What are the current funding trends in AI coding?

All funding deals in the AI code assistant market over the last years

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 really up in 2026?

Yes, AI code assistant funding is up slightly in 2026, but that headline hides a much weaker market underneath.

We calculated $1.132 billion of funding from January 1 to September 21, up 3.3% from $1.096 billion over the same period in 2025. Compared with 2024, the increase is larger at 31.7%.

Now look at who actually received that money. Deal count fell from nine last year to four today. The number of funded companies was cut in half, from eight to four. And once rounds above $50 million are removed, funding drops from $90.9 million in 2025 to just $6.75 million this year.

Capital is still available in huge amounts for a very small number of companies, while ordinary fundraising activity has become much thinner.

The comparison with 2024 makes the contrast even sharper. By September 21, current funding had already reached 82.8% of the entire 2024 annual total, despite only four deals this year versus fourteen across all of 2024. A lot of money is moving, but very few companies are receiving it.

Funding measure	2026 YTD	2025 YTD	2024 YTD	Change vs 2025
Total funding	$1.132B	$1.096B	$859.2M	+3.3%
Deals	4	9	11	-55.6%
Unique funded companies	4	8	10	-50.0%
Funding excluding >$50M rounds	$6.75M	$90.9M	$37.2M	-92.6%
Median round	$65.0M	$20.0M	$13.5M	+225%
Largest round	$1.000B	$900M	$320M	+11.1%
First financings	1	3	3	-66.7%
Top three share of funding	99.85%	94.4%	81.1%	+5.45 pp

Is Poolside making AI coding funding look much stronger than it really is?

Absolutely. Poolside's $1 billion financing represents 88.4% of all the AI code assistant funding we tracked this year.

Add Code Metal's $125 million Series B and two companies capture 99.4% of the total. Everything else comes down to two rounds worth a combined $6.75 million.

The headline is unusually distorted. Removing only Poolside leaves $131.8 million of funding this year. Doing the same exercise with the biggest round in 2025 leaves $195.9 million, while removing 2024's largest transaction leaves $539.2 million.

Poolside also makes the average round look bigger, pushes Developer Coding Copilots to the top of the category tables and sends North America's capital share close to 100%.

The concentration has become more extreme over time. The largest round represented 37.2% of comparable funding in 2024, 82.1% in 2025 and 88.4% currently. The top three transactions now absorb 99.85% of the market.

Calling 2026 a broadly stronger funding year would miss what has actually happened. We are looking at an exceptional Poolside transaction sitting on top of a much smaller financing base.

Are fewer AI coding startups getting funded now?

Yes. Far fewer AI coding startups are getting funded today, and this contraction is much bigger than the 3.3% increase in headline capital suggests.

Only four companies raised during the current period. There were eight in the same part of 2025 and ten in 2024.

The investor base narrowed with them. We counted four unique disclosed lead or co-lead investors this year, down from nine in 2025 and twelve in 2024. No current lead appears on more than one transaction.

Geographic breadth has shrunk too. Last year's comparable sample included companies from North America, Europe, Asia-Pacific and the Middle East. This year, only North America and Europe appear.

None of those individual changes would tell us much with such a small sample. Taken together, though, they point in the same direction: fewer startups are finding their way into funded rounds, and fewer investors are showing up as leads.

That is the part of the 2026 market a founder will probably feel most directly. A billion-dollar financing can push an industry total higher without making fundraising easier for everyone else.

Are AI code assistant rounds actually getting bigger?

The $65 million median round makes AI coding financings look much bigger today, but the underlying four deals do not support a broad round-size boom.

The current financings are $1.75 million for GitHits, $5 million for Adronite, $125 million for Code Metal and $1 billion for Poolside. Half the companies raised $5 million or less. The other half raised at least $125 million.

That is why the median lands at $65 million even though nobody raised anything close to $65 million. With an even-numbered sample, it simply falls halfway between Adronite and Code Metal.

The average is even less representative at $282.9 million because Poolside carries so much weight.

So, yes, the reported median rose from $20 million to $65 million and the average more than doubled. For a founder trying to understand today's fundraising environment, those numbers are close to useless without the distribution behind them.

What has genuinely changed is the distance between the small rounds and the giant ones.

