AI code assistants: what’s funding like now?

All funding deals in the AI code assistant market over the last 24 months

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

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

Is AI code assistant funding actually healthier now?

AI code assistant funding is much richer at the top today, but the market underneath has weakened.

We counted $3.51 billion of funding over the latest 12 months, up from $1.60 billion in the previous period. Taken alone, that 119.7% increase looks like a huge improvement.

The rest of the picture is less flattering. Deals fell from 11 to 9, while the number of funded companies dropped from 10 to 8. Capital from rounds below $50 million also slipped 5.8%, even as the median financing increased from $20 million to $32 million.

So money is clearly available right now, just much less evenly. Companies that convince investors can raise substantially more than before. Fewer companies are getting through that filter in the first place.

Metric	Latest 12 months	Previous 12 months	Change
Total funding	$3.514B	$1.600B	+119.7%
Deals	9	11	-18.2%
Funded companies	8	10	-20.0%
Median round	$32.0M	$20.0M	+60.0%
Average round	$390.5M	$145.4M	+168.5%
Capital from rounds below $50M	$89.45M	$94.94M	-5.8%
First-financing share	22.2%	27.3%	-5.1 pp
Follow-on share	77.8%	72.7%	+5.1 pp

Would AI code assistant funding still look strong without Cursor and the other huge rounds?

Once we strip out the three biggest AI code assistant rounds, the funding boom disappears.

Cursor raised $2.3 billion, Poolside received a $1 billion equity investment and Code Metal raised $125 million. Together, those three transactions account for 97.5% of all capital we counted in the latest period.

The previous market was concentrated too. Its three largest rounds contributed 94.1% of funding. What changed is the size of the winners. Both periods produced exactly three financings above $100 million, so the surge came from much larger checks rather than a new wave of megarounds.

The cleanest comparison is everything outside those top three deals. We get $89.45 million now versus $94.94 million previously. That turns the apparent 120% funding surge into a 5.8% decline.

Outlier test	Latest 12 months	Previous 12 months
Total funding	$3.514B	$1.600B
Largest round	$2.30B	$900M
Top-three capital	$3.425B	$1.505B
Top-three share of funding	97.5%	94.1%
Capital outside top three	$89.45M	$94.94M
Change outside top three	-5.8%	-
Number of $100M+ rounds	3	3

Are more AI coding companies getting funded these days?

AI coding funding is reaching fewer companies now, even though the companies that raise are getting larger checks.

We found eight funded companies in the latest period, down from ten. The deal count moved in the same direction, falling from eleven to nine.

The contraction also shows up outside the raw company count. New financings became less common, the market covered fewer regions, and six of the latest twelve monthly intervals had no AI code assistant financing at all.

That makes the market narrower than the dollar total suggests. A $3.5 billion year sounds broad until we look at how few companies actually shared it.

Are new AI coding startups still getting funded?

New AI coding startups can still raise today, but fewer newcomers are breaking into the funded market.

We counted two first financings in the latest period, compared with three previously. Their share of all deals fell from 27.3% to 22.2%.

There is a more encouraging detail underneath that decline. Those two companies raised $9.75 million between them, compared with $5.94 million across the three earlier first financings. The median first round nearly doubled from $2.5 million to about $4.9 million.

Kilo Code illustrates the upper end of that change with an $8 million seed round. GitHits raised $1.75 million for its code-context and dependency-search product. Investors are still willing to fund new entrants, but the entry door looks narrower and the successful entrants can start with more capital.

Are the big AI coding winners taking nearly all the money?

Existing AI coding winners are taking almost all the capital available in the market right now.

Follow-on rounds accounted for seven of the nine latest financings and roughly $3.50 billion of the $3.51 billion invested. First financings captured less than 0.3% of the money.

Cursor is the extreme case. After raising $900 million in its previous financing, the company returned for $2.3 billion. Poolside followed a $500 million Series B with a $1 billion NVIDIA equity investment. Code Metal raised $36.5 million and then came back only a few months later for another $125 million.

Once investors believe a company has real momentum, the funding market can let it pull away very quickly. That makes the headline market look extraordinary while doing much less for the broader startup population.

Are normal AI code assistant rounds getting bigger too?

AI code assistant rounds have genuinely moved up in size, even beyond Cursor and Poolside.

The median financing rose 60%, from $20 million to $32 million. Follow-on rounds moved from a $28.5 million median to $36.5 million, while the typical first financing rose from $2.5 million to just under $5 million.

The small end also thinned out. We counted four financings below $5 million in the earlier period and only one this time. Meanwhile, the $5 million to $20 million band grew from one transaction to three.

Investors seem comfortable putting more money into companies they like. The harder part these days is becoming one of those companies.

Is AI coding funding moving toward early-stage startups or mature companies?

AI coding deal activity has moved somewhat earlier-stage, while the money has moved even harder toward mature companies.

Seed and Series A rounds made up two-thirds of the latest deals, up from 54.5% previously. Yet those six early-stage transactions received only 2.5% of all funding.

