Elon Musk Spent $366M On The Midterms — Will It Work?

OpenAI

Across the shows: 28 speakers in other episodes, 12 bullish and 27 bearish calls, past 30 days

• Bradley Tusk · Tusk VenturesBearish

  OpenAI's campaign against Alex Bores backfired and raised his profile, so AI firms risk political collateral damage.

  OpenAI came out aggressively against Alex Bores, who passed frontier model regulation in Albany and was a leading candidate in the high-profile Manhattan congressional primary. Voters who heard that Sam Altman opposed Bores liked Bores more, which propelled his campaign. Bores still lost, but Tusk argues the effort helped him anyway. He says companies in social media, data centers, and frontier models and hyperscalers must be very careful because of this kind of collateral damage.

  “Sam Altman inadvertently in trying to kill Boris ended up helping them. So, yes, I do think these companies have to be very careful”at 12:43

• Gil Luria · D.A. DavidsonUncertain

  OpenAI is gross-margin profitable on compute but probably not overall profitable; full profitability is speculative.

  Luria walks through the numbers. On about $70B of revenue, including a $20B hyperscaler cut, OpenAI keeps about $50B, and about $30B of gross margin after compute costs. Its training costs are likely above $30B, so it is probably not profitable overall. Training costs are not ramping faster than revenue, and very large models limit how much more can be spent. He estimates net margins near 80% and gross margins near 60%, which he calls healthy for a company that did not exist a few years ago.

  “So OpenAI may not be profitable, it's probably not profitable right now on an overall basis, but it is profitable on”at 25:35

Anthropic

Across the shows: 28 speakers in other episodes, 11 bullish and 32 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBullish

  Anthropic's revenue is growing fast and incrementally profitable, and it may already be cash-flow positive.

  Luria expects Anthropic's revenue run rate, combined with OpenAI's, to be well above $100B, up from about $10B a year ago and zero two years ago. He estimates gross margins of 60-80%, depending on gross or net revenue, which is healthy. Anthropic may be cash-flow positive because its growth has been so fast it could not spend fast enough. On an incremental basis its revenue is quite profitable and still growing fast. He is in year one of a long investment cycle, and he thinks AI spending will likely produce positive returns over the next twelve months.

  “Anthropic may actually be cash flow positive because their growth has been so fast they couldn't spend fast enough. ... their revenue is actually quite profitable and is still growing very fast.”at 22:05

• Gil Luria · D.A. DavidsonUncertain

  Anthropic's public listing will be highly speculative, since long-run profitability including training and compute costs is unproven.

  Luria says it is completely speculative whether Anthropic and OpenAI will ever become profitable on an overall basis, including payments to hyperscalers, compute, and training. That is why he expects Anthropic's IPO to be highly speculative. He notes much current AI spending by consumers and companies goes to these two firms, which will not be the case forever. He also says the application layer, other models, and open-weight models will take a growing share of the market.

  “That's why as Anthropic goes public, it's going be highly speculative.”at 27:13

Broadcom

Across the shows: 2 speakers in other episodes, 2 bullish and 1 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBullish

  Broadcom is priced as if the AI cycle ends, yet it should capture most chip profit if the cycle runs to 2030.

  Luria says the market treats Broadcom, along with SK Hynix, Micron and Nvidia, as if the AI cycle is already over, while much of the semi complex trades at 30-50x. If the cycle continues through 2030, these four will take most of the profit. He likes Broadcom, though only to some extent.

  “Because if the cycle continues through 2030, those four companies will grab most of the profit.”at 31:22

SK Hynix

Across the shows: 1 speaker in other episodes, 1 bullish and 0 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBullish

  SK Hynix is priced as if the AI cycle ends, yet it should capture most chip profit if the cycle runs to 2030.

  Luria says the market treats SK Hynix, Micron, Nvidia and Broadcom as if the AI cycle is already over, while much of the semi complex trades at 30-50x. If the cycle continues through 2030, these four will grab most of the profit, so the market's valuation makes no sense. He lists SK Hynix among the chip names he likes.

  “Because if the cycle continues through 2030, those four companies will grab most of the profit.”at 31:22

NVIDIA

Across the shows: 29 speakers in other episodes, 23 bullish and 9 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBullish

  Nvidia is priced as if the AI cycle ends, yet it should capture most chip profit if the cycle runs to 2030.

  Luria says the market is treating Nvidia, Micron, SK Hynix and Broadcom as if the AI cycle is already over, trading at low multiples while much of the semi complex trades at 30-50x. He argues this makes no sense. If the cycle continues through 2030, these four will grab most of the profit. He lists Nvidia among the chip names he likes.