Where did the $20 million to $50 million AI coding rounds go?

The middle of the AI code assistant funding market has disappeared for now.

Last year's comparable period included Cline at $27 million, Moderne at $30 million and Unblocked at $20 million. Those deals created a real institutional middle between seed rounds and Cursor's huge financings.

There is nothing comparable in 2026. After Adronite's $5 million Series A, the next transaction is Code Metal at $125 million.

That gap is worth watching because $20 million to $50 million rounds often finance companies that have moved beyond the first product but have not yet become obvious category winners. Today, our data jumps almost directly from early-stage checks to late-stage-scale capital.

The comparison with 2024 adds another wrinkle. That year had no $20 million to $50 million rounds in the comparable period either, but there were two transactions between $50 million and $100 million and three above $100 million. The top end was much broader than it is today.

Deal-size bucket	2026 deals	2025 deals	2024 deals
Below $5M	1	3	4
$5M to <$20M	1	1	2
$20M to <$50M	0	3	0
$50M to <$100M	0	0	2
$100M+	2	2	3
Deals below $50M	2	7	6
Deals above $50M	2	2	5
Total deals	4	9	11

Are investors still backing new AI coding startups?

New AI coding startups are still getting funded, but almost none of the money is going to them this year.

GitHits is the only financing we classify as a first financing in the current period. Its $1.75 million round represents roughly 0.15% of all the capital we tracked.

There were three first financings during the same part of 2025, worth $5.94 million combined, and three in 2024, worth $17.21 million.

Meanwhile, the three follow-on financings in 2026 attracted about $1.13 billion.

The company-age figures add some useful nuance here. The median funded company is around three years old this year, the same as in 2025 and only one year older than in 2024. We therefore do not see a convincing shift toward old companies in the literal sense.

What has changed more clearly is financing history. Investors are putting the biggest checks into companies that have already raised and already built some evidence around their technology, customers or strategic value.

We treat first-financing status as directional rather than as a perfect reconstruction of every company's lifetime funding history. Even with that caveat, the difference between $1.75 million of first-financing capital and roughly $1.13 billion of follow-on capital is hard to dismiss.

Are early-stage AI coding startups having a harder time raising money?

Early-stage AI coding startups still account for half of this year's deals, but they are receiving almost none of the money.

GitHits and Adronite are the two Seed/Series A transactions. Together they raised $6.75 million, or around 0.6% of current capital.

Early-stage companies received 4.8% of comparable funding in 2025 and 11.1% in 2024.

Deal share tells a much calmer story. Early-stage rounds represented 54.5% of deals in 2024, 55.6% in 2025 and 50% today. Young companies have not vanished from the transaction count.

The financial gap has become much wider, though. A startup can still get a small early round while virtually all the big money flows elsewhere.

That is a more useful description of today's early-stage market than saying investors have abandoned seed or Series A. We still see those financings. We just do not see much capital behind them.

Which parts of the AI coding market are investors actually funding now?

Developer Coding Copilots dominate the money today, while Code Understanding Tools dominate the number of deals.

Poolside gives Developer Coding Copilots 88.4% of current capital from only one transaction. Code Understanding Tools account for two of the four deals but just 0.6% of funding. Code Metal gives Refactoring AI Tools another 11%.

That mix changed dramatically from last year, when IDE Code Assistants absorbed 91.7% of comparable capital. Cursor alone raised $105 million in January and $900 million in June.

Calling that a major rotation from IDE assistants into copilots goes too far. The companies creating the huge rounds changed, and the category tables moved with them.

The two-year view makes this clearer. Developer Coding Copilots were already a major capital destination in 2024 through companies such as Augment Code, Magic, Pieces and Codeium/Windsurf. IDE Code Assistants then surged in 2025 because of Cursor. Poolside has now pulled the capital center back toward copilots.

The broadest transaction activity is happening somewhere else. Code Understanding represented 18.2% of comparable deals in 2024, 44.4% in 2025 and 50% currently. That trend has survived several changes in the identity of the year's biggest winner.

Why are Code Understanding startups getting deals but such small checks?

Code Understanding has quietly become the most consistent source of AI coding deals, yet investors are still writing relatively small checks into the category.

