AI Code Assistant Funding Tracker (32 deals)

Our funding tracker covering the AI code assistant market

Below is the table listing all the deals. You may find our methodology at the end of this page.

And, if you want to gain a deeper understanding of the market and its current dynamics, get our report covering the AI Coding Market.

Is AI code assistant funding really a $6 billion market today?

As of today, AI code assistant funding does add up to almost $6 billion, but that headline makes the average company look far richer than it really is.

We tracked 32 financings across 26 AI code assistant companies from January 2024 through the end of our research window. Together, those rounds represent $5.98 billion.

Then we removed the nine financings above $100 million. The market fell to roughly $350 million. Around 94% of all the money disappeared.

That gap tells us more about AI coding funding than the $5.98 billion headline does. Seed and Series A activity exists across a fairly broad group of startups, while a tiny upper layer has access to hundreds of millions or even billions of dollars.

The median financing was $16.7 million. The average was $186.8 million, more than eleven times higher. Anyone looking only at aggregate capital or average round size would come away with a very distorted idea of what companies in this market usually raise.

How many AI code assistant companies are actually getting funded?

Today, the funded AI code assistant market we can identify consists of 26 companies, and most of them have appeared only once.

Twenty-two of those 26 companies recorded a single qualifying financing during the period. Three appeared twice. Cursor appeared four times.

That means 84.6% of funded companies have only one round in our research window.

The market has a slightly unusual shape. There is plenty of experimentation at the bottom: new copilots, repository-intelligence products, refactoring tools and specialized coding assistants keep getting funded. Very few have built a visible sequence of increasingly large rounds.

The difference gets clearer when we follow the money. The four companies that appear more than once account for roughly $5.24 billion of the $5.98 billion raised.

A reader seeing 32 financings might imagine a fairly broad race. Looking at the 26 companies behind those rounds, and especially at how rarely the same startup comes back for another financing, makes the field look much narrower.

Is AI code assistant funding basically Cursor and Poolside?

Cursor and Poolside have absorbed roughly 81% of all the AI code assistant funding we tracked, so the answer is uncomfortably close to yes.

Cursor alone raised $3.365 billion across four rounds in the period. That equals 56.3% of the entire market.

Poolside adds another $1.5 billion through its $500 million Series B and the later $1 billion equity investment included in our research. Together, the two companies represent about four dollars out of every five invested across the market.

After them, there is a steep drop. Magic raised $320 million, Augment Code $227 million, Codeium/Windsurf $215 million and Code Metal $161.5 million.

By the time we reach the tenth-largest company, cumulative capital is down to $20 million.

The top five companies hold 94.1% of the capital. Company-level concentration is even more revealing than individual mega-rounds because the giant checks keep returning to a small group of businesses.

Company	Financings tracked	Capital raised
Cursor	4	$3.365B
Poolside	2	$1.500B
Magic	1	$320M
Augment Code	1	$227M
Codeium / Windsurf	2	$215M
Code Metal	2	$161.5M
CoreStory	1	$32M
Moderne	1	$30M
Cline	1	$27M
Unblocked	1	$20M

What does a normal AI code assistant funding round look like?

A normal AI code assistant financing today is much closer to a $10–20 million round than to the giant checks dominating funding headlines.

The median deal is $16.7 million. Half of the 32 rounds are below $20 million, and those 16 financings collectively account for only around 1.3% of the capital.

Ten rounds are below $5 million. Another six fall between $5 million and $20 million. Five sit between $20 million and $50 million.

Then the distribution changes abruptly. Eleven rounds exceed $50 million, and those deals capture 96.2% of the money.

Company-level numbers point in the same direction. The median company raised $8 million during the period, compared with an average of almost $230 million.

For founders, operators or investors trying to benchmark an ordinary startup, the median figures are far more useful. A company raising $3 million, $8 million or $20 million is much closer to the center of this market than one raising $300 million.

Deal size	Number of deals	Share of deals	Capital raised
Under $5M	10	31.3%	$26.5M
$5M–under $20M	6	18.8%	$52.7M
$20M–under $50M	5	15.6%	$145.5M
$50M–under $100M	2	6.3%	$125M
$100M+	9	28.1%	$5.627B

Which part of the AI coding market is getting the serious money?

AI coding IDEs and broad developer copilots currently get almost all the serious funding, even though plenty of startups are being financed elsewhere.

IDE Code Assistants attracted $3.436 billion, or 57.5% of the capital. Developer Coding Copilots added $2.257 billion, or 37.8%.

