Inevitability Weekly #5
Happy New Year! We’re back with another edition of Inevitability Weekly
Gavin Baker Invest Like The Best Interview
I do believe at some level investing is the search for truth. And if you find truth first, and you're right about it being a truth, that's how you generate alpha, and it has to be a truth that other people have not yet seen, you're searching for hidden truths.
- Gavin Baker
A wide ranging interview covering the economics of AI, discussions of potential bear cases, the impact to adjacent tech industries and lots more. I’m always interested in hearing what Gavin has to say as he brings a really unique perspective to investing in tech across public and private markets.
On Google’s TPU Economics
A couple of things. So, one, for whatever reason, Google made more conservative design decisions. And part of that is, so, Google, let’s say, the TPU ... So, there’s front end and back end of semiconductor design, and then there’s dealing with Taiwan Semi.
And you can make an ASIC in a lot of ways. What Google does is they do mostly the front end for the TPU, and then Broadcom does the backend, and manages Taiwan Semi and everything. It’s a crude analogy, but the front end is, like, the architect of a house. They design a house. The back end is the person who builds the house. And then managing Taiwan Semi is stamping out that house like Lennar or D.R. Horton.
And for doing those two latter parts, Broadcom runs a 50% to 55% gross margin. We don’t know what on TPUs. Let’s say in 2027, TPU I think consensus estimates may be somewhere around $30 billion. Again, who knows? $30 billion I think is a reasonable estimate. 50% to 55% gross margins. So, Google is paying Broadcom $15 billion. That’s a lot of money.
At a certain point, it makes sense to bring a semiconductor program entirely in-house. So, in other words, Apple does not have an ASIC partner for their chips. They do the front end themselves, the back end, and they manage Taiwan Semi. And the reason is they don’t want to pay that 50% margin.
So, at a certain point, it becomes rational to re-negotiate this, and just as perspective, the entire opex of Broadcom’s semiconductor division is, round numbers, $5 billion. So, it would be economically rational now that Google is paying ... If it’s $30 billion, we’re paying them $15 billion. Google can go to every person who works in Broadcom semi, double their comp, and make an extra $5 billion. In 2028, let’s just say it does $50 billion. Now it’s $25 billion. You get triple their comp. And, by the way, you don’t need them all.
On Potential Bear Cases for AI
There’s one really obvious bear case and it is just edge AI, and it’s connected to the economic returns to ASI. In three years, on a bigger and bulkier phone to fit the amount of DRAM necessary, and the battery won’t probably last as long, you will be able to probably run a pruned-down version of something like Gemini 5 or Grok 4, Grok 4.1 or ChatGPT at 30, 60 tokens per second, and that’s free. This is clearly Apple’s strategy. It’s just, “We’re going to be a distributor of AI, and we’re going to make it privacy-safe and run on the phone. Then you can call one of the big models, the God models in the cloud, whenever you have a question.” If that happens, if 30, 60 tokens a second at a 115 IQ is good enough, I think that’s a bear case.
Other than just these scaling laws break. In terms of if we assume scaling laws continue, and we now know they’re going to continue for pre-training for at least one more generation ... and we’re very early in the two new scaling laws for post-training, mid-training, RLVR, whatever people want to call it, and then test-time computed inference ... we’re so early in those, and we’re getting so much better at helping the models hold more and more context in their minds as they do this test-time compute.That’s really powerful, because everybody’s like, “Well, how’s the model going to know this?” Well, eventually, if you can hold enough context, you can just hold every Slack message and Outlook message and company manual in a company, in your context, and then you can compute the new task and compare it with your knowledge of the world, what you think, what the model thinks, all this context. It may be that just really, really long context windows are the solution to a lot of the current limitations. That’s enabled by all these cool tricks like KV cache offload and stuff. I do think, other than scaling laws slowing down, other than there being low economic returns to ASI, edge AI is to me by far the most plausible and scariest bear case.
On Power Being A Bottleneck
Having watts as a constraint is really good for the most advanced compute players, because if watts are the constraint, the price you pay for compute is irrelevant. The TCO of your compute is absolutely irrelevant because if you could get 3X or 4X or 5X more tokens per watt, that is literally three or 4X or 5X more revenue.
If you’re going to build an advanced data center costs 50 billion, a data center with the ASIC maybe costs 35 billion. If that $50 billion data center pumps out 25 billion of revenue and your ASIC data center at 35 billion is only pumping out eight billion, well, you’re pretty bumped. So I do think it’s good for all of the most advanced technologies in the data center, which is exciting to me as an investor.
So as long as power is a governor, the best products are going to win irrespective of price and have crazy pricing power. That’s the first implication that’s really important to me. Second, it is in the only solutions to this, we just can’t build nuclear fast enough in America. As much as we would love to build nuclear quickly, we just can’t. It’s just too hard. NEPA, all these rules, it’s too hard. Like a rare ant that we can move and it could be in a better environment, can totally delay the construction of a nuclear power plant, one ant. That is America today.
