this, citing specific levels: $420 would only be an 11% premium ("nothing... never be accepted"), $493 would be roughly 30%, and $520 would be about 37%. Both agreed on a rough exchange ratio near three SpaceX shares per Tesla share at current prices, while flagging that a SpaceX rally before the deal closes would change the math and could disadvantage Tesla shareholders the longer the companies wait.

"Everyone knows it has to happen in the next 18 months guaranteed in my opinion." — Jo Bhakdi

Orbital compute: Bhakdi sees the first gigawatt in space by 2029-2030

Bhakdi said he is "very confident" in his modeling that SpaceX's Starship will be "more or less" ready by the end of next year, with Nvidia and SpaceX jointly developing an orbital compute rack (the "Vera Rubin 72") scheduled for Q4 2027. He projected SpaceX could reach scale deployment into orbit in 2028, calling this "not expected" by the market, with the first gigawatt in space arriving in 2029 in his bullish case or 2030 if conservative.

The host pressed on flight cadence, noting SpaceX has asked for a cadence of 450 flights a year. Bhakdi said satellite manufacturing and costs are "no problem," but flight cadence and reusability are the bottleneck, estimating roughly 75 flights needed in 2029 to hit a first gigawatt by year-end under his most bullish case. Both speakers flagged an unresolved point: whether Starship's upcoming test flight (referenced as "flight 15") successfully catches and reflies both booster and ship will be decisive for how fast this timeline compresses.

Terrestrial compute bottlenecks: cooling, turbine blades, and political permitting

Bhakdi said Musk has secured chips, land and financing for terrestrial data centers, with the remaining constraints being cooling/wiring and energy turbines. He said Musk discovered turbine manufacturers could not source blades from the handful of companies that make them, so Tesla began manufacturing blades itself using its "giga-casting" technology, which Bhakdi called "genius." He said Musk has also reportedly secured "20 gigawatts worth of cooling for next year," which Bhakdi read as a tell of where the real bottleneck lies.

On political resistance, Bhakdi said Musk claims to have "a magic hack" for permitting that he won't disclose, and Bhakdi said he does not fully understand it but finds Musk's confidence convincing. Both speakers argued red-state locations with cheaper power and friendlier permitting (citing Louisiana and Tennessee as examples) give Musk room to maneuver, though Bhakdi warned that "every time you hack something like that, the door closes," meaning terrestrial buildout gets politically harder each year — one reason he thinks the move to orbital compute is "perfectly timed."

Disagreement over how to value a gigawatt of compute versus a Cybercab

The host and Bhakdi disagreed on the revenue comparison between a gigawatt of AI compute and Tesla's Cybercab fleet. The host's model put revenue at roughly $11 billion per gigawatt after depreciation, while Bhakdi called that figure "totally off," arguing it should be closer to $40 billion, pointing to Tesla's disclosed 450-megawatt deal generating a $26 billion run rate. On Cybercabs, the host cited paid miles around "250 to 300... per day" at roughly $250 a mile, while Bhakdi said if those figures hold, the equivalent valuation would be "significantly higher" than the host's $30,000 gross-profit-per-cab estimate. They converged roughly around 300,000-500,000 Cybercabs being equivalent to one gigawatt of terrestrial compute, though both called their own numbers rough estimates pending further modeling.

"I think compute gets a similar amount... because compute the operating expenses are not that high." — Jo Bhakdi

Supply, not demand, is the constraint — and the Neoclouds are losing

Bhakdi argued the market misunderstands the AI compute story: demand is effectively unlimited, and the real bottleneck is build-out speed. He said Neocloud data-center build times have stretched from two and a half years to four years, while Musk's teams can build equivalent capacity in "12 weeks," a gap he called a severe competitive disadvantage for companies like Coreweave, which he said pay roughly $5 billion a year in interest costs on delayed builds. He cited a claimed rumor that Musk has locked up 40% of Nvidia's Vera Rubin chip supply.

The host added that an "agentic explosion" could drive a tenfold increase in CPU demand (benefiting companies like AMD), and raised an unverified claim that Tesla's Robotaxi/Cybercab fleet is now logging more paid miles in Austin than Waymo, citing "3.3 million miles, zero accidents" as a safety data point from the show.

Tesla's Megapack business and compute-energy convergence

The host said Tesla's Megapack battery storage business, which he had previously projected could be worth $1.1 trillion in market cap, now has disclosed figures of 130-140 gigawatts of capacity, roughly $33 billion in revenue and an estimated $11 billion in profit at a 33% margin. He argued combining Megapack storage with inference compute and Starlink connectivity could let Tesla scale compute capacity faster than traditional hyperscalers. Bhakdi said he does not know exactly how much "free energy" in deployed batteries could be repurposed for compute but noted Musk has hinted at "distributed compute" plans, speculating this could be an undisclosed element of Musk's permitting strategy.

Both cited a claim attributed to Musk that SpaceX/Tesla combined revenue could exceed "100 trillion in less than 10 years" absent a major world war, alongside a separate claim of "half a trillion to a trillion dollars" in compute-related revenue potential, while cautioning these figures are speculative and unconfirmed by specific disclosed numbers.

Politics as a market risk: midterms, AI backlash, and Silicon Valley's realignment

On the US midterms, the host cited historical data showing markets up "8%" three months after election day, "14%" at six months, and "20 to 32%" at twelve months across prior cycles, arguing this undercuts fears of a selloff regardless of outcome. Bhakdi said Wall Street prefers political gridlock over either party holding full control, and argued a "blue, red, red" outcome (Democratic House, Republican Senate and presidency) would be the best scenario for markets.

Bhakdi argued the deeper political risk is a growing Democratic Party line against "tech billionaires" and AI buildout specifically, which he said is pushing previously politically-mixed Silicon Valley figures into a more unified opposition front. He raised a Snow Crash-style scenario in which concentrated corporate power, held by figures like Musk, begins to rival or exceed state power, citing SpaceX's role in geopolitical infrastructure as an example, though he offered this as speculative framing rather than a firm prediction.