  “Because if the cycle continues through 2030, those four companies will grab most of the profit.”at 31:22

Micron Technology

Across the shows: 10 speakers in other episodes, 8 bullish and 2 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBullish

  Micron is priced as if the AI cycle ends, yet it should capture most chip profit if the cycle runs to 2030.

  Luria says the market is treating four chip companies, SK Hynix, Micron, Nvidia and Broadcom, as if the cycle is already over, trading at low multiples (about 18x and 4-5x), while much of the semi complex trades at 30-50x. He thinks this is inconsistent: if the cycle continues through 2030, these four will capture most of the profit, not just grow. Part of the gap reflects trading and the law of large numbers. He names Micron among the chip names he likes most.

  “Because if the cycle continues through 2030, those four companies will grab most of the profit.”at 31:22

CoreWeave

Across the shows: 3 speakers in other episodes, 1 bullish and 2 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBearish

  CoreWeave's equity goes to zero if the AI cycle turns; borrowing at 9% for about 1% returns destroys value.

  Luria calls CoreWeave the poster child for bad behaviour: borrowing at very high rates, about 9%, to fund a speculative asset that currently earns about 1% returns, with guidance of only 4-5% at best. He compares it to an individual investor taking a margin loan at 9% to buy Treasuries yielding 5%, which is value destruction. If the AI cycle sputters or turns, CoreWeave will be the first to go, its equity worth zero, and it may not be able to make its interest payments.

  “[CoreWeave] is gonna be the worst, the first to go. The equity value is gonna be zero.”at 32:30

Palantir Technologies

Across the shows: 2 speakers in other episodes, 2 bullish and 0 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBullish

  Palantir is a software winner from AI tools improving, adding value and accelerating growth, with much bigger upside ahead.

  Luria names Palantir in a group of software companies doing phenomenally well. They are all benefiting from AI tools getting better, adding more and more value, and have accelerating growth. They performed well recently, and he thinks they will be much bigger companies down the road.

  “You've got Palantir, Snowflake, Datadog, Shopify, CrowdStrike, ... extraordinarily good companies that are all benefiting from AI tools getting better.”at 33:42

Snowflake

• Gil Luria · D.A. DavidsonBullish

  Snowflake is a software winner from AI tools improving, adding value and accelerating growth, with much bigger upside ahead.

  Luria names Snowflake in a group of software companies doing phenomenally well. They are all benefiting from AI tools getting better, adding more and more value, and have accelerating growth. They performed well recently, and he thinks they will be much bigger companies down the road.

  “You've got Palantir, Snowflake, Datadog, Shopify, CrowdStrike, ... extraordinarily good companies that are all benefiting from AI tools getting better.”at 33:42

Datadog

• Gil Luria · D.A. DavidsonBullish

  Datadog is a software winner from AI tools improving, adding value and accelerating growth, with much bigger upside ahead.

  Luria names Datadog in a group of software companies doing phenomenally well. They are all benefiting from AI tools getting better, adding more and more value, and have accelerating growth. They performed well recently, and he thinks they will be much bigger companies down the road.

  “You've got Palantir, Snowflake, Datadog, Shopify, CrowdStrike, ... extraordinarily good companies that are all benefiting from AI tools getting better.”at 33:42

CrowdStrike Holdings

• Gil Luria · D.A. DavidsonBullish

  CrowdStrike is a software winner from AI tools improving, adding value and accelerating growth, with much bigger upside ahead.

  Luria names CrowdStrike in a group of software companies doing phenomenally well. They are all benefiting from AI tools getting better, adding more and more value, and have accelerating growth. They performed well recently, and he thinks they will be much bigger companies down the road.

  “You've got Palantir, Snowflake, Datadog, Shopify, CrowdStrike, ... extraordinarily good companies that are all benefiting from AI tools getting better.”at 33:42

Shopify

Across the shows: 1 speaker in other episodes, 1 bullish and 0 bearish calls, past 30 days

• Gil Luria · D.A. DavidsonBullish

  Shopify is a software winner from AI tools improving, adding value and accelerating growth, with much bigger upside ahead.

  Luria names Shopify in a group of software companies doing phenomenally well. They are all benefiting from AI tools getting better, adding more and more value, and have accelerating growth. They performed well recently, and he thinks they will be much bigger companies down the road.

  “You've got Palantir, Snowflake, Datadog, Shopify, CrowdStrike, ... extraordinarily good companies that are all benefiting from AI tools getting better.”at 33:42

US elections & politics

• Bradley Tusk · Tusk Ventures

  Musk's midterm spending will largely be wasted; its real return is favour-owing and deterrence, not flipping the House.

