The Startup Legal Brief, Issue #1, August 2026

The Startup Legal Brief: Issue #1

Where the money actually went this year, and which legal changes reach a company your size.

Jason Acevedo | August 14, 2026

Welcome to the first issue. The plan is a quarterly note for founders and investors: what the funding data actually says, the handful of legal changes that reach a company at your stage, and what is worth doing about them. No case summaries, no client alerts dressed up as insight.

The Big Picture: a record half-year that most seed-stage founders did not feel

US startups raised $412.7 billion in the first half of 2026, about 22% more than the $339.4 billion raised in all of 2025, according to the PitchBook-NVCA Venture Monitor. AI companies took $355.9 billion of that, roughly 86 cents of every venture dollar. The second quarter alone produced seven rounds of a billion dollars or more, worth $87.2 billion between them.

Then there is the part that does not make the headline. Financings of $100 million or larger absorbed 87.5% of the capital deployed, which leaves about $51 billion spread across every deal below that line. Crunchbase, counting North America rather than PitchBook's US figures, put seed and angel funding at roughly $4.9 billion in Q2, down 27% from the same quarter last year and down 15% from Q1. Early-stage dollars nearly doubled year over year, but early-stage deal count fell to a five-quarter low, which tells you the money reached fewer companies in larger increments.

The supply side is concentrating too. Andreessen Horowitz, Thrive Capital and Founders Fund together raised 48.1% of all the capital committed to US venture funds in the first half, about $34.8 billion out of $72.4 billion across 405 funds. First-time fund formation is tracking toward its weakest year since 2016. If the emerging manager who wrote your first check has gone quiet, that is a reasonable explanation.

What I take from this, sitting across the table from both sides of these deals: two founders can read the same record-breaking headline and be operating in completely different markets. A company with an AI product and a warm introduction is raising into the best conditions in years. A company with a sound business and no AI story is raising into a market with smaller checks, slower diligence, and an investor who is quietly worried about their own fundraise. Plan for the market you are actually in.

The Legal Spotlight: two changes that reach a company your size

1. Your entity choice now runs a tax clock

Here is a pattern I have seen several times this summer. A founder forms an LLC to keep things simple, builds for a year or two, then converts to a Delaware C-corporation in the middle of a financing. The conversion is routine. Delaware has a statutory procedure, lawyers do them constantly, and it closes.

The cost nobody flags is the calendar. Qualified small business stock under Section 1202 lets founders and early investors exclude gain on a sale, and for stock issued after July 4, 2025, the terms improved considerably: 50% of gain excluded at a three-year hold, 75% at four years, 100% at five, with the per-company cap raised to $15 million and the gross-asset ceiling raised to $75 million. Only C-corporation stock qualifies. The holding period starts when the corporation issues your stock, not when you founded the LLC. Two years spent as an LLC on the way to the same destination is two years of that clock you never get back.

Decide this from your capital plan. If you are raising from investors now, or plausibly within two years, form the Delaware C-corp early enough that the clock works for you. If you are building a business you intend to own and run on its own profits, an LLC is usually the better home, and often in your own state rather than Delaware. The longer version, including what Delaware actually costs each year, is in Delaware C-Corp or LLC for My Startup?

One boundary worth naming here, because it comes up: entity selection is a corporate decision and it is one I will help you make. The tax modeling underneath it sits with your accountant. On a choice this hard to unwind, make that call early rather than during a financing.

2. Your insurance may have quietly stopped covering the AI in your product

This is not a regulation, which is probably why almost nobody is writing about it. It is a change to the forms your carrier hands you at renewal, and it is arriving now.

ISO, which drafts the standard forms most commercial general liability policies are built from, issued generative AI exclusion endorsements carrying January 2026 edition dates. CG 40 47 excludes bodily injury, property damage, and personal and advertising injury arising out of generative AI from a CGL policy. CG 40 48 does the same for Coverage B alone. CG 35 08 reaches products and completed operations. The definition in the forms is wide: a machine-based learning system or model trained on data with the ability to create content or responses, including text, images, audio, video or code.

Some carriers are going further on their own paper. Berkley has an absolute AI exclusion running across D&O, E&O and fiduciary liability that reaches not only AI you deploy but failure to detect AI-generated communications and inadequate AI policies or training. Hamilton excludes claims involving generative AI and names ChatGPT, Bard, Midjourney and DALL-E in the text. How widely any of this has been adopted is genuinely unknown. Verisk said in July that it does not know how many insurers have picked up the ISO forms, so treat any adoption percentage you are quoted as invented.

