Fed Hesitates on Tariffs, The New Mag 7, Death of VC, Google's Value in a Post-Search World
Summary
- Philippe Laffont’s macro read is that the Fed’s hold at 4.25%-4.5% is not inherently bearish, because a cut might mean conditions are deteriorating while no cut can mean the economy remains strong. Consumer sentiment looks terrible, yet Visa, Mastercard, and company commentary show resilient spending even after adjusting for tariff front-loading. He called the drawdown a “tariff correction or a tariff tantrum, but not a tariff crisis.”
- The monetary-policy split is whether liquidity warnings justify pre-emption or the tariff regime demands patience. Chamath Palihapitiya said subprime indicators are “blinking yellow” and argued that Powell’s repeated language about waiting—22 instances or synonyms—supports a hypothesis of political motivation. David Friedberg countered that March CPI was still 2.4% against a 2% target, while a potentially durable 10% tariff regime, possible tax cuts, and new trade-driven orders could lift GDP, employment, and inflation.
- Laffont’s decisive tech call was “tokens are much greater than tariffs.” Microsoft reported processing 100 trillion AI tokens in Q1, including 50 trillion in March, while Laffont heard of chip and compute shortages across public and private companies. His thesis is that tariffs may eventually be offset by deals, deregulation, and tax breaks, leaving AI token demand as “maybe the most exciting trend” he has seen in 35 years.
- Eddy Cue’s claim that Apple’s search volume fell for the first time in 20 years turns every basis point of Google share shift into a modelable profit risk, but the panel disagreed sharply over how fast Google must cannibalize itself. David Sacks favored measured migration because search advertising generates roughly $200 billion and AI queries cost an order of magnitude more to serve; Chamath urged Google to assume share could fall from 99% to 75% within two years and make Gemini its front door. His warning: spending $75 billion annually on models while “stagegating the product” is the worst middle course.
- The Mag 7 trade is fragmenting into a new public-private basket of perhaps 25 companies. Laffont compared Google’s roughly $1.8 trillion valuation with OpenAI’s assumed $300 billion and asked whether Google should become “the next IBM” or re-engineer itself; meanwhile, companies such as SpaceX and Stripe cannot rationally be excluded merely because they remain private. The broader opportunity also includes established non-tech businesses that use AI to disrupt mature markets.
- The “death of VC” argument is a liquidity argument: without IPOs and acquisitions, venture cannot generate the mid-to-high-20% net return that long lockups require. Chamath’s portfolio math starts with a 10% after-tax target, versus 4%-5% short-term paper, 12%-13% hedge funds, and mid-teens private credit or equity; regulatory strangulation may be costing venture 500-1,000 basis points. Laffont’s sharper point was that capping blockbuster outcomes removes the lottery ticket that finances all the failures.
- Laffont’s proposed answer is a near-permanent interval fund spanning public equities, private companies, and cash, built to identify the next Mag 7 over ten years. It charges 1.25% and a 12.5% incentive fee, has an indicated minimum near $50,000, and began with a combined $1 billion from the Bezos and Dell family offices. The discipline is to wait until there is roughly a 75% probability a company is the category leader, then value it against public comparables rather than assuming every private round should simply double.
Deep dive
1. The Fed hold looks stronger than consumer sentiment
The Fed kept rates at 4.25%-4.5% after cutting 50 basis points in September and 25 in both November and December. Its statement said activity continued at a solid pace while acknowledging the paired risks of higher unemployment and higher inflation.
Laffont challenged the reflex that a cut must be good news: “What if the Fed is cutting because things are not so great?” Conversely, holding rates could mean the economy is strong. At Coatue, he tracks hard news divided by sentiment; he said it was the first time he had seen the news so good while sentiment looked so bad.
The clearest divergence is the consumer. Sentiment is “very, very weak,” but spending remains remarkably resilient in Visa and Mastercard results and company transcripts—including April and the latest week after adjusting for purchases pulled forward ahead of tariffs.