At the other end, three later-stage financings took 97.5% of the capital. There was also no transaction between $50 million and $100 million in either period.

The market is unusually split. Investors are still writing Seed and Series A checks, then a very small group of companies jumps to enormous nine- or ten-figure financings. The middle is remarkably sparse.

Which parts of the AI coding market are actually gaining momentum?

Code refactoring and modernization show the clearest funding improvement beyond a single giant AI coding deal.

Refactoring tools went from one financing worth $30 million to two totaling $161.5 million. Code Metal drives most of that increase, so we would not describe the whole category as a breakout yet, but both deal count and round size moved in the right direction.

Code-understanding companies also attracted more money. The number of deals stayed at four, while capital increased from $25.94 million to $44.95 million and the median financing more than doubled.

Developer copilots and IDE assistants tell a different story. Their dollar totals exploded because of Poolside and Cursor while their deal counts fell. Anyone looking only at category funding would miss that distinction.

AI coding category	Latest deals	Previous deals	Latest capital	Previous capital
Code Understanding Tools	4	4	$44.95M	$25.94M
Refactoring AI Tools	2	1	$161.5M	$30M
Developer Coding Copilots	2	3	$1.008B	$535M
IDE Code Assistants	1	3	$2.30B	$1.009B
Code Understanding median	-	-	$5.6M	$2.55M
Refactoring median	-	-	$80.75M	$30M
Developer Copilot median	-	-	$504M	$27M
IDE Assistant median	-	-	$2.30B	$105M

Is AI code understanding finally breaking out?

AI code-understanding startups are getting better funded, although deal creation has yet to accelerate.

We counted four code-understanding financings in each period. The improvement came from check size, with current funding reaching almost $45 million.

CoreStory accounts for much of that change through its $32 million Series A. Its pitch goes well beyond answering questions about code: the company extracts business logic and architecture from legacy software so developers and AI agents can work with complex systems more reliably.

Nozomio approaches the same underlying problem from the coding-agent side. Nia indexes repositories and documentation so agents such as Cursor and Cline receive better context. GitHits is attacking another version of the problem by giving coding agents access to dependency code, documentation and implementation examples.

These companies are all betting on a similar idea: generating code has become easier, while understanding the right context remains difficult. Funding has started to reflect that shift.

Why is AI code refactoring suddenly getting so much attention?

AI code refactoring is pulling in more money because enterprises have expensive legacy-code problems and are starting to pay for tools that can change those systems safely.

Code Metal gives us the strongest example. When the company announced its $36.5 million Series A, it said it had already secured eight figures of annual contracted revenue and had deployed across defense, automotive and semiconductor environments. A $125 million Series B followed only a few months later.

Moderne provides another version of the same thesis. Its $30 million Series B centered on large-scale code modernization and technical-debt remediation rather than simply helping developers write fresh code faster.

CoreStory pushes further into legacy-system understanding, where undocumented business logic can make modernization risky before anyone even starts rewriting software.

Investors are paying more attention to this part of AI coding because the customer problem is already huge and easy to understand. Banks, industrial companies, governments and other large organizations own enormous software estates that cannot simply be replaced because a new coding agent writes JavaScript quickly.

Do AI coding startups need more proof to raise a big round now?

The funding bar for large AI coding rounds looks higher today, with commercial proof showing up much more clearly around the companies raising the biggest checks.

Cursor offers the clearest example. The company reported more than $100 million in recurring revenue around its $105 million Series B, more than $500 million in ARR when it raised $900 million, and more than $1 billion in annualized revenue by the time of its $2.3 billion Series D. Millions of developers were using the product by then.

Code Metal paired its rapid fundraising with contracted revenue and deployments across industries where software failures carry serious costs. CoreStory's $32 million round also came with enterprise deployments and measurable claims around modernization work.

We cannot turn those examples into a clean market-wide percentage because milestones were not consistently disclosed for every company. Still, the companies absorbing the most capital today generally arrive with much more than a promising demo.

Are AI coding companies raising again faster?

Some AI coding companies are returning for much larger rounds within months, but the available sample is too small to claim that fundraising has broadly accelerated.

Reliable previous-round dates are available for only four of the seven latest follow-ons. Their median gap was roughly five months, compared with 4.7 months for the two usable observations in the earlier period. Statistically, that tells us very little.

The individual cases are more interesting. Code Metal moved from its $36.5 million Series A to a $125 million Series B in roughly three months. Cursor and Nozomio also returned quickly after earlier financings.

Rapid repricing is clearly possible when traction moves fast. We simply do not have enough comparable rounds to say the whole AI coding market is fundraising faster.

Are NVIDIA and other corporate investors becoming more important in AI coding?

Corporate investors are much more involved in AI coding funding now, especially around models, compute and mission-critical software.

Strategic investors appeared in four of the nine latest deals, up from two of eleven previously. Their participation rate therefore rose from 18.2% to 44.4%.