Together, those two categories account for roughly 95% of the dollars.

The deal counts are much less extreme. IDE assistants represent seven rounds. Developer copilots and Code Understanding Tools both recorded ten.

That comparison is particularly useful. Code Understanding produced exactly as many financings as Developer Coding Copilots, yet it attracted $77.4 million instead of $2.26 billion.

The median Code Understanding round was only $3.35 million. Developer Coding Copilots reached a median of $88.5 million, while IDE assistants reached $65 million.

Investors are clearly willing to fund repository search, codebase intelligence and contextual tools. We have not yet seen many of those businesses graduate into the same financial class as Cursor, Poolside, Magic or Augment.

AI coding category	Deals	Companies	Capital	Median round
IDE Code Assistants	7	4	$3.436B	$65M
Developer Coding Copilots	10	9	$2.257B	$88.5M
Refactoring AI Tools	4	3	$194.7M	$33.3M
Code Understanding Tools	10	10	$77.4M	$3.35M
Inline Code Completion	1	1	$12M	$12M

Why can AI coding IDEs and copilots raise so much more money?

The biggest AI coding companies can raise enormous checks because some of them now combine software economics with the cost structure of AI model companies.

Cursor is the clearest commercial example. When it announced its $2.3 billion Series D, the company said it had crossed $1 billion in annualized revenue, had millions of developers using the product and planned to invest heavily in its own research and models.

That gives investors something very different from an early developer-tool pitch. They are financing an already large software business while also paying for model research, inference and infrastructure.

Poolside shows the compute-heavy version. When the company raised $500 million, it said the financing had allowed it to scale its training cluster to 10,000 GPUs. Poolside described two races happening at once: improving model capabilities through talent, compute, research and data, while simultaneously building a commercial product and distribution engine.

Magic was even more striking. Around its $320 million financing, the company said it had only 23 people but 8,000 H100 GPUs, with plans to move toward tens of thousands of Blackwell GPUs.

Augment shows that large checks can arrive before a long public operating history exists. The company emerged from stealth with a $227 million Series B and $252 million of total funding.

Once proprietary models and very large compute budgets enter the equation, the amount of capital a coding company can plausibly consume changes quickly. That helps explain why the funding gap between the top platforms and a normal code-intelligence startup has become so large.

Why is AI code understanding getting plenty of deals but small checks?

AI code understanding currently looks like a healthy startup-creation category with an unproven path to large-scale funding.

We found ten financings in the category but only $77.4 million of capital.

The company examples make the pattern easier to see. GitHits raised $1.75 million. CodeKarma raised $2.5 million. Lightbase raised roughly $2.6 million. Greptile raised $4.1 million. Nozomio raised $6.2 million.

At the upper end, Unblocked reached a $20 million Series A and CoreStory raised $32 million.

The common need behind these products is easy to understand. Developers and coding agents need better context from large repositories, old systems, dependencies, documentation and internal architecture. As AI agents write more code, giving them accurate context may become even more valuable.

Funding has still stopped early. We found no Code Understanding Series B or later round in the entire period.

For now, investors seem comfortable financing the problem but have yet to show that they will repeatedly finance independent code-understanding companies at platform scale. That could change, especially if context becomes a critical infrastructure layer for coding agents, but the large rounds have not appeared yet.

Is AI code assistant funding actually accelerating right now?

AI code assistant funding is producing bigger dollar spikes, but the number of financings has remained surprisingly modest.

Looking across the 32 complete months in our monthly analysis, we get roughly one deal per month on average. The median is also one.

The busiest month contained only three deals.

Capital behaves very differently. A normal month had a median of roughly $8 million in funding, while the biggest month reached about $2.34 billion from only two financings.

Cursor’s $2.3 billion Series D explains almost all of that peak.

This is why funding headlines can make the AI coding market feel as if it suddenly exploded. One huge round can multiply monthly capital without bringing many additional companies into the market.

The underlying deal cadence looks much steadier. What has changed most dramatically is how much investors are prepared to give the few companies that break away from the pack.

How young are AI code assistant startups when they raise money?

AI code assistant startups are raising serious money extremely young, with a median age of only two years at financing.

The average is about 2.45 years.

Seed rounds naturally arrive fastest: the median company is effectively in its founding year. Series A, Series B and Series C companies all have a median age of roughly three years in our research.

Even rounds above $50 million occur at a median company age of only three years.

Cursor shows how compressed the timetable can become. Founded in 2022, it raised a $60 million Series A in 2024, followed by $105 million, $900 million and $2.3 billion rounds within a relatively short period.