Adronite raised $5 million this year and GitHits raised $1.75 million. Together they represent half of all current transactions and less than 1% of capital.

The same mismatch appeared before. Code Understanding reached 44.4% of comparable 2025 deals while taking only 2.4% of funding. In 2024, it represented 18.2% of deals and 0.8% of capital.

The products themselves solve increasingly important problems. Adronite focuses on understanding entire codebases rather than isolated files or snippets. GitHits is building a version-aware code and dependency context layer so coding agents can inspect open-source implementations instead of relying on incomplete model memory.

Those are useful capabilities as AI coding tools work across larger repositories and more complicated software systems. Investors clearly keep funding companies around this problem.

What we have not seen yet is a breakout financing on the scale of Cursor, Magic, Poolside or Code Metal. For now, Code Understanding looks like a broad company-formation theme rather than a major capital sink.

Why did Nvidia put $1 billion into Poolside?

Nvidia's Poolside investment is unusually strategic, which helps explain why it should not be treated like an ordinary $1 billion venture round.

The equity investment was announced alongside a separate $6 billion non-exclusive licensing arrangement for Poolside's model-development technology. More than 100 Poolside employees were also reportedly set to receive job offers from Nvidia.

We count only the $1 billion equity investment as funding. Combining the licensing agreement with the financing would badly overstate venture activity.

The wider relationship is still important for understanding the size of the check. Nvidia is deeply involved in the computing infrastructure used to train and run AI models, while Poolside has been building models and developer systems around software engineering. The transaction goes well beyond a financial bet on another coding assistant.

This is also why Poolside tells us relatively little about what a normal AI coding company could raise today. Nvidia had strategic reasons to value Poolside's technology, talent and model-building capabilities that do not apply across the whole startup market.

Why could Code Metal raise $125 million so quickly?

Code Metal raised $125 million because it is selling AI coding into expensive, mission-critical software problems where customers care about verification as much as speed.

The Series B came only a few months after a $36.5 million Series A. Salesforce Ventures led the new round, with investors including Accel, B Capital, Shield Capital and RTX.

Code Metal says customers include RTX, L3Harris, Toshiba and the U.S. Air Force. Its software translates and optimizes code across programming languages and hardware environments, with an emphasis on proving that the output remains correct enough for production use.

That puts Code Metal in a different part of the AI coding market from a general-purpose autocomplete tool. Defense systems, semiconductor environments, automotive software and old enterprise codebases often contain software that is expensive to rewrite and risky to modify.

Salesforce Ventures made that point explicitly when explaining the investment. Its thesis centered on decades of legacy C, C++, MATLAB and Ada code, the move toward GPUs and edge computing, and the difficulty of modernizing those systems manually.

The $125 million round therefore looks tied to a very specific combination of technical difficulty, customer value and strategic urgency. It also helps explain why Refactoring AI Tools can capture 11% of this year's capital from a single company.

Are corporate investors becoming more important in AI coding funding?

Corporate investors are playing a much bigger role in the largest AI coding rounds this year.

Nvidia is directly involved in Poolside and RTX participates in Code Metal. Those two financings represent $1.125 billion, or almost the entire current market.

Last year, Intel Capital participated in Moderne's $30 million Series B. In 2024, our research also found strategic participation around several companies, including Cursor and Magic.

The striking difference today is the size of the transactions where corporates are involved. The median strategic-participation round increased from $60 million in 2024 and $30 million in 2025 to $562.5 million this year.

Two deals are far too few to call this a permanent change in the whole market. They do show that the companies attracting the largest checks currently have close links to bigger technology or industrial players.

Poolside's relationship with Nvidia goes into model technology and compute. RTX is both an investor in and a disclosed customer of Code Metal. That overlap between customer, strategic partner and capital provider is especially important at the top end of AI coding funding right now.

Investor measure	2026 YTD	2025 YTD	2024 YTD
Deals with strategic investors	2	1	3
Strategic participation share	50.0%	11.1%	27.3%
Median strategic round	$562.5M	$30M	$60M
Deals with disclosed leads	4	8	9
Disclosed-lead coverage	100%	88.9%	81.8%
Unique disclosed leads/co-leads	4	9	12
Investors leading more than one deal	0	1	0
Total deals	4	9	11

Are AI coding investors becoming much more selective?