  With Republicans polling poorly, Tusk expects most of Musk's roughly $366M midterm spending to be wasted. He sees Musk's return in two things. First, candidates he helps elect owe him and can be called on for help with SpaceX, Tesla or The Boring Company. Second, being seen as the biggest spender creates fear and deterrence among politicians who think fundraising matters. He adds that Musk and Jared Kushner will likely be frequent congressional witnesses, since private citizens cannot fight subpoenas as government officials can.

  “I think it largely will be for waste. But I think the ROI for him comes in two things.”at 6:55

• Bradley Tusk · Tusk Ventures

  Musk's money won't stop Democrats flipping the House, and even Democratic wins would pass little legislation.

  Tusk says Musk's money will not change the fact that the House is going to flip in the midterms, a few weeks away. He also argues that legislation will not move even with a Democratic House and Senate, because the Senate needs 60 votes for cloture to beat a filibuster and Democrats will not reach that. Confirmations of judges may move at the margins.

  “His money's not gonna change the fact that the house is going to flip.”at 7:34

• Bradley Tusk · Tusk Ventures

  Campaign money has sharply diminishing returns; heavy spending like Musk's has limited impact on election outcomes.

  Most campaign money goes to broadcast TV, which few people watch now as viewing moves to streaming, and direct mail is discarded unread. Recent winners such as Trump and Mamdani won on energy, social media and fit with the moment, not money. Mamdani beat Cuomo, who spent about $70M against Mamdani's $17M, and Trump beat Harris and Clinton despite being outspent. Tusk argues money has an impact until a point, after which returns diminish very fast, and that billionaires over-index on money because it is their main resource.

  “It's the other way, actually. I think that it has an impact until a certain point, and then the returns become diminishing very fast.”at 8:28

• Bradley Tusk · Tusk Ventures

  Cheap social media ads now beat expensive broadcast TV at reaching voters, so broadcast spending is much less impactful.

  Technology has changed how voters are reached. Free or cheap channels like social media are now very impactful, while expensive, outdated broadcast television is much less so. Tusk's firm runs the clean energy super PAC Investment Tomorrow, which ran ads only on Rumble and Truth Social, not broadcast, against Chip Roy, not nice enough to Trump. Roy lost the Texas attorney general race by 11 points. The PAC has spent about $6M, and three incumbent members of Congress have already lost.

  “the very nature of technology has changed the way that you reach voters, which makes things that are free or very cheap, like social media, really impactful and makes things that are highly expensive and outdated, like broadcast television, much less impactful.”at 10:27

AI chips & semiconductors

• Gil Luria · D.A. Davidson

  AI value-chain companies report strong results, raised guidance and record visibility; the sell-off doubts durability, not current demand.

  Luria says every company in the AI value chain, from chip equipment makers to chipmakers, hyperscalers and model companies, reported very good results over recent weeks. They say market conditions are better than three months ago, outlook is improving, guidance is being raised, and visibility is better than ever. Investors are jittery because they do not believe the boom can last. Luria thinks companies need to explain better why they see such good visibility, but the reported numbers do not show weakness.

  “All we've heard over the last several weeks, I wanna just be clear about that, from all the companies throughout the AI value chain is not only are they reporting very good results”at 19:28

• Gil Luria · D.A. Davidson

  Chip demand holds whichever AI model is used, since open-weight and application-layer spending still needs GPUs, CPUs and memory.

  Luria argues that AI spending will broaden beyond OpenAI and Anthropic into an application layer built on top of them, mostly through hyperscalers, and into cheaper open-weight and open-source models. Revenue from these sources will keep funding investment even if it is not the two leading labs. Whether a model runs on a data center, on premise, or on a device, it still uses GPUs, CPUs and memory, so it does not matter which model it is for chip demand.

  “Because when we use an open weight model, open source model, we're still using GPUs, CPUs, and memory. It doesn't matter.”at 29:04

AI regulation & safety

• Bradley Tusk · Tusk Ventures

  Midterms won't change tech policy much; states will regulate AI and data centers bipartisanly whatever the outcome.

  Trump is strongly anti-regulation and would veto federal tech rules that Congress tries to pass, and Congress would not override him. Action will happen in the states, where Republican and Democratic governors alike are going after AI and data centers on a bipartisan basis. Tusk therefore expects relatively uniform tech outcomes whoever wins or loses in the midterms.

  “And I would argue that there's actually not gonna be much impact on tech in the midterms simply because Trump is so anti regulation.”at 11:22