Two things to do something about. The erosion tends to happen quietly, through revised base forms and narrowed carve-backs rather than an endorsement with an obvious title, and the same incident can be characterized as E&O, D&O, EPLI or cyber, each policy carrying differently worded language. Reading one policy will not find the gap. Then there is the part that catches people off guard: your customer contracts almost certainly require you to maintain E&O or CGL coverage at stated limits. If your carrier attaches an AI exclusion to a product you build with generative AI, you may be sideways with an insurance covenant in an MSA you signed last year, without anyone at either company noticing.

At your next renewal, ask the broker in writing whether an AI exclusion has been added on any line, and read the definition of AI rather than the exclusion itself. A Berkley-style definition captures machine learning you have been running since 2022.

The Deal Radar

Pre-seed is holding steady in dollars and shrinking in deals. Carta counted $3.19 billion across more than 11,500 pre-seed instruments in Q2 2026, against $3.22 billion across 14,825 instruments a year earlier. Nearly identical money, roughly 3,300 fewer deals, with the average instrument up 27% year over year to about $276,000. AI companies took 49% of pre-seed dollars in the first half. Carta also notes that at the 90th percentile, caps on SAFEs above $2.5 million can reach $100 million, with the caveat that getting past $2.5 million usually means ten or more stacked instruments. Useful context the next time someone tells you their cap is just where the market is.

The seed to Series A gap keeps stretching. Blended industry estimates put the gap past 24 months, with a substantial share of seed companies raising an extension before they get to a priced round, usually on a SAFE at flat or a modest step-up. Treat the specific percentages carefully, because the underlying figures are synthesized rather than reported. The planning point is not in doubt. If you raised a seed in 2025, budget for an extension conversation you did not plan on, and read the MFN and pro rata terms in your existing SAFEs before you paper a new instrument. Those provisions are where extensions get expensive, and they get read closely for the first time at exactly the wrong moment.

Two filings got easier. FinCEN published a final rule on August 14 permanently exempting US-formed companies from beneficial ownership reporting under the Corporate Transparency Act. Not a pause and not an interim rule this time, a final one effective on publication. Foreign entities registered to do business here still report, but not as to their US-person owners or applicants. The statute itself is still on the books, so a future administration could revisit this by rulemaking, and it is worth being precise with anyone who tells you the CTA was repealed. It was not. Separately, 83(b) elections can now be filed online through IRS Form 15620, and the system returns a confirmation PDF. That confirmation quietly solves the diligence problem every Series A runs into, which is the founder who cannot produce a certified mail receipt from three years ago. Two caveats: only Form 15620 goes online, not a custom election statement, and it needs a personal IRS account, which is real friction for founders outside the US. The 30-day deadline has not moved and there is still no relief for missing it. Whether the election makes sense for you is a conversation with your accountant. What changed is the filing mechanics, and they got better.

The exit window is open at the very top of the market. SpaceX listed in June, and both OpenAI and Anthropic have reportedly filed confidentially. This is a useful signal about late-stage marks and secondary pricing. It has not yet changed what a Series A investor will pay for your company.

One thing to do this month

Pull every SAFE and convertible note you have outstanding and put them in one table: date, amount, valuation cap, discount, pre-money or post-money, and whether the holder has an MFN clause, a pro rata right, or a side letter.

Most founders have never seen their own instruments side by side. They get signed months apart, often on different templates, and sometimes on a template the investor supplied. An hour of work answers the two questions that decide your next round: how much of the company converts at what price, and who already has a contractual right to something in that financing whether you offer it or not. An MFN clause that looked harmless in isolation reaches forward and pulls the best terms you ever gave anyone into every other instrument carrying one. If the mechanics are unfamiliar, start with Cap Table Mastery and Do I Need a Lawyer for a SAFE?

If an extension or a priced round is anywhere in your fall, build the table before you start negotiating. I have sat in too many rooms where a founding team saw the combined effect of its own paperwork for the first time while a term sheet was on the table.

Get the next issue

The Startup Legal Brief goes out quarterly to founders and investors. Funding data worth knowing, the legal changes that actually reach an early-stage company, and one concrete thing to do about them. No client alerts, no case summaries, and nothing you need to read twice.

Quarterly, nothing else. Unsubscribe whenever you like and I will not chase you about it.

Jason Acevedo is a startup and venture capital partner at Klehr Harrison and the founder of Acevedo Venture Law, counseling founders and investors from formation through exit.

Working through one of these this quarter? Start with a short scoping call. I will tell you honestly what the work involves and roughly what it should run for your situation, including through our startup reduced fee program. Get in touch.

This brief is general information, not legal advice, and it does not create an attorney-client relationship.