Humans need to come first. We need to have a human-centric view of the world. But the solutions are natural gas and solar. And the great thing about these AI data centers is apart from the ones that you’re going to do inference on, you can locate them anywhere.
So I think you were going to see, and this is why you’re seeing all this activity in Abilene because it’s in the middle of a big natural gas basin and we have a lot of natural gas in America because of fracking. We’re going to have a lot of natural gas for a long time. We can ramp production really fast. So I think this is going to be solved. You’re going to have power plants fed by gas or solar. And I think that’s the solution. And you’re already, all these turbine manufacturers were reluctant to expand capacity, but Caterpillar just said, “We’re going to increase capacity by 75% over the next few years.” So the system on the power side is beginning to respond.
Gavin also shared his thoughts on the Nvidia/Groq deal.
The Prison of Financial Mediocrity
My take is that social media and our higher order needs have conditioned people that are positioned well below the financial upper class to feel like they are already at a loss. The zeroth line has been repositioned. It is why you see unironic takes about the poverty line being at $150k. This generation isn't gambling to survive, they're gambling to actually have a life.
A thought provoking piece on the current financial dynamics the young generations are facing. This piece succinctly captured what seems to be a super pervasive sentiment amongst millennials and gen z: anxiety about the level of uncertainty for how AI will shape the future combined with a widening wealth gap that leaves many feeling they can’t attain their “dream life” (which may be skewed because of social media) without taking on greater financial risks.
On the breaking of the old “deal” of work
The implicit deal used to be simple: show up, work hard, stay loyal, and you’ll be rewarded. Companies offered pensions. Tenure meant something. Your house appreciated while you slept. The system worked if you trusted it.
That deal is dead.
Staying at one company for 20 years is now a career liability, not an asset. Wages grew 8% while housing costs doubled and debt payments for young people increased ~33%. The math doesn’t support patience anymore.
Looking at the bigger picture, I used to think it was bad, but with the advent of AI and the economic impact they are going to have (even with current technologies), I think it is only going to get significantly worst.
On AI and Social Media Creating Anxiety
Previous generations had limited visibility into how others lived. You compared yourself to your neighbors, your coworkers, maybe some celebrities in magazines. The reference class was narrow. Now the reference class is infinite. A 25-year-old making $70k is constantly fed content from people their age making $2mn, living in Bali, “working” four hours a day. The baseline for “enough” keeps moving.
You never catch up. No matter what you achieve, social media will show you what you’re missing. The spread between your life and the life you “should” have is maintained algorithmically, forever uncollapsible.
So you have AI shrinking your timeline AND social media ensuring you never feel like you’ve arrived. The pressure to escape, NOW, FAST, before it’s too late, compounds daily.
The anxiety is pervasive. Every white-collar worker has done the mental exercise: “Could AI do my job? When?” And most of them don’t love their answers. Even if they think they’re safe for now, “for now” keeps getting shorter.
The Numbers and the Trades
Prediction Markets: Polymarket and Kalshi did $10B+ in volume in November 2025 alone. Combined annual volume is approaching $40B. In 2020 this was essentially zero. The growth rate is vertical.
Sports Betting: Legal sports betting revenue went from $248M in 2017 to $13.7B in 2024. Gen Z and Millennials account for 76% of betting activity. Activity on online sportsbooks rose 7% year-over-year for both cohorts.
TransUnion’s report identified these gamblers as “speculators”: urban renters, heavy users of crypto apps, concentrated in mobile trading platforms. Young people locked out of traditional wealth-building, seeking asymmetric payoffs in the only markets that offer them.
If this diagnosis is right, that a generation of economically locked-out young people will continue seeking agency through high-variance financial products, then you want to be long anything that serves that demand.
The platforms win regardless of whether the users win. You are looking for platforms that doesn’t care if you win your bet or if your prediction is right. You are looking for businesses that extract fees from activity, and activity is growing.
Dan Wang 2025 Letter
https://danwang.co/2025-letter/
A very thoughtful and wide ranging piece reflecting on the year, touching on topics such as culture in Silicon Valley, the future of AI, the arms race between The US and China and much more.
On AI Future and SV Culture
While critics of AI cite the spread of slop and rising power bills, AI’s architects are more focused on its potential to produce surging job losses. Anthropic chief Dario Amodei takes pains to point out that AI could push the unemployment rate to 20 percent by eviscerating white-collar work.
I wonder whether this message is helping to endear his product to the public.
The most-read essay from Silicon Valley this year was AI 2027. The five authors, who come from the AI safety world, outline a scenario in which superintelligence wakes up in 2027; a decade later, it decides to annihilate humanity with biological weapons. My favorite detail in the report is that humanity would persist in a genetically modified form, after the AI reconstructs creatures that are “to humans what corgis are to wolves.” It’s hard to know what to make of this document, because the authors keep tucking important context into footnotes, repeatedly saying they do not endorse a prediction. Six months after publication, they stated that their timelines were lengthening, but even at the start their median forecast for the arrival of superintelligence was later than 2027. Why they put that year in their title remains beyond me.
