Founder stories,
partner perspectives,
and industry insights.

Consumer fintech is in a transitory phase. Consolidation is taking place as value accrues to larger incumbents. We’ve reached peak fragmentation where there is an app for every financial task.

Banking access has largely been solved for younger generations in the US. So we think the next phase will be defined by AI-enabled applications in wealth and spend. And these are increasingly consolidated in vertical superapps. We are tentatively calling this 3rd era of consumer fintech the sophistication era (name suggestions welcome!).

Today there is a relatively small group of VC-backed consumer fintech companies that are worth over $10B in market cap. Just 7 on US exchanges as of this article’s writing: Intuit, PayPal, Block, Nubank, Coinbase, Grab, and Zillow. Eight if you include Mercado Libre (~50% fintech revenue). Intuit is mostly B2B revenue and Zillow isn’t exactly a fintech. So maybe 6.

On one hand, 5 of those IPO’d during the 2nd era of personal finance, the mobile era, so you could argue it was the most successful era of personal finance yet.

On the other hand, many fintech investors feel burned by capital intensive models and CAC wars, so much so that there’s a narrative that consumer fintech is dead. We heavily disagree.

We believe the opportunity for tech-enabled personal finance has never been stronger, and we expect the 3rd era of consumer fintech to continue to advance toward the promise of a personalized finance future.

What might the 3rd era of personal finance look like?

In its most idealistic form, the goal of personal finance is for it to be completely, well, personalized. The complete personalization of finance means each user receives the specific financial help they need at that moment, tailored to their individual financial situation.

We’re not there yet. Today’s rising consumer products all point toward a step before that; the fastest growing categories of consumer fintech activity today revolve around more sophisticated spending and intelligent optimization. Savvy consumers are optimizing their marginal dollar.

Consumer finance’s 3.0 era will be dominated by sophisticated optimization of marginal dollars around 1) wealth management and 2) consumer spending that are 3) increasingly accessed through vertical superapps.

There are large macro trends driving these categories.

In wealth, we are facing the largest wealth transfer ever seen. You all know the details, so we’ll spare a long description, but more than $60 trillion of wealth is expected to change hands in the coming years as baby boomers pass on, and more than 80% of recipients (millennials) are expected to change advisors. There are over 25 million millionaires in the US, so this is not an insignificant number of people. We are also seeing this play out from a bottoms-up perspective: startups growing the most right now in wealth management are tackling several problems serving this need:

• Retirement Plan Administration
• Wealth Transfer
• Estate Planning
• Alternative Asset Access
• Sophisticated Tax Strategies
• Global Accounts as Volatility Hedges
• Yield-bearing products, Fixed-Income Tech

Consumers are becoming more savvy about how they spend their marginal dollar, optimizing for personal efficiency. The macro backdrop of increased spend choices is that this is enabled because payment rails and payment interfaces are undergoing huge changes: the emergence of new government backed rails like Pix, UPI, and FedNow, plus the popularity of stablecoins and crypto payments, mean the card networks are on watch.

We’ve unbundled the bank, now it’s time to unbundle the card networks. We are also seeing this play out from a bottoms-up perspective: startups growing the most right now in spend are tackling several problems serving this need:

• Multiplayer Finance
• Asset backed cards for lower APR
• Savings, Refunds, & Robo Claims
• Pay by Bank
• Digital currencies
• Climate Alignment
• Top of Wallet Spend Optimization
• Debit Rewards + Loyalty
• Global Accounts for Borderless Spend
• Verticalized Spend

In vertical superapps, the macro backdrop is rebundling, taking place after 15 years of bull-market fueled consumer fintech mania due to: macro downturn overfunding and underbuilt startup economic models; consumer fear due to bank collapses (SVB/FRB); and the US’s unique financial market structure not being compatible with the birth of financial superapps like we’ve seen with WePay, Alipay, Mercado Libre, Grab, etc., in other countries.