Poolside, founded in 2023, reached a $500 million Series B the following year.

The broader cohort is young as well. Most of the activity comes from companies founded after 2021, with particularly strong representation from the 2022–2025 generation.

Investors in this market are making very large bets before these businesses have anything close to the operating history normally associated with late-stage software financing.

Why do so few AI coding startups make it past Series B?

AI coding startups start disappearing quickly after Series B, and the few that continue tend to raise dramatically larger checks.

We counted 12 Seed rounds, seven Series A rounds and six Series B rounds.

From there, the ladder drops to three Series C financings and one Series D+ round. Poolside’s later $1 billion investment sits separately as Growth Equity because its structure does not fit neatly into the usual venture ladder.

The median tickets tell the other half of the story. Seed sits at $3.15 million. Series A jumps to $27 million. Series B reaches $115 million. Series C reaches $320 million.

By that point, we are dealing with a tiny population of companies.

Early-stage rounds make up about 63% of the classified financings but receive only 4.1% of the capital. Later-stage rounds take nearly everything else.

The funding ladder gets selective at roughly the same moment that building frontier coding products becomes much more expensive. Companies that make it through Series B can start needing capital for proprietary models, compute, international enterprise distribution or highly specialized engineering.

Stage	Deals	Capital	Median round
Seed	12	$50.7M	$3.15M
Series A	7	$194M	$27M
Series B	6	$1.052B	$115M
Series C	3	$1.370B	$320M
Series D+	1	$2.300B	$2.300B
Growth Equity	1	$1.000B	$1.000B
Unknown	2	$10M	$5M

Are investors still backing new AI coding startups?

Investors are still backing new AI coding startups, but fresh entrants receive very little of the market’s total capital.

We classify nine of the 32 rounds as first financings. Together, they raised about $36 million.

The remaining 23 are follow-on financings representing approximately $5.94 billion.

The typical checks are far apart too. A first financing has a median size of $2.59 million, compared with $36.5 million for a follow-on.

The flow of new startups has not stopped. ChipAgents, CodeKarma, Kilo Code, GitHits, Lightbase, Nia, Recurse ML, Refact.ai and Supermaven all entered our financing window through first rounds.

The bigger divide appears after that initial backing. Only four companies show multiple qualifying rounds during the period, yet those repeat fundraisers absorb 87.7% of the capital.

The market keeps creating new entrants. The much rarer event is a startup convincing investors to come back quickly with a much larger check.

Is AI code assistant funding basically an American market?

AI code assistant startups exist outside the United States, but the scale capital is currently overwhelmingly American.

North America accounts for 25 of the 32 financings and roughly $5.96 billion of capital. The United States alone contributes 24 deals across 18 companies.

Europe has five funded companies in our research, including Recurse ML, Refact.ai, Nia, Lightbase and GitHits. Their combined funding comes to only about $10.4 million.

The Middle East contributes Tabnine’s $8 million financing. Asia-Pacific contributes CodeKarma’s $2.5 million round.

No company headquartered in Europe, the Middle East or Asia-Pacific records a round above $50 million during the period.

North America therefore represents around 78% of the deal count and 99.65% of the dollars.

The international startup layer is real. What we have not found outside North America is an equivalent to the American scale-up layer where coding companies receive repeated nine-figure financings.

For now, creating an AI coding startup is a global activity. Financing one at Cursor, Poolside or Magic scale remains heavily concentrated in the United States.

Which investors keep coming back to AI coding companies?

Very few lead investors keep appearing across AI coding deals, despite the amount of money flowing into the market.

We could identify a lead investor on 90.6% of the financings and counted 35 different leads.

Only three appear more than once: Accel, Thrive Capital and LocalGlobe.

Thrive is the clearest example of sustained backing. Its three appearances are all tied to Cursor, spanning the company’s Series A, B and C.

Accel also appears three times across the financing set, while LocalGlobe’s two appearances sit at the much smaller early-stage end.

Elsewhere, the market draws money from a broad collection of firms including Andreessen Horowitz, Bain Capital Ventures, General Catalyst, Kleiner Perkins, Bessemer Venture Partners, Emergence Capital, Acrew Capital and Salesforce Ventures.

We see more continuity around specific companies than around one specialist group of VCs underwriting the whole AI coding category.

That fits a market this young. Investors clearly want exposure, but a settled group of repeat category specialists has not really formed yet.

Do NVIDIA and other strategic investors really matter in AI coding funding?

Strategic investors show up in fewer than one-third of AI coding rounds, but they are much more common around the big checks.