It’s easy for conversations in San Francisco to collapse into AI. At a party, someone told me that we no longer have to worry about the future of manufacturing. Why not? “Because AI will solve it for us.” At another, I heard someone say the same thing about climate change. One of the questions I receive most frequently anywhere is when Beijing intends to seize Taiwan. But only in San Francisco do people insist that Beijing wants Taiwan for its production of AI chips. In vain do I protest that there are historical and geopolitical reasons motivating the desire, that chip fabs cannot be violently seized, and anyway that Beijing has coveted Taiwan for approximately seven decades before people were talking about AI.
Silicon Valley’s views on AI made more sense to me after I learned the term “decisive strategic advantage.” It was first used by Nick Bostrom’s 2014 book Superintelligence, which defined it as a technology sufficient to achieve “complete world domination.” How might anyone gain a DSA? A superintelligence might develop cyber advantages that cripple the adversary’s command-and-control capabilities. Or the superintelligence could self-recursively improve such that the lab or state that controls it gains an insurmountable scientific advantage. Once an AI reaches a certain capability threshold, it might need only weeks or hours to evolve into a superintelligence.
And if an American lab builds it, it might help to lock in the dominance of another American century.
On China’s Market Competition
Probably the most underrated part of the Chinese system is the ferocity of market competition. It’s excusable not to see that, given that the party espouses so much Marxism. I would argue that China embodies both greater capitalist competition and greater capitalist excess than America does today. Part of the reason that China’s stock market trends sideways is that everyone’s profits are competed away. Big Tech might enjoy the monopolistic success smiled upon by Peter Thiel, coming almost to genteel agreements not to tread too hard upon each other’s business lines. Chinese firms have to fight it out in a rough-and-tumble environment, expanding all the time into each other’s core businesses, taking Jeff “your margin is my opportunity” Bezos with seriousness.
Third, western elites keep holding on to a distinction between “innovation,” which is mostly the remit of the west, and “scaling,” which they accept that China can do. I want to dissolve that distinction. Chinese workers innovate every day on the factory floor. By being the site of production, they have a keen sense of how to make technical improvements all the time. American scientists may be world leaders in dreaming up new ideas. But American manufacturers have been poor at building industries around these ideas. The history books point out that Bell Labs invented the first solar cell in 1957; today, the lab no longer exists while the solar industry moved to Germany and then to China. While Chinese universities have grown more capable at producing new ideas, it’s not clear that the American manufacturing base has grown stronger at commercializing new inventions.
On Geopolitics and Potential for Improvements
The United States isn’t so good at celebrating its history either. 2026 is the 250th anniversary of the country’s founding. Where are the monuments to exalt that history? Most of the planned celebrations look small bore. Why hasn’t the federal government built a technological specimen as sublime as the Golden Gate Bridge, the Hoover Dam, or the Apollo missions? Probably because planning for any project should have commenced 10, 20, or 30 years ago. No president would have gotten around to starting a project that has no chance of being completed in his term. Lack of action due to the expectation of long timelines is one of the sins of the lawyerly society.
But American problems seem more fixable to me than Chinese problems. That’s why I live here in the US. I made clear in my book that I am drawn to pluralism as well as a broader conception of human flourishing than one that could be delivered by the Communist Party. The United States still draws many of the most ambitious people in the world, few of whom want to move to China. Even now a significant number of Chinese would jump to emigrate to the US if they felt they could be welcomed. But this enduring American advantage should not excuse the US from patching up its deficiencies.
I learned recently that the Bay Area has 26 separate transit agencies; is it really a triumph of democracy to have so many unconsolidated efforts? I wonder whether we can accuse the California government of subverting the will of the people by making so little progress on its high-speed rail, which was approved by referendum in 2008; California rail authorities take more pride in creating jobs than doing the job. I am tempted to use the language from American foreign policy at home. Why talk about American credibility only in terms of combat? Why shouldn’t the failure to deliver on big projects, after spending so much money, constitute a more severe blow to the credibility of the American project? Is the state of the US defense industrial base really deterring adversaries?
I won’t belabor issues with American public works or manufacturing. I’ll suggest only that the US ought to be acting with greater curiosity on how to do better. It doesn’t have to become China; but it should better study China’s successes. There is a 21st century playbook for becoming an industrial power and China has written it. This playbook consists of infrastructure development, solicitation of foreign investment, industrial subsidies, and the creation of industrial ecosystems. I hope that the US will stop attributing all of China’s successes to stealing. If such a program would be sufficient for building a world-class industry, then American spooks should dedicate their formidable capabilities to extracting Chinese industrial secrets. The reality is that there is little to be learned from blueprints. By failing to recognize China’s real strengths — the industrial ecosystems pulsating with process knowledge — the US is only cheating itself.
Roundup
AI’s trillion-dollar opportunity: Context graphs
The Bitter Lessons: Thoughts on US-China Competition
Modeling for Fundamental L/S Equity
Thanks for reading! Be back same time next week and please never hesitate to send me interesting things to read! My dms are open.



Good read, especially wrt The Prison of Financial Mediocrity
Legend