Acquisitions are underway, startups are dying, and finances are increasingly embedded in workflows and commerce experiences. We are seeing this play out from a bottoms-up perspective: startups we’re meeting that are benefiting from consolidation pressures include:

• Vertical saas fintech (hospitality, restaurant, construction, education)
• Embedded fintech
• BaaS
• Data aggregators
• Growth stage consumer fintechs via m&a and cross selling

Peak Fragmentation

It makes sense that this third era is unfolding now. US consumers already have a plethora of digital banking solutions. The average American has 5-7 debit cards and 6-8 credit cards. Consumers have an overabundance of cards.

In fact, we think we’ve already reached the peak unbundling phase of consumer fintech. There are only 2 business models – unbundling and re-bundling – and after decades of unbundling the bank, we are already starting to see more consolidation.

For example, every mobile banking app of the last decade is now cross-selling their users additional banking products (see part 1 in of this series). The US market is saturated with banking options.

Consumers have their digital banking foundation set. So what do consumers need now?

They need a way to make sense of all their financial activity across institutions. Not just aggregation, but an intelligence layer that optimizes spend. That finds sophisticated wealth and tax solutions. That delivers insights about how to best allocate that consumer’s marginal dollar.

The need for an intelligent solution to optimize consumer spend couldn’t have been timed better; it is arriving just as generative AI sweeps tech. Similar to our identification of SaaS entering its 4th era, the Era of Cognition, personal finances are entering their 3rd era, the Sophistication Era.

Past eras

The second era of consumer fintech, the mobile or neo banking era, is ending. Before this, the first 2 eras of consumer fintech in the US were dominated by digital banking and brokerage platforms moving online.

The first era saw PayPal, eTrade, TurboTax (Intuit), Capital One, Mint, etc. grow during the 90s and 2000s as primarily desktop solutions, digitizing previously offline activities (PayPal – spend, eTrade – invest, TurboTax – taxes, Capital One – borrow).

The second wave was mobile first and saw CashApp, Robinhod, Affirm, Coinbase, SoFi, etc., take market share by transitioning those personal finance activities – spending, investing, and banking – into a seamless, on-the-go experience.

The third era, unfolding now, is all about sophisticated spending and wealth management. Despite having all the fintech apps we need, we still can’t answer basic personal finance questions like, “What two things should I do to most improve my financial situation?” or “How much did I spend across all accounts this month?”

As one Twitter user pithily put it, “there is an app for everything but you still cannot do everything in one app.”

AI’s impact on Sophisticated Wealth + Spend Choices

AI will impact wealth and spend. On the wealth side, sophisticated planning is the focus. On the spend side, sophisticated marginal dollar spend is the big focus.

Sophisticated Wealth

We’re seeing AI applied to wealth management strategies in limited contexts today: Roboadvising has been around for a long time and is automation of savings/investing its most basic form. We are also seeing more AI-directed portfolio strategies roll out and reaching strong product-market fit with tech-savvy users, but still require a lot of education for the average consumer.

Instead we are seeing the consumer opt for more sophisticated wealth management choices through tech-enabled tax-advantaged strategies, estate planning, alternatives access, global accounts as volatility hedges, and much more.

Asset management is also a large line item in JPMorgan’s revenue breakdown. While we believe it will take longer for consumers to trust a fully automated wealth solution, there is a large profit pool up for grabs. 80% of millennials are expected to change advisors when inheriting wealth from their parents: that’s over $50 trillion up for grabs for new financial institutions serving millennials.

We believe it will require a human-in-the-loop for the majority of users for several more years as consumers gain comfort with AI touching their nest egg.

Sophisticated Spend

We’re already seeing AI applied to “found money” opportunities, especially liabilities, as users become more savvy about sophisticated spend.

Today, we are already seeing consumers gain comfort with delegating responsibility for select areas of the liabilities side of their balance sheet to AI. Over the next few years, we should expect consumers to gradually increase comfort level to include all liabilities. Several years after that, trust and education levels should increase such that assets (checking, savings, retirement) can begin to be automated for the masses through AI.

After both of those are achieved, then the vision of AI as the consumer fintech AI interface laid out above can begin to unfold.

