Crypto’s First Step Into US Housing: From Wallets To Mortgages
Last week, the Federal Housing Finance Agency (FHFA) instructed Fannie Mae and Freddie Mac to develop proposals to consider cryptocurrency holdings—but only those held on US-regulated centralized exchanges—as assets in single-family mortgage loan risk assessments.1 The directive does not require converting crypto to US dollars and does not include crypto held in self-custodied wallets. It also calls for risk controls, like accounting for crypto market volatility and limiting how much crypto can count toward reserves.
This directive introduces a novel bridge between blockchain-based capital and the $12 trillion US mortgage market. Should the FHFA finalize this rule, setting the precedent for wider adoption across mortgages and cryptoassets, blockchain-native balance sheets could impact the mortgage market significantly by streamlining underwriting, lowering transaction costs, and enabling token-linked mortgage instruments.
According to a 2025 Harris Poll commissioned by CryptoSlate, 21% of US adults own digital assets. Of the ~55 million individuals who own cryptoassets, 6 million hold more than $100,000 on average. Given the wealth that individuals have been accumulating on-chain, the impact of cryptoassets on credit markets could become meaningful.2
According to iEmergent’s 2024 Home Mortgage Disclosure Act (HMDA) data,3 the number of mortgage originations in the US last year was ~6 million, valued at $1.82 trillion, suggesting an average loan size of ~$340,000. On that basis, if just 5% of mortgage borrowers included crypto-based assets in their applications, ~305,000 would qualify for a mortgage in this new framework—supporting $100 billion in originations. Each additional percentage point of adoption would increase mortgage loan volume by ~$20 billion.
Based solely on penetration rate × average ticket size, our methodology assumes no leverage multipliers or velocity changes, so the upside could be materially higher. This regulatory development aligns with ARK’s thesis that crypto will reshape legacy financial systems with greater transparency, automation, and interoperability.
2. Tesla Launched Its Robotaxi Business In Austin
Last week, Tesla launched its robotaxi service in Austin for a flat fare of $4.20, albeit for a limited group of users through an invite-only robotaxi app.4 The historic rollout focused on safety,5 with operations limited to a region in Austin roughly half the size of Waymo’s6 and safety monitors in the passenger seat. According to early reports, most robotaxi rides were very smooth,7 with a few edge cases highlighting areas for improvement.8 Now, Tesla is increasing the number of invitees.9
Its commercialization in Austin puts Tesla on track to tap into ARK’s expected ~$8-10 trillion market opportunity in global robotaxi revenue in 2030, as shown below. Network providers like Tesla and Waymo could account for half of that revenue base at software-as-a-service (SaaS)-like margins.
*$10 trillion is the addressable market, not the revenue we expect in 2030, as we do not expect autonomy to penetrate all addressable miles. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Forecasts are inherently limited and cannot be relied upon.
Powered by two unique advantages, its end-to-end AI autonomous solution and vertically integrated manufacturing, Tesla is likely to scale more quickly and cost-effectively than competitors. If, as Elon has suggested, Tesla offers robotaxi service in many cities by the end of this year,10 these two competitive advantages should deliver a premium service at a lower cost, undercutting the US pricing umbrella established by Uber and Lyft. ARK will continue to track the rollout and scaling of Tesla’s robotaxi network in the coming months.
3. AlphaGenome Could Transform Genomic Interpretation
Imagine understanding how a dish will taste by knowing its ingredients. Essentially, AlphaGenome has pursued that approach to DNA. Developed by Google DeepMind, AlphaGenome is leveraging AI to transform the understanding of genomic data.11
AlphaGenome’s strength is its depth, not breadth. Trained on approximately 700 meticulously selected genomes, each enhanced by ~7,000 functional genomic experiments, AlphaGenome can predict 15 distinct molecular outcomes directly from raw DNA sequences, achieving substantial improvements relative to the status quo ranging from 2% to 42%.
In our view, AlphaGenome’s opportunity lies not in sequencing another 2.5 million genomes but in harnessing existing but underutilized functional genomic data. Data from thousands of RNA-seq, ATAC-seq, and ChIP-seq experiments have been stranded in databases, representing an immense reservoir of molecular data waiting to be activated. Paired with emerging architectural breakthroughs, such as sparse attention mechanisms and multimodal integration, AlphaGenome’s potential is vast.
We believe AlphaGenome could evolve in phases, as follows:
- Near-term (1 year): Enhanced model scaling and improved data curation could enable tissue-specific predictions not possible before AlphaGenome.
- Medium-term (3 years): Integrating AlphaGenome with spatial genomics and single-cell sequencing could lead to virtual biopsies.
- Long-term (5 years): With exponential growth in datasets and plummeting computation costs, AlphaGenome could lead to molecular biology interpretations in real time.
DeepMind has engineered AlphaGenome to transform genomic medicine into a responsive, dynamic system much like advanced AI tools have transformed coding. Its real-world impacts should manifest in millions of clinical diagnoses delivered faster, more affordably, and more equitably than ever before.