Laffont treats sentiment as lagging rather than leading: markets fall, sentiment deteriorates, then recovering markets improve the next survey. He also saw two backstops—the government budged when equities broke, while the Fed promised to restore malfunctioning liquidity without simply bailing out stocks.
2. Liquidity is blinking yellow, but the panel split on politics
Chamath’s warning came from subprime lenders: the spread between Credit Acceptance and Capital One, especially unusually high price-to-book levels, has historically preceded liquidity trouble. His conclusion was narrower than a generic “Fed put”: measures of consumer credit health are “blinking yellow.”
He counted “waiting,” or synonyms for it, 22 times in the Fed release and called that “an incredible amount of verbal gymnastics.” He argued that officials are acting as much from political motivation as financial metrics, because a cut would help Trump ahead of the midterms, while acknowledging this was the explanation he could come up with for ignoring the indicators.
Jason Calacanis pressed him on whether Powell was retaliating over Trump’s firing threats. Chamath rejected that narrower formulation: his claim was that officials are choosing to ignore historically useful leading indicators, and “the only reasons that I can come up with to ignore them are political reasons.”
Friedberg’s pushback — worth keeping: March CPI was 2.4% against the Fed’s 2% target, with another report imminent, while mortgage delinquencies appeared flat, probably because so much debt was refinanced at low rates. “I think they’re going to wait for data.”
3. A durable 10% tariff makes rate cuts harder
Friedberg treated the UK agreement as the first concrete template: even one of America’s friendliest partners retained a 10% tariff on imports into the US. If that becomes the floor, less friendly or less reciprocal countries could face higher rates.
That creates a potentially sizable federal revenue stream, which could finance tax reductions while also affecting inflation, GDP, and employment. The agreement removed the UK’s 2% tax on large technology companies, and Howard Lutnick had indicated a forthcoming $10 billion Boeing order.
Calacanis said a giant US retailer told him only about 50% of tariff costs would reach prices, not 100%. He was initially surprised that a surplus country still received 10%, but interpreted the market’s positive reaction and softer China rhetoric as hope for “a bit more of a win-win.”
4. “Tokens greater than tariffs” reset the technology trade
Laffont was candid about his macro limits—“I think I’ve predicted like seven of the last three recessions”—and said services-heavy technology was relatively insulated. Semiconductors and computer assembly were not: autos already faced 25% sector tariffs, with possible pharmaceutical and semiconductor measures still unresolved.
The uncertainty helped produce a 25% peak-to-trough market decline and made Laffont temporarily conservative. What changed his mind was an emerging feedback loop: executives could argue their cases in Washington, officials could observe what broke, and policy could then be readjusted.
Microsoft’s AI disclosure was the larger pivot. It processed 100 trillion tokens in Q1, 50 trillion in March alone; Laffont saw the near-vertical curve as evidence that reasoning engines require much more compute, matching the chip and compute shortages he heard about across his private and public companies.
His framing was “tokens are much greater than tariffs.” The selloff had reflected both trade fears and doubts about AI ROI; rising capex and scarce capacity answered part of the latter. If deals, deregulation, and tax breaks eventually offset tariffs, “what are we left with? We’re left with tokens.”
5. AI can reopen every mature market
Laffont said Sergey Brin had described personal uses of AI in management decisions and offered the provocative observation that “managers are the first to go.” Laffont added that he was discussing AI first principles with his management team and planning an offsite around the subject.
The opportunity, in Jason Calacanis’s view, is selecting ordinary companies whose management teams use AI to accelerate growth and create leverage. Mature markets normally converge until competitors look alike; now “every mature market can be completely disrupted,” reopening differentiation across the economy.
6. Search erosion makes Google’s profit stream newly modelable
Apple executive Eddy Cue said Apple’s search volume fell for the first time in 20 years as users moved toward ChatGPT and Perplexity. Bloomberg’s report erased about $100 billion from Google’s market capitalization within an hour, although Google responded that overall queries—including those from Apple platforms—were still growing.
Friedberg said the “search, click, repeat paradigm is over,” but the replacement could be chat, voice, earbuds, or another interface. Google owns competitive models and users. Sacks noted that search ads produce roughly $200 billion while an AI query costs an order of magnitude more to serve.