As we meet founders today in mid 2023, there are only a few reliable areas where consumers trust AI to fully handle their tasks end to end: 1) debt removal 2) subscription cancellation 3) find savings + new money.

This is for a fairly straightforward reason: most consumers are happy to sign up for a service that can help them decrease their liabilities and increase their assets without risk of losing their nest-egg. It’s unclear whether the 3 product categories listed above will create enough touch points with the user to be a strong wedge into a longer banking relationship.

AI will take profits from incumbents

Highly valued consumer fintechs of the last era (Nubank, PayPal, Block, etc) succeeded by attacking structurally high profit margin incumbencies (Brazil banking, international money transfer, external bank transfers, acquiring), which created space for lower friction digital products to emerge.

Consumers’ personal balance sheet – deposits and liabilities – are a bank’s revenue opportunity. Continued advancement in AI will enable consumers to get a better deal.

Just look at JPMorgan’s 2022 revenue breakdown: over 50% of their $128B in revenue came from interest income, or revenue earned from interest bearing assets – a combination of yield bearing assets, lending, etc.

Some areas where we think there are large remaining profit pools in the USA :

• 6% buy and sell commission for home sales
• Wire fees
• Interchange fees from card networks: acquiring + issuing
• 1% AUM fees for wealth management
• 20-25%+ APRs for consumer credit (earned wage access is even higher)

Where else can tech remove friction from a big profit pool business for incumbents now that AI is here? Feel free to write to us with your thoughts.

AI will improve the consumer fintech business model

We outlined some of the common complaints about consumer fintech business models at the start of this piece. There are really 4 areas a consumer fintech can have a cost advantage: cost to acquire, cost to underwrite, cost to service, and cost of capital.

AI has been used to improve the cost of underwriting for a very long time in lending (see our fintech AI blog post). We only expect this to improve.

The area we’re seeing the greatest change in is generative AI’s ability to reduce the cost to serve users. From robo debt recovery calls, to AI-enabled capital markets benchmarking, to extending the number of users a wealth manager can service, to automatic refund cancellation, to sophisticated tax strategies for the masses, automation is reducing COGS by 30 to 80% across companies we’ve met.

Incumbent banks continue to have a massive advantage in operating with a lower cost of capital. We are starting to see fintechs securing better debt terms using AI-driven software.

Acquisition costs are becoming a more even playing field. We’ve seen consumer fintechs using AI chatbots as acquisition tools. We’d love to hear from founders on even more ways you’re improving costs with AI.

Where incumbents bank advantage remains

The two areas where incumbents continue to have a large advantage over consumer fintechs: 1) distribution – cost of acquisition of subsequent products 2) cost of capital advantage – they can fund their working capital cycles at nearly the Fed Funds rate, the lowest cost available.

The counter attacks to big bank’s distribution and cost of capital advantage are 1) turning commerce experiences into banking experiences – enabling anyone to become a bank (consolidation theme) and 2) encouraging users to fund their own risk: incentivizing the right type of behavior / risk alignment by introducing fee-based products and subscriptions tied to greater financial freedom on platform (collateralized assets that bring down the cost of capital – spend theme).

We think there is still a large opportunity to innovate in these two remaining bank strongholds as the 3rd era of sophisticated finance unfolds.

If you’re looking around, seeing the last era of neobanks still early in their journey and wondering if consumer fintech is dead, think again. The category is evolving into one where sophisticated spending, sophisticated investing tools, and vertical superapps will dominate. This shift has created a very uncertain time in consumer fintech. We are in the midst of a paradigm shift from the second era of consumer fintech to the third.

There is so much opportunity ahead.

If you enjoyed this, don’t hesitate to reach out. We can be reached at fintech_contact@lsvp.com. Checkout other related articles including The Successful Consumer Fintech Path, Fintech x AI, and Lightspeed’s 2023 Fintech Trends.

