This Week in Startups
Why a Do-Nothing Congress could be great for markets | E2359
This debrief was written and voiced by AI from the episode's audio.
🎧 AI Debrief
🎙️ Full Episode
Key Moments
D.A. explains why split control of Congress is market-positive: gridlock prevents policy shocks, trade wars, and destabilizing chaos. This uncertainty reduction outweighs tech industry concerns about political opposition to data centers and AI infrastructure.
Jason identifies the real political problem: older equity owners benefit from Uber, Tesla, SpaceX, and automation, while young drivers and workers see only job threat. This wealth gap across generations, not just political ideology, is driving electoral shifts and anti-tech sentiment.
Jason argues that Marc Andreessen's detailed historical arguments about technology adoption won't move voters. Instead, when ordinary people (stay-at-home parents, teachers, small business owners) experience tangible AI benefits—cheaper services, time savings, personal tutors—sentiment will flip. Technology narrative won't fix this; tangible results will.
Sequoia is sunsetting the scouts program that Jason and Sam Altman pioneered. Reason: market structure has fundamentally changed. A seed round can now be $5M, $50M, or $500M—t-shirt sizes don't apply. Plus, best founders now don't want VC money at all. The program built for information asymmetry is irrelevant when founders are capital-efficient and controlling.
Definition and implications: founders building capital-efficient, profitable businesses with AI can raise on their own terms or not at all. 'Alocorns' with wings—they fly over funding rounds. This is a structural shift that makes mega-funds struggle to deploy capital while making small, focused funds with founder networks suddenly valuable again.
Tomasz predicts GPU scarcity and margin compression. AI model makers currently enjoy 80%+ gross margins that are unsustainable long-term. When inference capacity is rationed like bandwidth (queries per month), margins collapse. This forces a resurgence in custom application development because paying cloud rates becomes uneconomical.
D.A. highlights emerging perpetual markets and synthetic securities platforms (Hyperliquid, Cauchy) where retail investors can bet on private company valuations, milestones, and indices with high leverage, 24/7, with no intermediaries. This could disrupt IPO timelines and force price discovery in private markets before traditional exits.
Capital-efficient AI-native founders don't need big funds and prefer their own networks because they maintain control. This is a massive shift: the best founders are now financing through syndicates, secondary markets, and social capital rather than mega-funds. Small, founder-focused funds suddenly become more valuable than $10B behemoths.
Jason quantifies his own ROI: spending $40K monthly on Claude, Grok, Cursor, and other AI tools to build internal software (AngelList alternative for syndicates, accelerator tools, podcast automation) that would have cost $2M and 12 full-time employees to build traditionally. This is real value creation, but it assumes inference prices stay cheap—a bet that Tomasz says will fail by 2027.
Autonomous vehicles save lives and reduce accidents, but they threaten driver jobs. So politicians can't win the argument in favor regardless of facts. Jason sees non-zero risk of an anti-self-driving wave across the country, even though it's wrong on the merits. This is the core political problem with AI infrastructure: tangible job threat beats abstract safety benefits.
Search Episode
Summary
This Week in Startups explores the political aftermath of the 2024 election, where Democrats gained control of the House and Senate, creating a divided government. Despite overnight market fears of a correction, markets rallied—a phenomenon D.A. Wallach attributes to uncertainty reduction. The conversation pivots to how political gridlock actually benefits business by preventing chaos, while examining technology's narrative problem: AI and automation threaten jobs, making data centers and autonomous vehicles political liabilities despite their economic and safety benefits. The panel discusses how AI adoption must reach everyday consumers through tangible benefits (cheaper homes, better healthcare, personal agents) before sentiment shifts. The episode also covers venture capital structural changes at Sequoia, the rise of capital-efficient "alocorn" companies, unsustainable margin compression in AI infrastructure, GPU constraints coming in 2027, and emerging tokenized derivative markets that could disrupt traditional securities regulation entirely.
Key Takeaways
Divided government reduces uncertainty and prevents policy chaos, which is ultimately good for markets and business despite political gridlock
highAI's political vulnerability stems from job displacement fears that are difficult to counter with historical technology adoption arguments—messaging matters more than facts
highThe next wave of venture-backed companies will be capital-efficient "alocorns" (profitable, minimal funding needs) rather than traditional venture-dependent startups, fundamentally changing VC fund economics
highGPU scarcity and inference costs will likely ration AI access in 2027, shifting from unlimited consumption to rationed bandwidth models similar to internet pricing
highVenture capital funds face margin compression across the ecosystem—unsustainable 80%+ gross margins at AI labs will normalize lower as competition increases and open-source models mature
mediumTokenized derivatives and synthetic markets could circumvent traditional securities regulation, enabling retail investors to gain direct exposure to private companies without intermediaries
mediumSequoia's consolidation and complexity reduction reflects industry-wide trend toward mega-funds and away from experimental programs, concentrating power among established players
mediumNotable Quotes
“Divided government, even if it means sclerosis, even if it means that nothing gets done in DC, what it does mean is that nothing crazy is likely to happen. And that's good for you. That's good for business.”