The profile changed as well. Strategic participation had previously appeared in two Series B deals. We now see corporate capital across Series A, Series B, Series D+ and Growth Equity.

NVIDIA sits at the center of that shift. The company joined Cursor's financing and later made a $1 billion equity investment in Poolside alongside a separately reported $6 billion licensing transaction. Google also invested in Cursor. RTX and Bosch backed Code Metal, whose customers include companies operating in the same industrial and defense markets.

At the same time, the market has not collapsed into a tiny circle of repeat VC firms. We counted 14 distinct disclosed lead investors in the latest period, versus 11 previously. The capital is concentrated by company much more than it is by investor name.

Investor signal	Latest 12 months	Previous 12 months
Strategic-backed deals	4	2
Strategic participation rate	44.4%	18.2%
Change in participation	+26.3 pp	-
Stages with strategic participation	4	1
Unique disclosed lead investors	14	11
Lead investors appearing more than once	1	1
Main repeat lead	Accel	Thrive Capital

Is AI code assistant funding becoming more US-centric?

AI code assistant funding is more US-centric today, particularly when we look at where funded companies are based.

Eight of the nine latest deals came from North America. The United States alone accounted for 88.9% of transactions, up from 54.5% in the previous period.

Europe still appears through GitHits, while Asia-Pacific and the Middle East disappear from the current set. The earlier period included CodeKarma in India, Tabnine in Israel and several European companies.

Dollar concentration changes much less because North America was already taking almost all the capital one year earlier. What has narrowed is the geography of companies getting checks.

With only nine deals, we would not treat one year of data as evidence that Asian or Middle Eastern AI coding startups have permanently lost investor interest. For now, though, the financing pipeline is plainly more US-heavy.

Is AI code assistant funding arriving steadily throughout the year?

AI code assistant funding has become far more stop-start, with huge bursts of capital separated by long quiet stretches.

Half of the latest twelve monthly intervals had no financing at all, compared with four previously. The median amount invested per monthly interval fell from about $16.4 million to less than $1 million.

The average went in the opposite direction and climbed to almost $293 million a month. Cursor, Poolside and the other large rounds pull that figure sharply upward.

For anyone trying to understand how much capital is routinely available to an AI coding startup, the median gives a better feel for the current market than the average. Funding these days arrives in clusters rather than through a steady flow of transactions.

Are AI coding investors playing it safer now?

AI coding investors look more selective about which companies they fund, while becoming more aggressive once a company clears the bar.

Several pieces fit together here. Fewer companies received financing. First-time rounds became less common. The median age of funded companies rose from one year to three years, and very small financings became rarer.

At the same time, investors did not respond by shrinking every check. The median round increased, rapid follow-ons remained possible, and proven companies attracted extraordinary amounts of capital.

The pattern looks like a higher threshold rather than a broad retreat from risk. Investors appear willing to take very large financial bets when revenue, usage, strategic value or customer adoption gives them enough conviction.

What are AI coding companies actually raising money for now?

The biggest AI coding companies are increasingly raising money to scale research, proprietary technology and commercial execution at the same time.

Cursor's earlier Series B was framed around expanding the team and investing in systems, models and product research. By the Series D, the company was already operating at much greater commercial scale and said the new capital would deepen its research work further.

Code Metal's $125 million Series B had a broader operating agenda: more engineering capacity, faster product development, expanded commercial and government partnerships, and a larger go-to-market operation.

Smaller and earlier companies still look more product-heavy. Nia's earlier financing focused on growing the technical team, improving the agent and expanding access.

There is no clean market-wide use-of-funds percentage because many companies disclose only broad descriptions. The pattern we can see is a lifecycle shift among the winners: research remains central, but the largest financings now have to support customers, distribution and organizational scale alongside the technology.

What really changed over the last 12 months?

Priority	What changed	What it means for AI code assistant funding
1	Headline funding more than doubled while underlying sub-megaround capital slipped	Looking only at total dollars badly overstates the improvement
2	Funding reached fewer companies	Access to capital became narrower despite the much larger total
3	Existing winners captured almost all current dollars	The biggest beneficiary of today's market is the proven company coming back for another round
4	Completed financings became larger	Investors are putting more money behind companies that pass a tougher selection process
5	Refactoring, modernization and code understanding improved	The market is expanding beyond pure code generation toward context, legacy software and verifiable changes
6	Strategic investors became much more visible	Compute providers, customers and technology partners increasingly have reasons to invest directly
7	Funding became more episodic	A few enormous financing windows now shape annual totals far more than steady deal activity

Overall, AI code assistant funding looks more mature today, but also more concentrated and harder to enter. The strongest companies can raise sums that would have looked extraordinary one year earlier, while the wider market has not received more conventional venture capital.

The misleading headline would simply be that AI code assistant funding rose 120%. The more useful conclusion is that a handful of proven winners raised dramatically more money while fewer companies got funded overall.