Truly transformative technologies build new markets, upend existing ones, and impact every aspect of how we live, work, and play. At Lightspeed, we’ve been fortunate to have witnessed and participated in the rise of multiple transformative technologies over the last 25 years – from the rise of the internet to the rise of mobile, and now the rise of AI – specifically generative AI.

These transformative cycles are both powerful and chaotic at the same time. And yet, we’ve seen incredible Founders time and again connect the dots between the emergence of a technical innovation and the massive opportunity it unlocks. This happens up and down the stack, across sectors and geographies. A similar narrative is playing out in AI right now. We have been working with extraordinary companies building AI-native applications across sectors, such as enterprise SaaS, robotics, consumer, healthcare, and fintech, as well as engineering the underlying infrastructure to help develop and run AI-based applications. The following image highlights some of these amazing teams, and the sector they are building in. You can also read more about each teams’ ambitious vision for the future on our companies page.

There’s no pre-determined sequence in which transformative technologies play out in the broader technology ecosystem. Often, things start with the emergence of one new infrastructure capability which catches the attention of developers, who then start building on top of it. This virtuous cycle of applications pushes the bounds of the infrastructure, accelerating further innovation, leading to ever more broader application of that technology within applications, and so on.

A similar cycle of infra and apps making each other better is playing out in AI. OpenAI’s introduction of transformer based models (GPT2/3) led to the emergence of applications in areas such as copywriting, blog writing etc. The increased developer interest led to continued innovation by OpenAI (GPT3/3.5/4, ChatGPT, Whisper) but also led to a Cambrian explosion of open source models from Meta, Stability, Mosaic, Databricks, and many more. This then feeds rapid incorporation of generative AI within applications, as well as the rise of generative AI operational, security, and observability platforms.

In subsequent articles, we’ll expand on the major markets outlined in the above map to outline the areas we are actively investing in, and the themes we are excited about. Our focus at Lightspeed is on having prepared minds, so we can connect the dots and dream together with our Founders. And we aren’t done yet – we’d love to talk to you if you have a strong point of view on how AI will create opportunities in this rapidly changing economy.

More Reading from Lightspeed

• Report
  • Eight AI Startups Winning The Race For Tech Talent – Meet the Generative Eight — and The Next Gen of AI companies building the future
• Enterprise Applications 
  • SaaS 4.0: Say Hello to The Era of Cognition – Generative AI will eventually be a native part of every enterprise software company. Here’s how we see the winners emerging.
• AI Infrastructure 
  • Will Enterprise AI Models be “Winner Take All”? – How a Cambrian explosion of models creates a broad set of opportunities in AI Infrastructure
  • Meet Europe’s Next Great Generative AI Startup: Mistral AI – Founded by key members of Meta’s and Alphabet’s AI research teams, Mistral AI has closed a seed round of over €105 million, led by Lightspeed.
  • Contextual AI: Making Enterprise-Ready LLMs Available For All – Announcing Lightspeed’s latest investment in the AI space
• FinTech
  •  Fintech x AI: The Lightspeed View – The future of finance: Not all AI is created equal
• Consumer 
  • Startups v. Incumbents—The Battle For AI’s Application Layer – If your AI product is going head to head with an incumbent, their distribution advantage will probably kill your startup…unless you fight back with a different game.
  • Fictional Reality—Why the Real Metaverse Doesn’t Require a Headset – AI is letting us live out alternative realities on the digital platforms where we already spend our time
• Gaming
  • Gaming x AI Market Map: The Infinite Power of Play – An (inevitably incomplete) overview of the emerging landscape of artificial intelligence and procedural generation in interactive media
• Generative Event Recaps
  • Scott Belsky on Creativity in Generative AI – An AI-enhanced transcript of the conversation with Adobe’s Chief Strategy Officer and EVP of Design and Emerging Products
  • How Founders Can Spot the Next Big Opportunity in AI – Spotify’s VP of Personalization believes there is one underrated trait that all founders need for success.
  • Fintech in AI: Five Lessons For Founders – Takeaways from our conversation with Grace Liu, Alloy’s Director of Product; Sasha Orloff, founder and CEO of Puzzle.io; and Leif Abraham, co-founder and co-CEO of Public.com.
• Lightspeed AI Reading List