We identified strategic participation in 28.1% of financings.

Those rounds have a median size of $125 million. Deals without a strategic investor have a median of $8 million.

NVIDIA is the most obvious recurring name. It appears around Cursor and Poolside, two companies for which model training and inference capacity are central to the product.

The relationship with Poolside is especially easy to understand. The company was already scaling toward a 10,000-GPU cluster around its $500 million Series B. NVIDIA later led the $1 billion equity investment included in our financing research.

Strategic money also appears where the value is industrial rather than compute-driven. Code Metal attracted RTX and Bosch Ventures while building verified AI code translation for defense, automotive and semiconductor environments. Moderne has Intel Capital among its investors while working on large-scale enterprise code modernization.

Code Metal gives us a useful example of how far that specialized path can go. The company moved from a $36.5 million Series A to a $125 million Series B within a few months, with software aimed at environments where a bad code transformation can create much larger consequences than a poor autocomplete suggestion.

Strategic investors therefore appear most relevant when an AI coding product depends on expensive infrastructure or sits deep inside an industry where technical validation and customer access are especially valuable.

What is still missing from AI code assistant funding today?

The biggest gap in AI code assistant funding today is depth: we can see many early experiments and a few huge winners, with surprisingly little in between.

Later-stage financing remains thin. Beyond Series B, only a handful of companies appear.

Code Understanding is the clearest category gap. Ten companies received funding, yet none reached Series B during our window.

Geography creates another obvious hole. Several European startups are present, but we found no large European scaling round. The same applies to the Middle East and Asia-Pacific.

Repeat fundraising is rare too. More than four-fifths of the companies appear only once.

Inline Code Completion barely exists as a standalone funded category in this research, with only Supermaven’s $12 million financing.

There is also very little evidence of a broad middle class of heavily funded businesses. Once we move past the small-round population, capital rapidly concentrates into Cursor, Poolside, Magic, Augment, Codeium/Windsurf and Code Metal.

Some of these gaps may narrow as the market matures. For now, though, the financing record looks much more like an early winner-selection process than a deep, established software category.

What can a normal AI coding startup realistically raise today?

A normal AI coding startup today should think in single-digit millions at Seed and tens of millions at Series A, rather than benchmarking itself against the spectacular rounds in the headlines.

Our Seed median is $3.15 million. Series A is $27 million.

That range covers a large part of the actual market. Recent examples include GitHits at $1.75 million, CodeKarma at $2.5 million, Greptile at $4.1 million, Nozomio at $6.2 million, Kilo Code at $8 million, Supermaven at $12 million, Unblocked at $20 million and Cline at $27 million.

Crossing into Series B changes the benchmark sharply. The Series B median in our research is $115 million, largely because the companies reaching that stage include businesses such as Cursor, Poolside, Augment and Code Metal.

Those are difficult comparables for an ordinary startup. Their fundraising stories involve unusually strong commercial traction, proprietary-model ambitions, heavy compute requirements, strategic industrial relevance or some combination of the four.

For most founders, $2–5 million is a far more credible first reference point than $100 million. A $20–30 million institutional round is already a meaningful step up. Once investors start writing nine-figure checks, the company has usually moved into a much smaller group with a very different cost base or strategic value.

What should we actually take away from AI code assistant funding?

AI code assistant funding is already a multibillion-dollar market, but the breadth of that funding is much smaller than the headline suggests.

We found a large experimental layer where startups can still raise a few million dollars around coding agents, code understanding, refactoring and developer workflows. New companies continue to enter the market, and the Seed pipeline remains active.

Serious capital is much more selective. IDEs and broad coding copilots absorb roughly 95% of the money because the leading companies increasingly combine software distribution with model research and expensive compute.

Cursor and Poolside dominate the financing picture, with roughly 81% of total capital between them. Remove the biggest rounds and the market looks much closer to conventional venture software.

The companies receiving those huge checks are also remarkably young. Investors have been willing to make billion-dollar capital decisions around businesses only a few years old, which shows how aggressively the market is trying to identify long-term winners.

Series B currently looks like the decisive part of the ladder. Plenty of startups get financed before it. Very few keep climbing afterward, and the ones that do start raising at a completely different order of magnitude.

Geographically, the same divide appears again. Startup creation reaches Europe, the Middle East and Asia-Pacific, while almost all of the scaling capital stays in North America.

The clearest way to read the market today is as a selection race. Investors are still funding many ideas, but they are concentrating extraordinary amounts of money behind a very short list of companies they believe could own major parts of how software gets built.