AI coding funding looks much more selective today: investors are still willing to write enormous checks, but only a handful of companies are reaching that point.

The pattern shows up in several places at once. We counted four funded companies instead of eight last year, one first financing instead of three, four unique disclosed lead investors instead of nine and only two represented regions instead of four.

The deal-size distribution says the same thing from another angle. Small financings still happen. Giant financings still happen. What is currently missing is the broader middle where a larger group of companies can raise meaningful institutional rounds.

The current winners also have unusually strong reasons for investors to concentrate capital around them. Poolside has a deep strategic relationship with Nvidia. Code Metal is already being used in defense, aerospace and other mission-critical environments. Cursor's $900 million round last year came after the company said it had passed $500 million in annual recurring revenue and was being used by more than half of the Fortune 500.

Those are high bars.

Selectivity does not mean investors have lost interest in AI coding. Once a market attracts serious customers and very large outcomes, investors can become more aggressive around the companies they trust while becoming less generous with everyone else.

That is what the 2026 funding distribution currently looks like.

Could AI code assistant funding still change a lot before the end of 2026?

Yes. AI code assistant funding can still move dramatically because one large transaction is enough to reshape the entire year.

We saw exactly that in 2025. Funding through September 21 stood at $1.096 billion, while the full-year total eventually reached $3.479 billion. The deal count went from nine during the comparable period to fourteen for the whole year.

That makes mechanical annualization especially unreliable in this market. A handful of late rounds can add billions of dollars.

For now, 2026 has reached only 32.5% of last year's full-year funding total. At the same time, it has already reached 82.8% of the entire 2024 total.

The more interesting question over the next few months is what kind of financing comes next. Another billion-dollar transaction would make the annual headline bigger without fixing the lack of breadth. Several $10 million to $50 million rounds, new first financings or a wider group of lead investors would change the underlying market much more.

What would show that AI coding fundraising is getting healthier again?

A healthier AI code assistant funding market would show up first in the middle of the distribution, not necessarily in the annual total.

We would want to see more companies raising rather than another isolated megadeal. The return of $20 million to $50 million Series A and Series B rounds would be particularly useful because that part of the market is empty today.

More first financings would tell us that investors are building a new cohort instead of mainly adding money to companies that have already raised. A broader lead-investor base would make the same point from the investor side.

Code Understanding is another area worth watching. These startups already generate a large share of deal activity. If one or two start moving from small rounds into substantial Series A or Series B financings, it would suggest investors are assigning more value to the infrastructure that gives coding agents reliable context across large repositories.

The geographic mix could broaden again too. Europe's 25% current deal share comes from a single $1.75 million financing, while North America receives 99.85% of the money. More substantial European or Asian rounds would make the market less dependent on a handful of U.S. companies.

As of now, though, the strongest evidence still points toward concentration rather than renewed breadth.

What is actually changing in AI code assistant funding?

AI code assistant funding is becoming much more concentrated: 2026 has slightly more capital than the same part of 2025, while funding activity underneath the headline has contracted sharply.

The clearest change is breadth. Four companies have raised this year versus eight in 2025 and ten in 2024. Capital excluding rounds above $50 million has fallen from $90.9 million last year to $6.75 million today.

The second big change is the barbell in deal sizes. We have two transactions at $5 million or below, then nothing until $125 million and $1 billion. The $20 million to $50 million layer that produced Cline, Moderne and Unblocked financings last year has disappeared.

Existing winners are receiving nearly all the money. First-financing activity has fallen to one company, while approximately $1.13 billion has gone into follow-ons.

Category leadership is also being driven by individual companies more than broad waves of investment. Cursor made IDE Code Assistants look dominant in 2025; Poolside now does the same for Developer Coding Copilots. Code Understanding is more interesting as a breadth story because its deal share has kept rising without a corresponding surge in capital.

Strategic relationships have become especially visible at the top. Nvidia's Poolside transaction and RTX's participation in Code Metal connect financing directly with technology, customers and industrial strategy.

The figure we would treat most cautiously is the $65 million median round. Four extremely uneven transactions make that number look far more representative than it really is.

As of today, the clearest description of the AI code assistant funding market is a narrower market with deeper pockets for a very small number of winners.