Companies on our Portfolio Map (57 and counting)

• AI/ML Ops
  • Snorkel
  • Fiddler
  • Claypot
• Consumer, Commerce
  • Big Sur AI
  • Pocket FM
  • Volley
  • ShareChat
  • Circle Labs
• B2B Enterprise
  • Appzen
  • Glean
  • CueIn
  • Moveworks
  • Pepper Content
  • Rephrase.ai
  • Yellow.ai
  • ThoughtSpot
  • Typeface
  • Aquant
  • Blink
  • People.ai
  • NextBillion.ai
  • Osmos
• Productivity
  • Tome
  • Tana
  • Xembly
  • Granola
  • Butler Labs
  • Gushwork
  • Rattle
  • Tonkean
  • Shortwave
  • Magical
  • Thena
• Security Observability GRC, Safety
  • Patronus AI
  • Portkey
• Vertical Applications
  • Navan
  • Personetics
  • Eightfold
  • Zest AI
  • Notch
• Healthcare
  • Fathom
  • Triana
  • Innovaccer
  • Soda Health
  • Seismic Therapeutic
• Gaming, Media
  • Inworld
  • Believer
  • Gardens
• Autonomy & Robotics
  • Reliable Robotics
  • Dexterity
  • Apex.AI
  • Aurora
  • ReSim.ai
  • Smartex
• AI Infra and Foundation Models
  • Mistral AI
  • Alif Semiconductor
  • Stability.ai
  • Contextual.ai

Over the past decade, we at Lightspeed have had a front row seat to the incredible innovations in AI/ML thanks to the amazing founders we’ve had the privilege to partner with. We’ve been working with their companies, the platforms they’ve built, and the customers they serve, to better understand how enterprises are thinking through GenAI. Specifically, we have investigated the foundation model ecosystem with questions like, “Will the best performing model have winner-take-all dynamics?” and, “Will enterprise use cases all default to calling, say, OpenAI APIs, or will actual usage be more diverse?” The answers will determine the future growth of this ecosystem, and in what direction energy, talent, and dollars will flow.

Categorizing the Model Ecosystem

Based on our learnings, our belief is that there’s a Cambrian explosion of models coming in AI. Developers and enterprises will pick models best suited for the “job to be done,” even though the usage at exploratory stages might look a lot more concentrated. A likely path for enterprise adoption could be the use of big models for exploration, gradually moving to smaller specialized (tuned + distilled) models for production as they learn more about their use case. The following visual helps outline how we see the Foundation Model ecosystem evolving.

The AI model landscape can be divided into 3 primary, though somewhat overlapping, buckets:

Bucket 1: “Big Brain”

These are the best of the best, the leading edge of models. This is where the exciting magical demos that have captivated us all are coming from. They are often the default starting point for developers when trying to explore the limits of what AI can do for their apps. These models are expensive to train, and complex to maintain and scale. But the same model could take the LSAT, the MCAT, write your high school essay, and engage with you as a chat-bot-friend. These models are where developers are currently running experiments and evaluating AI usage within enterprise applications.

But, they are expensive to use, inference latency is high, and can be overkill for well-defined constrained use cases. A second issue is that these models are generalists that can be less accurate on specialized tasks. (For instance, see meta studies such as this one from Cornell.) Lastly, today, in almost every case, they are also black boxes that can present privacy and security challenges for enterprises who are grappling with how to utilize these without giving away the farm (their data!). OpenAI, Anthropic, Cohere are examples of companies in this bucket.

Bucket 2: “Challenger”

These are also high-capability models, with skills and abilities just below the most cutting edge. Models such as Llama 2 and Falcon are the best representations in this category. They are often as good as some of the Gen “N-1” or “N-2” models from the companies building the Bucket 1 models. By some benchmarks, Llama2, for instance, is as good as GPT-3.5-turbo. Tuning these models on enterprise data can bring their abilities to be as good as Bucket 1 models on specific tasks.