Chamath’s concern begins when Google falls from an effective 99% share: every basis point of erosion can now be translated into economic value. He urged management to model a decline to 75% within two years and ask, “What will go wrong?” before competitors provide the answer.
Laffont framed the valuation puzzle as Google at roughly $1.8 trillion versus OpenAI at perhaps $300 billion. Search may supply about 60% of revenue but perhaps 85%, or even 110%, of profit because other divisions absorb investment.
7. Google must choose between migration and self-cannibalization
Sacks favored an incremental path: Google already has the models, application, distribution, and testing culture. It can expand AI answers, steer users toward chat, or change defaults only after determining how much behavior and monetization survive the transition.
Chamath wanted more aggression. Waiting for internal evidence ignores unknown product moves at OpenAI, Apple, and Meta—the “sword of Damocles” that can drop without warning. Reactive announcements would also damage the morale of Google’s strong engineers and product managers.
His capital-allocation test was blunt: if Google wants harvest mode, it should preserve cash rather than spend $75 billion annually building models. If it spends the money, “we’ve made the cake, let’s sell the cake”; funding the product while stagegating consumer adoption is “the worst outcome.”
Chamath favored continued investment, arguing that harvesting a cash cow and repurchasing shares “never really works.” Google’s chance to remain exceptional requires risk: “The man in the arena, he who takes the risk usually gets the spoils.”
8. Google’s underused advantage is the size of the new query bucket
Jason pointed to Gemini’s unexpected access to his Calendar, its integration with Docs, and Google products with one billion or two billion monthly users. Sacks proposed YouTube as a wedge: answer questions across transcripts, synthesize changing opinions, and generate supercuts instead of returning ten links. Jason endorsed the idea.
Friedberg reframed market share around total activity. Jason estimated that AI lets him initiate perhaps five times as many tasks because he no longer delegates research to employees or consultants; owning 80% of a market three times larger could be superior to owning 99% of old-style search.
Chamath named the failure case: Google loses most of the old bucket while capturing only 10%-20% of the new one. Jason’s countervailing possibility was that Gemini queries, Calendar activity, and YouTube viewing could produce more targeted, valuable advertising than classic search.
9. The next dominant basket will contain roughly 25 companies
Laffont rejected both a hundred-stock universe and the concentrated portfolios where five positions represent 80% of capital. His instinct is about 25 names because investing mixes skill, luck, mistakes, and surprises; he learned that humility after starting on January 1, 2000 and being “reduced to ashes.”
The old Mag 7 worked as a highly correlated basket that absorbed capital and attention. AI is breaking that correlation, just as “FAANG” eventually gave way to another label, creating room to ask which public and private businesses actually matter for the next decade.
A valid basket cannot own a merely adequate public company while excluding SpaceX or Stripe solely because they do not trade daily. Laffont’s stated job for the new vehicle is simple: “I’ve got to build for you in 10 years the new Mag 7.”
10. Missing exits are breaking venture’s incentive structure
Laffont’s explanation for delayed IPOs combined reputational and regulatory risk with increasingly sophisticated private markets. He cited—explicitly without verifying it—an anecdote that roughly 35% of the S&P had encountered a government-agency issue, while private markets increasingly resemble public markets that trade only a few times annually.
The more damaging constraint is blocked M&A. If large companies cannot acquire startups, investors lose one of their best routes to monetize risk; Laffont asked why he should fund a small private company when he can simply buy the public incumbent.
Jason proposed freer combinations below perhaps $750 billion or $1 trillion to create a “Mag 70,” while restraining dominant platforms capable of forcing free bundled products onto users. Laffont rejected market capitalization as the test: antitrust should target conduct that removes choice, not punish size or orthogonal acquisitions.
Jason conceded that smaller companies in the same arena—Coinbase, Robinhood, and E*TRADE, for example—could cause equivalent harm. The exchange narrowed the real issue from “big company bad” to whether a transaction enables monopoly pricing, dumping, or foreclosure in the affected market.