— D.A. Wallach
Core thesis: political gridlock reduces uncertainty and prevents destabilizing policy shocks that hurt capital allocation and business planning
“The thing that's gonna turn this around as crazy as it seems is when the stay at home mom or dad, the teacher, your everyday American starts using an agent, and the agent starts making their life better...That's the kind of stuff that will get people excited about AI and about technology again.”
— Jason Calacanis
Explains why messaging about AI remains ineffective until ordinary people experience direct benefits; current political opposition stems from abstract job fears rather than concrete gains
“Intelligence wants to be free. And hardware can't be free. And so, you know, I don't know how long this takes, but, ultimately, you know, you're gonna have to have hardware companies that earn a big enough profit to attract the people who know how to do these crazy complicated things.”
— D.A. Wallach
Suggests sustainable margins must remain high enough to incentivize innovation in chipmaking and infrastructure, contrasting with deflationary AI model pricing
“It's like we will start to buy AI the way that we buy Internet. When you buy Internet, you buy a certain amount of bandwidth...We are going to start to buy AI in the same way, which is, like, how many effectively questions can I ask, and how quick is the response? It's gonna be rationed.”
— Tomasz Tunguz
Predicts 2027 GPU constraints will force consumption-based pricing models rather than current unlimited access, fundamentally changing AI economics
Predictions
GPU availability will become constrained by 2027, forcing AI access to be rationed and priced like bandwidth rather than offered as unlimited consumption
— Tomasz Tunguz · Confidence: high · 2027
Local inference will become dominant as cloud GPU costs rise; most non-coding work (90% per Stanford research) will execute on local models with cloud spikes for heavy compute
— Tomasz Tunguz · Confidence: high · 2027
AI margins will compress significantly as open-source models undercut proprietary pricing and competition intensifies, though not to zero due to training costs requiring amortization
— Multiple speakers · Confidence: high · 2-3 years
Mainstream consumer adoption of AI agents will drive positive political sentiment reversal as people experience direct life benefits (lower costs, time savings, better decisions)
— Jason Calacanis · Confidence: medium · 12-24 months
Tokenized derivatives and perpetual markets will bypass traditional securities regulation, enabling retail investors direct access to private company exposure without intermediaries
— D.A. Wallach · Confidence: medium · 2-5 years
Alocorn companies (profitable, minimal funding needs) will become the dominant venture-backed model, leaving traditional mega-funds unable to capture early stage opportunities
— D.A. Wallach · Confidence: medium · 3-5 years
Action Items
If running a venture-backed company, evaluate whether you actually need outside capital or should pursue alocorn path (capital-efficient profitability) to maintain founder control
Divided government and uncertain policy make self-funded growth increasingly attractive versus raising large rounds; capital-efficient businesses have better exit timing control
For VC firms, consider whether 10B+ fund sizes can deliver LP returns competitive with historical venture benchmarks; if not, consider specialization in smaller, more agile funds
Sequoia's consolidation reflects industry-wide shift toward mega-funds, but mega-funds face geometric scaling challenges in returns; smaller, focused funds may outperform
Invest in or build AI products targeting everyday consumer use cases with measurable time/money savings rather than B2B enterprise plays to shift political narrative
AI adoption sentiment depends on consumer experience, not tech industry arguments; focus on applications that touch ordinary lives (robotaxis, personal agents, home automation)
Evaluate infrastructure and power plays (chips, nuclear energy, local compute hardware) as defensive bets against GPU scarcity; these will become valuable if inference is rationed in 2027
Tomasz's prediction of GPU scarcity and rationing creates investment opportunity in alternative compute/power solutions that reduce cloud dependency
Build relationships with family offices and offshore investors now if you plan late-stage syndicate participation; regulatory restrictions on retail access to private markets will likely increase
Secondary and derivative markets are fragmenting retail access; direct family office networks will become more valuable as accredited investor rules tighten
Links Mentioned
Incorporation, banking, payroll, benefits, accounting, taxes, and back office administration for startups
https://every.io/
SOC 2, HIPAA, and security compliance platform; provides continuous monitoring and compliance automation
https://www.vanta.com/twist
Pre-vetted software engineer recruitment platform with 10+ years of vetting experience
http://G2i.co/twist
AI-powered recording device and meeting transcription platform with offline capability
https://Plaud.ai/twist
Family office and institutional investor database with detailed decision-maker tracking across 4,000-5,000 family offices
https://www.fintrx.com/