Many of these models are open source (or close enough) and once released have led to immediate improvements and optimizations by the open source community.

Bucket 3: “Long Tail”

These are “expert” models. They are built to serve a narrow purpose, like classifying documents, identifying a specific property in an image or video, identifying patterns in business data, etc. These are nimble, inexpensive to train and use, and can be run in data centers or on the edge.

A quick look at Hugging Face is enough to get a sense for how vast this ecosystem already is and will grow to be in the future, thanks to the breadth of use cases it serves!

Matching Use Cases to Models

While early, we’ve seen some of the leading development teams and enterprises already thinking about the ecosystem in this nuanced way. There’s a desire to match the use to the best possible model. Perhaps even use multiple models to serve a more complex use case.

The factors used in evaluating which model/s to use often include the following:

1. Data privacy and compliance requirements, which impact whether the model needs to run in enterprise infrastructure, or if data can be sent to an external hosted inference endpoint.
2. Whether the ability to fine tune a model is critical or strongly desired for this use case.
3. What level of inference ‘performance’ is desired (latency, precision, expense, etc.).

The actual list is often much longer than just the above and is reflective of the tremendous diversity in use cases that developers would like to use AI for.

Where are the opportunities?

There are several implications of this emerging ecosystem:

1. Evaluation Frameworks: Enterprises will need access to tooling and expertise that can help evaluate which model to use for which use case. Developers will need to decide how best to evaluate the suitability of a specific model for the ‘job to be done.’ The evaluation would need to be multi-factor and include not just the performance of the model, but also the cost, the level of control that can be exercised etc.
2. Running and Maintaining Models: Platforms to help enterprises train, fine tune, and run models, especially the Bucket 3, long tail models, will emerge. Traditionally, these have broadly been referred to as ML Ops platforms, we expect that definition will expand to include generative AI as well. Platforms such as Databricks, Weights and Biases, Tecton and others are rapidly building towards this.
3. Augmentation Systems: Models, particularly hosted LLMs, need retrieval augmented generation to deliver superior results. This requires a secondary set of decisions to be made, including:
   • Data and meta-data ingestion: How to connect to structured and unstructured enterprise data sources and then ingesting both data, as well as meta data on things such as access policies.
   • Generating and storing embeddings: Which model to use to generate embeddings for the data. And then how to store them: Which vector database to use, specifically based on performance, scale and functionality desired?

Opportunities exist here to build enterprise class RAG platforms which can take away the complexity associated with selecting and stitching together these platforms:

1. Operational Tooling: Enterprise IT will need to build guardrails for engineering teams, manage costs, etc.; all the tasks that they handle today for software development need to be expanded now to include AI usage. Areas of interest for IT include:
   • Observability: How are the models doing in production? Is their performance improving/degrading with time? Are there usage patterns that might impact the choice of model in future versions of the application?
   • Security: How to keep AI native applications secure. Are these applications vulnerable to a new class of attack vectors that need new platforms?
   • Compliance: We expect AI native applications and usage of LLM will need to be compliant with frameworks that relevant governing bodies are already beginning to work on. That’s in addition to the existing compliance regimes around privacy, security, consumer protection, fairness etc. Enterprises will need platforms that help them stay compliant, run audits, generate proof of compliance and associated tasks.
2. Data: Platforms to help understand the data assets an enterprise has, and how to use those assets to extract the maximal value from new(er) AI models, will see rapid adoption. As one of the largest software companies on the planet once said to us, “our data is our moat, our core IP, our competitive advantage.” Monetizing this data using AI, in a way that drives additional “differentiation without diluting defensibility” will be key. Platforms such as Snorkel play a critical role in this.

This is an amazing time to be building platforms in AI infrastructure. Adoption of AI will continue to transform entire industries, but it will require supporting infrastructure, middleware, security, observability, and operations platforms to enable every enterprise on the planet to adopt this powerful technology. If you are building to make this vision become a reality, we’d love to speak with you!

Learn more about our investment in Inworld AI, the leading character engine for AI NPCs.