Who is winning AI coding now?
In our AI code assistant market deck, you will find everything you need to understand the market
SUMMARY
Cursor is winning AI coding now. It has the largest revenue base among independent startups, the strongest proof of daily professional use, and the broadest product across editing, agents, review, and enterprise deployment.
The market has split into two races that increasingly overlap. Cursor and Cognition compete for professional engineering work, while Replit, Lovable, Emergent, and Bolt compete to turn an idea into a working application with less traditional coding.
Cursor's lead is commercial, not just technical. Its officially confirmed annualized revenue floor already exceeds the latest disclosed or responsibly estimated figures for its startup rivals, and reports suggest the gap widened sharply after that confirmation.
Replit is the strongest end-to-end platform rather than the strongest editor. Its advantage is the infrastructure underneath the agent: development environment, execution, databases, hosting, deployment, collaboration, and a user base of more than 50 million.
Cognition is the most dangerous direct challenger because it owns both an interactive editor and a more autonomous agent. The unresolved question is not whether Devin and Windsurf are ambitious enough, but whether the combined system can deliver dependable results and transparent commercial traction at scale.
Lovable has turned simplicity into a serious business. Its growth shows that a large part of the AI coding market begins outside engineering teams, with founders, designers, marketers, and operators who want a polished web product rather than a better programming environment.
Benchmarks do not settle the race. Cognition currently leads one important terminal benchmark, while Cursor remains ahead in broader adoption and product maturity; independent studies also show that coding agents can still slow experienced developers or introduce persistent defects.
Enterprise evidence is unusually important here because user counts are easy to inflate. Named deployments involving thousands of engineers, production repositories, review workflows, and long-lived applications carry more weight than registrations, generated projects, or one-off demos.
The strongest moats come from workflow data and infrastructure, not model access alone. Cursor learns from daily developer behavior, Replit controls the application runtime, Cognition can capture long agent trajectories, and StackBlitz owns distinctive browser-based infrastructure.
The large AI platforms remain the biggest threat to the startup ranking. Claude Code already exceeds Cursor on disclosed coding-product revenue, while GitHub, OpenAI, and Google can bundle models, repositories, cloud infrastructure, security, and enterprise contracts.
Our ranking is clear but not permanent: Cursor first, Replit second, Cognition third, Lovable fourth, and Emergent fifth. Cursor has the best overall package today; the challengers each lead a narrower part of the market and have credible ways to close the gap.
This market map, featured in our AI code assistant market deck, highlights top companies and startups in the AI code assistant market
Who is winning AI coding now?
Which AI coding startups actually belong in this race?
Nine AI coding startups deserve serious comparison today, but only Cursor, Replit, Cognition, and Lovable are large enough to fight for the overall lead.
We include three types of company because buyers increasingly compare them. Cursor, Cognition's Windsurf, and Augment Code help professional engineers work inside existing codebases. Cognition's Devin handles longer assignments with more independence. Replit, Lovable, Emergent, and Bolt turn plain-language instructions into working applications, often for people who are not full-time developers. Qodo focuses on reviewing, testing, and governing the growing amount of AI-written code, while Poolside builds coding models and private enterprise deployments.
The boundaries keep moving. Cursor now runs cloud agents and code review. Cognition connects Devin with Windsurf. Replit, Lovable, Emergent, and Bolt all combine code generation with deployment tools. A customer may start by comparing two app builders, then consider an editor or agent once the project becomes more complex.
We exclude GitHub Copilot, Claude Code, OpenAI Codex, Gemini Code Assist, Amazon Q Developer, and Vercel's v0 from the startup ranking. Those products shape the market, but they belong to larger technology companies or general AI laboratories. Sourcegraph remains relevant as a code-intelligence company, though its current strategy is broader than a direct race to become the main AI coding product. Magic remains closer to a research bet than a scaled commercial competitor.
Funding totals are approximate because private rounds, secondary transactions, and acquired-company financing are disclosed unevenly.
| Startup | Main AI coding product | Approximate cumulative funding |
|---|---|---|
| Cursor, operated by Anysphere | AI-native editor, cloud agents, code review, and proprietary coding models | $3.4B |
| Cognition | Devin autonomous agent, Windsurf editor, and proprietary SWE models | More than $1.6B |
| Replit | AI app building, cloud development, hosting, and deployment | $922M |
| Poolside | Coding models, agents, and private enterprise deployments | $626M |
| Lovable | Natural-language web application builder | $545M |
| Augment Code | Enterprise coding agents and large-codebase context | $252M |
| Emergent | Full-stack web and mobile app builder for nontechnical users | $230M |
| StackBlitz, operator of Bolt | Browser-based full-stack app builder using WebContainers | $135M |
| Qodo | AI code review, testing, quality, and governance | $120M |
Is there a clear AI coding startup leader, and who has the most revenue now?
Cursor is the clear overall leader among AI coding startups and makes about four times as much annualized revenue as its nearest disclosed rivals.
Cursor confirmed more than $1 billion in annualized revenue in late 2025, and Bloomberg later reported that the figure had doubled within three months. Lovable says it has exceeded $500 million. Sacra estimates Replit at approximately $525 million, up from $300 million at the end of 2025. Emergent recently reported a $120 million run rate after crossing $100 million earlier in the year.
Cursor also has the best proof of professional use. A recent company update says more than 70% of Fortune 500 companies now use the product. Customer studies describe 30,000 Nvidia developers committing three times more code, a rollout to 3,000 Stripe engineers, Salesforce teams moving more than 30% faster, and Amplitude shipping three times more production code. These are vendor case studies rather than independent tests, but the scale is hard to dismiss.
Cognition is harder to place. Devin grew from $1 million in annual recurring revenue to $73 million in nine months. Cognition then said the Windsurf acquisition more than doubled revenue and that combined enterprise revenue rose another 30% over the following seven weeks. Those disclosures put Cognition above $190 million, but they do not reveal the company's current total. Unverified estimates circulating in the market are much higher, so we leave them out.
Bolt last disclosed a $40 million run rate in early 2025. Augment Code, Qodo, Poolside, and Sourcegraph have not released current figures that can be compared responsibly with the leaders.
With Cursor indexed at 100, Replit and Lovable sit near 25 and Emergent near 6. Cognition may be above Emergent, but we need a fresh company figure before placing it precisely.
| Startup | Best available annualized revenue evidence | Indexed against Cursor at 100 | Confidence |
|---|---|---|---|
| Cursor | About $2B reported; more than $1B officially confirmed | 100 | High on the official floor, medium on the latest report |
| Replit | About $525M estimated by Sacra | 26 | Medium |
| Lovable | More than $500M company-reported | 25 | Medium to high |
| Cognition | Historical disclosures imply more than $190M | More than 10 | High on the floor, low on the current position |
| Emergent | About $120M company-reported | 6 | Medium |
| Bolt | $40M last disclosed in early 2025 | 2 | Low for the current position |
If you want more recent data on this point, please see our latest AI code assistant market report.
As this chart shows, and as featured in our AI code assistant market deck, search interest in AI code assistants has increased significantly
Which AI coding startup is growing fastest now?
At large scale, Lovable is growing fastest, Cursor is adding the most revenue in dollars, and Emergent has made the sharpest recent jump from a smaller base.
Lovable doubled from $100 million to $200 million in four months and later reached $400 million. Its latest milestone added another 25%, a slower pace than the early surge but still exceptional at this size. Lovable also says its users create about one million new projects each week.
Cursor's reported run rate doubled in three months, adding roughly another billion dollars of annualized business. No other startup in the field has publicly added that much over a comparable period.
Emergent moved from $50 million to $100 million in annualized revenue in one month, raised a $130 million Series C soon afterward, and reported more than 12 million applications built since launch. It remains much smaller than Lovable or Replit, but it has entered the serious challenger group.
Sacra estimates Replit rose from $300 million at the end of 2025 to $525 million by April 2026, a gain of about 75% in four months. Cognition's last comparable figure is older: enterprise revenue rose 30% in the seven weeks after it bought Windsurf.
Who has the biggest AI coding user base, and is Cursor still the default for professional developers?
Replit reaches the largest disclosed AI coding audience, but Cursor remains the default independent tool for professional developers.
Replit reports more than 50 million users. Its audience existed before the current AI boom, so the number includes students, traditional programmers, hobbyists, founders, and employees building internal tools.
Cursor counts millions of developers rather than tens of millions of registered users. Recent deployments show deeper professional use: Nvidia reports adoption across 30,000 developers, Stripe rolled it out to 3,000 engineers, and Dropbox indexed more than 550,000 files. Coinbase says Cursor reduced the time from idea to production by 90%, while Faire says cloud agents doubled pull-request throughput.
Cursor now covers repository search, multi-file changes, parallel agents, cloud handoffs, pull-request review, browser inspection, and its own Composer models. Cognition's Windsurf is the closest startup alternative, combining an AI editor with direct access to Devin for longer background tasks.
Lovable reported more than 50 million projects built and about one million new projects per week. Roughly 80% of its users were solo builders in its first user report, and most came from outside engineering. Emergent says most of its applications are built by people who had never coded before. Bolt had around five million registered users in its last widely reported disclosure.
Anthropic says Claude Code exceeds $2.5 billion in run-rate revenue and that weekly active users doubled from the start of 2026. Cursor therefore leads the specialist startup field, while professional developers increasingly mix editors, terminal agents, cloud agents, and GitHub-native assistants.
This chart, featured in our AI code assistant market deck, illustrates yearly VC funding for AI code assistant startups
Is Replit or Lovable winning AI app building?
Replit currently leads AI app building as a full platform; Lovable is faster and simpler for turning an idea into a polished web product.
Lovable lets a founder, product manager, marketer, or designer describe a product, adjust the interface visually, connect common services, and publish it without first learning a development environment. It now includes hosting, databases, authentication, payments, collaboration, and enterprise controls.
Zendesk says a process that took six weeks fell to three hours. Deutsche Telekom uses Lovable to turn stakeholder discussions into functional prototypes within days. Another customer said a four-week project involving 20 people became a four-day sprint with four people.
Replit covers more of the software lifecycle. Agent 4 can create the application, work with parallel agents, manage the environment, run the code, connect a database, deploy the service, and keep the project operating on Replit's cloud. UKG says the platform increased its capacity to gather product feedback by 400%.
Lovable remains the better choice for quickly producing a strong web interface. Replit becomes more attractive as the application grows in technical depth. Emergent now challenges both with web and mobile support and downloadable code.
If you want more recent data on this point, please see our latest AI code assistant market report.
Which AI coding startup has the strongest enterprise customers and deployments?
Among AI coding startups, Cursor has the strongest evidence of large enterprise rollouts and real production use.
The clearest evidence comes from deployments with named scale. Nvidia uses Cursor across 30,000 developers, while Stripe rolled it out to 3,000 engineers. Cursor says more than 70% of the Fortune 500 now use the product, and customer studies describe gains at Salesforce, Amplitude, Coinbase, Faire, PayPal, Dropbox, National Australia Bank, and Wayfair.
Cognition has the strongest customer list among Cursor's challengers. Devin and Windsurf are used by Goldman Sachs, Citi, Dell, Cisco, Palantir, Ramp, Nubank, and Mercado Libre. Mercedes-Benz is now deploying both products across its global engineering organization for legacy modernization, cloud-native development, and logistics. Infosys has also partnered with Cognition to deploy Devin internally and across client work.
Replit reaches users inside 85% of the Fortune 500 and names Atlassian, Labcorp, PayPal, Zillow, Talkdesk, Adobe, and UKG. The phrase "users inside" can include scattered adoption, so it does not mean Replit has completed company-wide rollouts across 85% of the Fortune 500. UKG provides the clearest deep example because Replit describes use across prototyping, product development, internal tools, and a reusable design framework.
Lovable has credible corporate use at Klarna, Deutsche Telekom, Zendesk, and other unnamed global companies. The work often begins in product, design, sales, or operations rather than central engineering.
This chart, featured in our AI code assistant market deck, breaks down Anyshpere’s playbook in AI code assistants
Which AI coding startup has the most mature and best-performing product?
Cursor has the most mature professional AI coding product, while Cognition currently leads the main public startup benchmark.
Cursor covers repository search, autocomplete, multi-file edits, terminal work, browser checks, background agents, code review, pull requests, plugins, team controls, and model choice. Cursor also acquired Graphite, strengthening its position around review and collaboration.
Replit has spent more than a decade building the cloud layer underneath its agent. Users receive a development environment, package installation, execution, collaboration, databases, hosting, deployment, monitoring, and account controls. Agent 4 can work across multiple artifacts and run parallel agents, while Replit remains responsible for keeping the resulting application online.
Cognition covers the widest range of professional AI coding workflows. Windsurf supports interactive work, while Devin can take a larger assignment and continue independently. Devin Desktop, Devin CLI, code review, Windsurf integration, government deployments, and local-to-cloud handoffs have moved the company well beyond its original autonomous-agent demo.
A recent Terminal-Bench 2.0 snapshot placed Cognition's SWE-1.7 at 81.5%, Poolside's Laguna model at 70.2%, and Cursor's Composer 2.5 at 69.3%. Cognition leads that benchmark, which measures difficult tasks performed through a command-line environment.
Cursor's own CursorBench tells a different story. Composer 2.5 scored below several frontier models but completed tasks at a reported average cost of $0.44, far below many higher-scoring options. Since Cursor created the benchmark from its own sessions, we treat the result as product research rather than neutral proof of superiority.
Independent studies remain more sobering. METR's randomized study of 16 experienced open-source developers found that early-2025 AI tools made them 19% slower on 246 tasks, even though the developers expected to become faster. A newer study of AI-generated commits across thousands of GitHub repositories found that more than 15% of commits from every tool studied introduced at least one issue, and almost a quarter of the tracked issues remained in the latest version of the codebase.
Lovable is already mature for web app creation, but its scope is narrower. A security incident earlier this year exposed data from public Lovable projects to authenticated users; Lovable fixed the flaw within two hours and later admitted that both the product and its first response had fallen short. Emergent is developing quickly but has had less time to prove reliability at enterprise scale.
Cognition wins the benchmark comparison. Cursor still wins the broader product comparison because more large engineering teams have deployed it in daily work.
If you want more recent data on this point, please see our latest AI code assistant market report.
Which AI coding startup gives customers and investors the best value?
For a working developer, Cursor offers the clearest customer value; Lovable turns funding into revenue most efficiently, while Replit includes the most in one subscription.
Cursor's Pro plan starts around $20 per month and includes editor assistance, agents, repository context, and access to several models. Heavy usage and higher tiers can push spending far above the entry price.
Windsurf also starts around $20 per month after moving from credits to quotas. It pairs the editor with Cognition's SWE models and a route into Devin's more autonomous workflows.
Lovable and Bolt begin around $25 per month. Lovable charges through build credits, while Bolt meters tokens and becomes more expensive as projects grow.
Replit Core costs about $20 per month when billed annually, while Replit Pro costs about $90 annually billed or $100 month to month. The price includes credits, parallel agents, databases, deployment, hosting, and collaboration.
Lovable has raised about $545 million, and its annualized revenue now sits close to that amount, or roughly $0.92 of run-rate revenue for every dollar raised. Replit produces about $0.57 using Sacra's latest revenue estimate, while Cursor sits close to $0.59 against approximately $3.4 billion raised.
Emergent's ratio is near $0.52. Cognition has raised more than $1.6 billion but has not disclosed a fresh revenue total. Poolside has raised $626 million with far less public evidence of commercial scale.
| Product | Typical individual starting price | Best value for | Main cost risk |
|---|---|---|---|
| Cursor | About $20 per month | Professional developers working daily in existing codebases | Heavy model and agent use can require higher tiers |
| Windsurf | About $20 per month | Developers wanting an editor plus Devin workflows | Quotas and packaging may change as Devin integration expands |
| Replit | About $20 per month billed annually; Pro about $90 per month billed annually | Building, deploying, and operating an application in one platform | Usage credits can make complex builds harder to budget |
| Lovable | About $25 per month | Fast, polished web products for founders and business teams | Repeated revisions consume build credits |
| Bolt | About $25 per month | Browser-based full-stack web prototypes and applications | Token costs rise with project size |
This chart, featured in our AI code assistant market deck, illustrates yearly funding for AI code assistant startups
Which AI coding startup has the strongest moat?
Cursor has the strongest moat in AI coding today because daily workflow habits, enterprise rollouts, proprietary models, product data, and capital reinforce one another.
Cursor sees how professional developers search, accept, reject, edit, test, and review generated code inside real repositories. That feedback can improve context retrieval, agent behavior, review, and its Composer models.
The editor also creates habit. Enterprise controls, indexed codebases, shared rules, review workflows, and internal training deepen the switching cost once a company standardizes on Cursor.
Replit owns a different moat through infrastructure and distribution. More than 50 million users can move from learning or prototyping into AI app building without changing platforms, while Replit can create, run, host, and deploy the resulting software.
Cognition may eventually build the richest training dataset. Windsurf captures interactive decisions, while Devin records longer trajectories across planning, testing, failure, and recovery.
StackBlitz owns distinctive browser infrastructure through WebContainers. Qodo has a credible position around verification and governance. Lovable and Emergent rely more heavily on product design, brand, community, and distribution.
Cursor's main weakness is its continued dependence on frontier models from companies that compete directly with it.
If you want more recent data on this point, please see our latest AI code assistant market report.
Which AI coding startup has the strongest momentum now?
Cursor has the strongest AI coding momentum now because it is advancing on revenue, products, models, and enterprise rollouts at the same time.
Cursor released Cursor 3 as a unified agent workspace, added organization-level controls, opened an SDK for programmatic agents, improved Bugbot so it runs more than three times faster while costing 22% less, and partnered with SpaceX for model-training compute.
Lovable has lately pushed beyond web app generation. The product can now analyze spreadsheets, documents, presentations, and data files, then turn that material into applications. New planning features also map the build before writing code.
Cognition has added Devin inside Windsurf, launched Devin Desktop and CLI workflows, expanded code review, previewed stronger SWE models, and won a global Mercedes-Benz deployment.
Replit launched Agent 4, which the company says is ten times faster than Agent 3, raised another $400 million, and deepened its enterprise links with Databricks and other partners. Emergent raised $130 million, reached a $1.5 billion valuation, and continued adding applications at high speed.
This chart, featured in our AI code assistant market deck, compares the main business model options for AI developer tools platforms
Can Anthropic, GitHub, OpenAI, or Google take the AI coding lead from startups?
Anthropic already leads Cursor on disclosed coding-product revenue, and the large AI platforms remain the biggest threat to every startup in this ranking.
Anthropic says Claude Code has passed $2.5 billion in run-rate revenue and doubled its weekly active users since the start of 2026. Claude Code also sits close to Anthropic's frontier models, allowing the company to improve the model, terminal agent, enterprise contract, and pricing together.
Microsoft reported 4.7 million paid GitHub Copilot subscribers, up 75% year over year. GitHub already owns the repositories, pull requests, actions, security tools, and enterprise relationships surrounding much of professional software development.
OpenAI is pushing Codex across the desktop, IDE, cloud, and mobile. A recent study using Codex activity found that active users grew more than fivefold during the first half of 2026, while more than 10% of users managed at least three concurrent agents in a typical week.
Google can combine Gemini models, Google Cloud, Android, developer infrastructure, and existing corporate agreements.
Cursor still clears the bar for an independent product because its workflow remains strong enough to justify a separate subscription. Replit owns the application runtime, while Lovable and Emergent reach users who do not begin inside GitHub or a terminal.
Which AI coding startups are actually ahead?
Cursor ranks first among AI coding startups now, and the lead over Replit, Cognition, and Lovable is too large to explain through funding or publicity alone.
Cursor wins on the facts buyers care about most in AI coding. It has the largest startup revenue base, the deepest professional adoption, several deployments involving thousands of engineers, a mature product spanning editor work and cloud agents, proprietary models, code review, and the capital to keep investing. No challenger currently matches that full package.
Replit ranks second because it controls more of the path from an idea to a running application than any other startup. The company has the largest disclosed user base, revenue around one quarter of Cursor's reported level, strong recent growth, and years of cloud infrastructure underneath Agent 4. Replit would move closer to first by hitting its public revenue target and proving that large customers keep running important applications on the platform.
Cognition ranks third and remains Cursor's most dangerous direct challenger. Devin, Windsurf, strong enterprise customers, proprietary SWE models, and large financing give Cognition the right pieces. The ranking stays below Replit because current revenue, retention, agent success rates, and post-acquisition integration remain less transparent. A fresh revenue disclosure near Replit's level, combined with more global rollouts like Mercedes-Benz, could change the order quickly.
Lovable ranks fourth. Its revenue and growth are already close to Replit's, and its product is often better for nontechnical builders. We keep Lovable below Cognition because the product covers a narrower part of software development and because long-term retention, gross margins, and the share of projects that remain in production are still unclear.
Emergent enters fifth after building a nine-figure revenue business, raising $230 million, and producing an eight-figure application count in roughly a year. Bolt follows because its browser technology remains distinctive, though its freshest commercial figures trail Emergent. Augment Code and Qodo lead useful enterprise niches. Poolside has serious technology and capital but has yet to prove comparable demand.
The ranking can change under clear conditions. Replit can challenge Cursor by turning its large user base into durable enterprise software revenue. Cognition can pass Replit by proving that Devin and Windsurf work as one dependable system at scale. Lovable can rise by showing strong retention and production use after the first burst of app creation. Emergent can join the leading group by sustaining its current growth for another year. Cursor can lose the overall market lead to Claude Code even while remaining the top startup.
Our answer to the original question is firm: Cursor is winning the AI coding startup race today. Replit is the strongest end-to-end platform, Cognition is the closest professional challenger, Lovable leads frictionless web app creation, and Emergent is the fastest new entrant worth watching.
| Rank | Startup | Why it holds this position now |
|---|---|---|
| 1 | Cursor | Largest startup revenue base, strongest professional adoption, mature editor-agent suite, proprietary models, and repeated large enterprise deployments |
| 2 | Replit | Biggest user base, strong revenue growth, and the most complete platform for building, deploying, and operating applications |
| 3 | Cognition | Best direct challenge through Devin plus Windsurf, strong enterprise customers, proprietary SWE models, and substantial capital |
| 4 | Lovable | Exceptional growth and the easiest strong product for nontechnical web app builders, with less proof of long-term production depth |
| 5 | Emergent | Nine-figure run rate, more than 12M applications, fast international growth, and a strong position in web and mobile app building |
| 6 | StackBlitz, operator of Bolt | Distinctive browser infrastructure and a proven app-building product, but older and smaller disclosed commercial figures |
| 7 | Augment Code | Credible enterprise context and agent platform for complex codebases, without leader-level public traction |
| 8 | Qodo | Strong specialist position in review, testing, and governance as AI-written code creates a larger quality bottleneck |
| 9 | Poolside | Valuable private-deployment and model strategy, but limited commercial proof and greater infrastructure execution risk |
If you want more recent data on this point, please see our latest AI code assistant market report.
This chart, featured in our AI code assistant market deck, illustrates how market revenue is distributed across customer segments in the AI code assistant market
OUR METHODOLOGY
This analysis asks which independent AI coding startup is ahead based on the evidence available today. We compare commercial traction, growth, user adoption, enterprise deployments, product maturity, technical performance, customer value, competitive positioning, strategic moat, and market momentum before drawing the overall ranking.
We treat AI coding as a market with several overlapping product categories. Professional editors and coding agents, autonomous software-engineering agents, app-building platforms, code-quality tools, and private coding-model providers are compared when buyers increasingly consider them for the same budget or workflow.
The ranking covers independent startups only. GitHub Copilot, Claude Code, OpenAI Codex, Gemini Code Assist, Amazon Q Developer, and Vercel v0 are used as external market benchmarks because they shape pricing, product expectations, and competitive pressure, but they are not included in the startup table.
We prioritized the freshest publicly available evidence. Direct company disclosures, named customer deployments, primary product documentation, current pricing pages, funding announcements, independently run benchmarks, technical research, and tier-one reporting received the most weight.
Not every metric is treated equally. Revenue and named production deployments carry more weight than registrations, generated-project counts, valuation headlines, or isolated benchmark wins. Company case studies are useful when they include concrete scale, but we do not treat vendor-reported productivity gains as neutral experiments.
Funding totals and annualized revenue figures remain approximate because private rounds, secondary transactions, acquisitions, and run-rate milestones are disclosed unevenly. When a current commercial figure could not be verified with enough confidence, we used a conservative floor or left the company unranked on that metric.
The final order is an editorial synthesis rather than a fixed scoring formula. A company ranks highly only when it performs well across several dimensions that matter to developers, enterprise buyers, customers, and investors; one benchmark, funding round, or growth headline is not enough.
Key sources used for this analysis include: Cursor's product, revenue, and customer updates, Cursor pricing, Cognition's Devin and Windsurf updates, Devin documentation, Windsurf product information, Replit's company and product updates, Replit pricing, Lovable's company updates, Lovable pricing, