🌡️ SF Tech Week Founder Meetup: Are We in an AI Winter or Just Getting Started?

[maya_builds]

The indie hacker/founder dinner tonight got HEATED. Topic: “AI Winter or AI Spring?”

Half the room: “AI funding is at all-time highs! We’re in a golden age!”

Other half: “It’s a bubble. Most AI startups will die. This is 2021 crypto vibes.”

Both sides brought receipts.

The “AI Spring” Argument

Data point 1: AI funding DOMINATES venture capital

From the a16z partner who spoke:

• 2024: $131.5B in AI VC funding (33% of ALL VC)
• Q4 2024: 50.8% of all global VC went to AI companies
• H1 2025: Generative AI funding already surpassed entire 2024

Source: Crunchbase, a16z “AI companies that startups actually pay for” report (Oct 2, 2025)

Data point 2: Software/AI is eating the world (for real this time)

• Software/AI now 45% of all VC funding
• Every industry being disrupted: healthcare, legal, finance, education
• Enterprises actually deploying and PAYING for AI

Data point 3: This is JUST the beginning

• LLMs are 2-3 years old
• Compare to: Internet in 1995, mobile in 2009
• We’re in the “innovation” phase, not the “decline” phase

The optimists: “We’re in the first inning of a 20-year transformation.”

The “AI Winter” Argument

Data point 1: Concentration risk is INSANE

From the skeptical founder who bootstrapped a profitable SaaS:

• OpenAI raised $40B in Q1 2025
• That ONE round skewed the entire market
• Remove mega-rounds: VC is actually getting MORE selective

Data point 2: Path to profitability is unclear

• Most AI startups have no revenue model
• “We’ll build audience then monetize” = 2021 thinking
• Only 12% of gen AI companies are profitable

Quote from the founder: “Show me the unit economics. I’ll wait.”

Data point 3: Exits are a mirage

• 43 new unicorns in 2025 vs 340 in 2021
• Exit timelines extended to 12+ years
• M&A market frozen (antitrust, valuation mismatch)

The skeptics: “This ends like crypto 2022. 90% of AI startups dead in 2 years.”

The Data I’m Wrestling With

On one hand:

• AI funding at record highs ($131.5B in 2024)
• H1 2025 already beat 2024 total
• Real enterprise adoption (not just hype)
• Technology genuinely transformative

On the other hand:

• Valuations disconnected from revenue (seed rounds at $20M pre with $0 revenue)
• Most AI startups are thin wrappers around OpenAI API
• “AI-washing” everywhere (every pitch deck has AI buzzwords)
• VCs MORE selective despite high funding numbers

My Take as an Indie Hacker

I’m building an AI-powered tool for developers. Bootstrapped, no VC.

Here’s what I’m seeing in the trenches:

The market is REAL:

• Developers actually want AI coding tools
• They’re paying $20-50/month (Cursor, Replit, etc.)
• GitHub Copilot has millions of paying users

But competition is BRUTAL:

• OpenAI could release Codex 2.0 tomorrow and crush us
• Hard to differentiate (we all use similar models)
• Pricing pressure (race to the bottom)

My survival strategy:

1. Niche down: I’m not competing with Cursor. I’m building for Rust developers specifically.
2. Own the customer: Not just a wrapper. Building features OpenAI can’t/won’t.
3. Revenue from day 1: No “free tier to get users then figure it out later”
4. Capital efficient: No office, no fundraising, profitable in month 3

The SF Tech Week Vibe Check

Walking around SF this week, the energy is:

• High optimism (AI is transforming everything!)
• High anxiety (What if my startup is the one that fails?)

Parallels to 2021:

• Easy to raise seed rounds
• Sky-high valuations
• Every startup claims to use AI
• FOMO among investors

Differences from 2021:

• Technology actually works (LLMs are real)
• Enterprises deploying at scale (not just hype)
• VCs MORE selective (despite high funding totals)
• Exit environment worse (longer timelines, fewer IPOs)

The Question That Stumped Everyone

During dinner, someone asked:

“If we’re NOT in a bubble, why are seed valuations 3x higher than fundamentals justify?”

“If we ARE in a bubble, why is enterprise AI spending growing 40% Q/Q?”

Nobody had a good answer.

My Questions for This Community

For founders:

• Are you building a VC-backed AI startup or bootstrapping?
• How are you thinking about defensibility (when everyone has access to same models)?
• What’s your timeline to profitability?

For investors/operators:

• Which AI business models are actually working?
• What separates the 10% that will succeed from the 90% that will fail?

For everyone:

• AI Winter, AI Spring, or something in between?

I’m genuinely trying to figure this out. My indie hacker livelihood depends on reading this market correctly.

Sources:

• SF Tech Week founder dinner (Day 4, ~40 founders and investors)
• Crunchbase AI funding data 2024-2025
• a16z “AI companies that startups actually pay for” report (Oct 2, 2025)
• GoingVC Top VC Trends 2025
• Allvue Systems VC Trends 2025
• SF Examiner “SF AI startups benefitting from 2025 VC boom”

[product_david]

@maya_builds This resonates hard. I’m at a Series A AI startup and we’re living this tension every day.

Inside a VC-Backed AI Startup Right Now

Our metrics (can’t name company, but this is representative):

Funding:

• Raised $12M Series A in March 2025 at $50M post-money valuation
• $0 revenue at time of raise
• Pitch: “AI-powered X for enterprise”

Current state (7 months later):

• $400K ARR (annual recurring revenue)
• Burn: $800K/month
• Runway: 14 months
• Team: 25 people

Valuation math:

• Revenue multiple: 125x (!!!)
• For context: Normal SaaS trades at 10x revenue
• We’re priced for “exponential AI growth”

Are we in a bubble? You tell me.

What’s Actually Working (And What’s Not)

Not working: AI wrappers

Startups that are just thin wrappers around GPT-4:

• Can’t defend against OpenAI moving down the stack
• Commoditized instantly
• Investors have wised up (this strategy worked in 2023, not anymore)

Working: Vertical AI with proprietary data

Harvey AI (legal), Glean (enterprise search), etc:

• Built deep expertise in one vertical
• Have proprietary data moats
• Hard to replicate

Not working: Consumer AI

• Character.AI burned $150M, struggling to monetize
• User engagement high, revenue low
• Ad model doesn’t work (privacy concerns with AI)

Working: B2B AI with clear ROI

• Developer tools (Cursor, Replit)
• Enterprise workflow automation
• Companies that can show “AI saves you $X/month” → get paid $X/2

The Business Model Question

You’re right to focus on this. Most AI startups have no clear path to profitability.

The failed models:

“Free to build audience, monetize later”

• Worked for social media (ad model)
• Doesn’t work for AI (compute costs too high)
• You’re just burning money on inference

“Usage-based pricing”

• Sounds good in theory
• Reality: Your costs (API fees) scale with usage
• Margins razor-thin
• Hard to get to profitability

“Enterprise will pay anything for AI”

• True in 2023-early 2024
• Not true anymore in late 2025
• Enterprises are getting price-sensitive
• “AI fatigue” setting in

The working models:

Seat-based SaaS with AI features

• Charge per user, not per token
• AI improves product, but isn’t the pricing model
• Margins improve as you optimize inference costs

Outcome-based pricing

• “Pay for results, not API calls”
• Example: “Pay per contract analyzed” not “pay per token”
• Aligns incentives with customer value

Hybrid: Base fee + usage

• Base SaaS fee covers costs
• Usage pricing captures upside
• Most sustainable model I’m seeing

Your Indie Hacker Advantage

@maya_builds your approach is actually BETTER than most VC-backed startups:

1. Niche down (Rust developers specifically)
2. Own the customer (features OpenAI can’t/won’t)
3. Revenue from day 1
4. Capital efficient

Compare to us:

1. Trying to serve everyone (enterprise, SMB, startups)
2. Raising VC = pressure to grow fast, not sustainably
3. Gave away free tier for 6 months to “build audience”
4. Team of 25 in expensive SF office

You’ll probably outlive 80% of VC-backed AI startups.

The Uncomfortable Truth

Here’s what I learned from our board meeting last week:

Investors KNOW most AI startups will fail.

They’re playing a portfolio game:

• Invest in 50 AI startups
• 45 will fail
• 3 will return capital
• 2 will be home runs (100x returns)

That portfolio math works for THEM. It doesn’t work for US (founders).

When I joined this startup, I thought: “We’re special. We’ll be in the 10% that succeed.”

Now I’m less sure. We have 14 months of runway and no clear path to Series B.

The bar for Series B in late 2025:

• $2M+ ARR (we’re at $400K)
• 100%+ YoY growth (we’re at 50%)
• Clear path to $10M ARR (unclear for us)

AI Winter, Spring, or Selective Fall?

My answer: It’s a selective market.

AI Spring for:

• Companies with real revenue and unit economics
• Vertical AI with defensibility
• B2B with clear enterprise ROI
• Capital-efficient bootstrapped companies

AI Winter for:

• Thin wrappers around OpenAI
• Consumer AI with no monetization
• Companies burning millions with no revenue
• “AI-washing” without real AI value

The split is happening NOW. By end of 2026, it’ll be obvious who’s in which camp.

My Advice (From the Trenches)

If you’re fundraising:

• Only raise if you have clear path to revenue
• Understand: VC money is not validation, it’s debt you repay with equity
• Ask: “Would this work bootstrapped?” If no, be very careful.

If you’re bootstrapping (like @maya_builds):

• You’re playing a different game (sustainable growth vs. VC exponential)
• Your advantage: No pressure to raise Series B
• Focus on profitability, not growth-at-all-costs

If you’re joining an AI startup:

• Ask to see revenue, burn rate, and runway
• Red flag: No revenue, high burn, vague monetization plan
• Green flag: Real customers paying real money

The Question I’m Asking Myself

“If we can’t raise Series B in 14 months, what’s the plan?”

Options:

1. Grow into Series B metrics (need 5x ARR growth)
2. Get acquired (but M&A market is frozen)
3. Cut burn and extend runway (lay off 50% of team)
4. Shut down

This is the reality most AI startups aren’t talking about publicly.

Sources:

• Our internal metrics and board discussions
• a16z report on AI business models that work
• Conversations with other Series A AI companies at SF Tech Week
• Crunchbase data on AI startup outcomes

[sales_jenny]

Sales perspective on the “AI Winter or Spring” debate: Follow the CUSTOMER MONEY, not the VC money.

What I’m Seeing in Enterprise Sales (October 2025)

I sell AI products to enterprises. Here’s what’s changed in the last 12 months:

2024 (AI Spring):

• “We need AI!” (FOMO buying)
• Deals closed in 4-6 weeks
• Minimal price negotiation
• Everyone had AI budget

2025 (AI Reality Check):

• “Show us ROI” (skeptical buying)
• Deals taking 4-6 MONTHS
• Heavy price negotiation
• AI budgets being scrutinized

The Enterprise AI Buying Cycle Has Matured

Phase 1 (2023): Experimentation

• “Let’s try AI and see what happens”
• Small deals, lots of POCs
• Low friction, high curiosity

Phase 2 (2024): Enthusiasm

• “AI is transforming our industry!”
• Medium deals, faster closes
• Budget flowing freely

Phase 3 (2025): Rationalization ← WE ARE HERE

• “Half our AI pilots failed. Let’s be strategic.”
• Larger deals but MUCH longer sales cycles
• CFOs asking “What’s the ROI?”

The Questions I’m Hearing from Buyers

2024 questions:

• “What can your AI do?”
• “How fast can we deploy?”
• “Who else is using this?”

2025 questions:

• “What’s the total cost of ownership?”
• “How does this integrate with our existing stack?”
• “What happens if we don’t renew? (Data portability, vendor lock-in concerns)”
• “Can you show us 3 customers with measurable ROI?”
• “How is this different from just using ChatGPT Enterprise?” ← This is the killer question

The ChatGPT Enterprise Problem

OpenAI launched ChatGPT Enterprise at $60/user/month.

For many use cases, it’s “good enough.”

Why would enterprise pay an AI startup $50K-200K when they can:

• Pay OpenAI $60/user for ChatGPT Enterprise
• Get GPT-4, custom models, admin controls
• No vendor risk (OpenAI isn’t going bankrupt)

The only answer: You need to be 10x better for a specific use case.

“We’re better” isn’t enough. “We’re 10x better for X vertical” is required.

Real Data from My Pipeline

I work for an AI startup (Series A, similar to @product_david’s company).

Our pipeline metrics:

Q1 2024:

• 100 leads → 25 qualified opps → 8 closed deals
• Close rate: 32%
• Average deal size: $50K
• Sales cycle: 45 days

Q3 2025:

• 100 leads → 15 qualified opps → 3 closed deals
• Close rate: 20%
• Average deal size: $35K (price compression!)
• Sales cycle: 120 days

Revenue per 100 leads:

• Q1 2024: $400K
• Q3 2025: $105K

That’s a 74% DECLINE in sales efficiency.

Why Deals Are Dying

I’ve lost 12 deals in the last 2 months. Here are the reasons:

“Budget frozen” (5 deals)

• CFO said “no new AI spend until we see ROI from existing”

“Building in-house” (3 deals)

• “We’ll just use GPT-4 API and build ourselves”

“Not now” (2 deals)

• Pushing to 2026 budget cycle

“Went with competitor” (1 deal)

• Lost to cheaper alternative

“POC didn’t show ROI” (1 deal)

• We saved them 10 hours/week but couldn’t quantify dollar value

The Startups That Are Still Closing Deals

At SF Tech Week sales meetup, compared notes with 15+ sales leaders.

Who’s still hitting quota:

Developer tools with clear productivity metrics

• “We make developers 30% faster” = measurable ROI
• Cursor, Replit, etc. still growing strong

Vertical AI with compliance angle

• Healthcare AI that ensures HIPAA compliance
• Financial AI that reduces regulatory risk
• Sell on “avoid $1M fine” not “save time”

AI that replaces headcount

• Brutal but true: “Replace 2 FTEs” sells better than “make FTEs more efficient”
• CFOs love clear cost savings

Who’s struggling:

Horizontal AI platforms

• “AI for everyone!” = AI for no one
• Too generic, can’t prove ROI

Consumer AI pivoting to enterprise

• No enterprise GTM experience
• Product built for consumers, not IT procurement

AI “nice-to-haves”

• If it’s not solving urgent pain, it’s not closing in 2025

My Prediction: Bifurcation

AI Spring for startups with:

• Clear ROI (dollars saved or revenue generated)
• Vertical focus (deep expertise in one industry)
• Enterprise-ready (security, compliance, support)

AI Winter for startups with:

• Vague value prop (“AI makes you more productive!”)
• Horizontal approach (trying to serve everyone)
• Consumer product trying to sell to enterprise

Advice for @maya_builds and Other Founders

You asked about survival strategies. Here’s what’s working in sales:

1. Niche down EVEN MORE

• You said “Rust developers” - perfect
• Specific > generic in this market

2. Lead with ROI, not features

• “Reduces debugging time by 40%” > “AI-powered code analysis”
• Quantify value in dollars or hours

3. Product-led growth > sales-led

• Let customers try before they buy
• Usage-based pricing (easier to get started)
• Land and expand (start small, grow within account)

4. Focus on retention, not just acquisition

• Retaining 1 customer > acquiring 1 customer (cheaper)
• Churn is the silent killer in this market

5. Build community

• Indie hackers trust other indie hackers
• Word-of-mouth > outbound sales in tight market

AI Winter or Spring? Both.

Spring: Technology is real, adoption is growing, enterprises ARE deploying AI

Winter: Bubble valuations, most startups will fail, sales cycles lengthening, buyer skepticism rising

For VC-backed startups burning millions: Winter is coming

For capital-efficient startups solving real problems: Endless spring

Sources:

• My sales pipeline data (Q1 2024 vs Q3 2025)
• SF Tech Week sales leader meetup (Day 4)
• Enterprise AI buyer conversations (200+ in 2025)
• Competitor intelligence from deals won/lost

[finance_fred]

Finance/operations perspective: Let’s look at the NUMBERS to answer “Winter or Spring?”

The Funding Data (What’s Real, What’s Noise)

I analyze VC markets professionally. Here’s how to read the conflicting signals:

Headline: “AI funding at all-time highs! $131.5B in 2024!”

Reality: Strip out mega-rounds (OpenAI $40B, Anthropic $7B, etc.) and the market looks VERY different.

Funding Distribution Analysis

2024 AI VC funding breakdown:

• Top 10 mega-rounds: $78B (59% of total)
• Everyone else: $53.5B (41% of total)

For context:

• 2021 (peak): Top 10 rounds were 31% of total
• 2024: Top 10 rounds are 59% of total

Translation: Money is concentrating in fewer, larger bets. The “long tail” of AI startups is actually getting LESS funding.

The Valuation Reality Check

From Crunchbase and EY data:

Seed stage valuations:

• 2021 peak: $12M median post-money
• 2025: $8.4M median post-money
• Change: -30%

Series A valuations:

• 2021 peak: $40M median post-money
• 2025: $32M median post-money
• Change: -20%

BUT: AI startups command 2-3x premium over non-AI

So AI seed round in 2025 = $16-25M post-money (still high, but down from 2024)

The Exit Reality

This is where “AI Spring” narrative falls apart:

Unicorns created:

• 2021: 340 new unicorns
• 2024: 73 new unicorns
• 2025 YTD: 43 new unicorns

IPOs:

• 2021: 1035 IPOs globally
• 2024: 108 IPOs globally
• 2025 YTD: 92 IPOs (on pace for ~120)

M&A:

• Median acquisition price DOWN 40% from 2021
• Antitrust scrutiny killing big tech M&A
• Strategic buyers more cautious

Average time to exit:

• 2015: 8 years
• 2025: 12+ years

What This Means for Startups

If you raised in 2023-2024 at high valuations:

You’re priced for exponential growth. Series B/C requirements:

• Series A → Series B: Need 3-5x ARR growth
• Series B → Series C: Need 3-4x ARR growth

Example:

• Raised $12M Series A at $50M post, $400K ARR
• Series B target: $60-80M post, need $2-3M ARR
• That’s 5-7x growth required

If you don’t hit those numbers:

• Down round (valuation cut, major dilution)
• Extension round (flat valuation, runway only)
• Death (shut down)

If you’re capital efficient:

You can ride out the cycle:

• Don’t need to raise at inflated valuations
• Can wait for market to stabilize
• Can grow into reasonable valuation

The Unit Economics Test

Here’s the framework I use to evaluate AI startups:

Key metrics:

1. CAC (Customer Acquisition Cost)

   • How much to acquire 1 customer?
   • AI startups: $5K-50K depending on segment

2. LTV (Lifetime Value)

   • How much revenue from 1 customer over lifetime?
   • Need LTV/CAC ratio of 3:1 minimum

3. Gross Margin

   • Revenue minus COGS (including AI inference costs)
   • Traditional SaaS: 80-90% gross margins
   • AI SaaS: 50-70% gross margins (AI costs eat margin)

4. Burn Multiple

   • How much you burn to generate $1 of ARR
   • Good: <1.5x
   • Okay: 1.5-3x
   • Danger: >3x

Most AI startups I analyze:

• LTV/CAC: 1-2x (need 3x)
• Gross margin: 40-60% (need 70%+)
• Burn multiple: 3-8x (need <2x)

Verdict: Unit economics don’t work at current scale. Need to get to $10M+ ARR to achieve efficiency.

AI Winter or Spring? Here’s My Model

I built a financial model to predict AI startup outcomes:

Assumptions:

• 500 AI startups raised Series A in 2023-2024
• Average raise: $15M at $60M post
• Average burn: $1M/month
• Required ARR for Series B: $3M

Projections for 2026:

Tier 1 (Top 10%): Hit Series B metrics

• 50 startups raise Series B
• Valuations: $150-300M
• These are the “AI Spring” success stories

Tier 2 (Next 20%): Extension rounds

• 100 startups raise extension/bridge rounds
• Flat or down valuations
• Buy 12-18 more months to hit metrics

Tier 3 (Next 30%): Acquihires or fire sales

• 150 startups get acquired for <$50M
• Founders and investors lose money
• Talent joins bigger companies

Tier 4 (Bottom 40%): Shut down

• 200 startups shut down completely
• Zero return for investors
• This is the “AI Winter” for them

So Which Is It?

Both.

• Spring for top 10-20% (real businesses with traction)
• Winter for bottom 40-60% (no path to profitability)

The question isn’t “AI Winter or Spring for the market?”

The question is: “Which tier is YOUR startup in?”

Advice for Founders Reading This

Test 1: Runway

• If you have <12 months runway, raise NOW or cut burn IMMEDIATELY
• Market is slowing, don’t wait

Test 2: Unit economics

• Calculate your LTV/CAC and burn multiple TODAY
• If they’re bad, fix them before raising next round

Test 3: Revenue trajectory

• If you’re not growing 10%+ month-over-month, why not?
• Investors expect exponential AI startup growth

Test 4: Differentiation

• Can customers get 80% of your value from ChatGPT Enterprise?
• If yes, you’re in trouble

For @maya_builds Specifically

Your metrics:

• Profitable in month 3
• Niche focus (Rust developers)
• Capital efficient
• Revenue from day 1

You’re in Tier 1 even without VC funding.

@product_david’s startup:

• $800K/month burn
• $400K ARR (burn multiple: 24x)
• 14 months runway
• Need $2M ARR for Series B (5x growth in 14 months)

That’s Tier 3 or 4 unless things change FAST.

The Uncomfortable Truth

66% of VCs expect exits to increase in next 12 months (source: EY VC survey)

But the data says:

• IPO market still weak
• M&A still constrained
• Time to exit extending

Either VCs are delusional or they know something I don’t.

My bet: They’re being optimistic because they NEED exits to raise their next funds.

Final Answer

AI Winter or Spring?

For the market overall: Spring (real technology, real adoption, real revenue)

For individual startups: Winter for 60%, Spring for 40%

For VC funds: Winter (most funds won’t return capital on 2023-2024 vintages)

For employees: Mixed (top startups hiring, bottom startups laying off)

For customers: Spring (more AI tools, better pricing, competition drives quality)

The bifurcation is happening NOW. By end of 2026, it’ll be clear.

Sources:

• Crunchbase State of Startups Q2 2025
• Bain Global Venture Capital Outlook
• EY Q1 2025 VC Investment Trends
• Deloitte 2025 VC Trends
• Wise Venture Capital Trends 2025
• My proprietary financial models and portfolio analysis