11. Capping the jackpot also eliminates the losing tickets
Laffont compared venture investing with a lottery: people tolerate frequent losses because one ticket might return $1 billion or $2 billion. Cap the prize at $30 million and participation falls, even though $30 million remains enormous. “The hope that you get OpenAI” finances the willingness to fund failure.
Chamath extended the mechanism to national competitiveness. If making money is treated as derogatory and the upside is constrained, capital retreats to simple activities; societies stagnate because fewer people fund difficult, uncertain projects. He pointed to falling investment in both China and Canada under different regimes.
Europe was his cautionary model: “too many administrators, too many hall monitors,” without enough risk capital or gigantic outcomes. His prescription was less regulation, more competition among large companies, and viable routes for small businesses to be acquired or become public.
12. Venture no longer clears the return required by its illiquidity
Exit activity spiked in 2021—examples included Rivian, Affirm, Robinhood, Duolingo, Roblox, and major acquisitions such as Afterpay and Mailchimp—then flatlined through 2022-2025. Laffont’s question was why recent issuance looked worse than otherwise ordinary periods such as 2004-2006 or 2013-2015.
Chamath’s LP framework begins with a roughly 10% annual after-tax target. Short-term securities might yield 4%-5%, hedge funds 12%-13%, and private credit or equity mid-teens with five- or six-year lockups, forcing capital farther along the risk curve.
Venture’s much longer illiquidity requires mid-to-high-20% net returns. Chamath estimated that regulatory restrictions on IPOs and acquisitions could be removing 500-1,000 basis points. Jason said that at the current pace, the asset class is hard to justify except “almost philanthropically.”
The lost return also breaks Silicon Valley’s diaspora: employees become founders, angels, mentors, and LPs after successful exits. Laffont added that American founders often leave wealth to foundations, so the same capital later funds public purposes and competes with government over what deserves support.
13. Coatue’s interval fund bridges public discipline and private upside
Laffont designed the vehicle around three freedoms: own differentiated public equities, invest in private companies, and hold substantial cash when conditions are unattractive. He rejected the indexing pressure that makes active managers closet benchmarks and keeps them fully invested even when market multiples look extreme.
His rough model was Berkshire Hathaway: a trillion-dollar company divided approximately among cash, public equities, and private holdings. Investors grant Coatue five to seven years but receive limited periodic liquidity; in exchange for near-permanent capital, Coatue charges 1.25% and a 12.5% incentive fee.
The indicated minimum was about $50,000, with broader eligibility, UBS initially distributing the product, and simpler 1099 reporting rather than numerous K-1s and capital calls. The Bezos and Dell family offices committed a combined $1 billion, and Laffont said they would also put substantial personal capital into it.
Laffont aimed for a $1.301 billion launch so he could claim the largest launch ever based on his comparison with an early Blackstone fund, though he said he did not know whether he would reach that figure. The economics are intentionally reciprocal: compounding a 12.5% incentive fee for longer may benefit Coatue more than charging 20% over a shorter-lived fund.
14. Private imagination still needs public-market valuation discipline
Jason noted that the vehicle could move throughout the capital structure, own debt, or acquire 100% of a business, but its north star is the next dominant-company basket. Laffont said venture may not be the right model for that task: the odds change dramatically as a company scales, and he prefers paying more once he can establish roughly a 75% chance that it is the leader.
He would likely want exposure to categories such as humanoids and robotaxis, but said it was still early to identify winners. Jason’s challenge was price: secondary stakes in fashionable private companies can carry $30 billion-$40 billion valuations before meaningful revenue.
Laffont criticized the private-market habit of saying, “If the last round was 100, well, this round’s 200,” without asking why. Public comparables provide revenue, profit, earnings, and multiple discipline: the test is what the company would be worth if it already traded publicly.
His ideal investor holds both faculties at once—the private-market “telescope into the future” and the public-market voice saying, “Hey, slow down, Chamath. This is like 80 times earnings.” Public positions can fall by half immediately; private marks rise repeatedly, “and then one day it just goes to zero.”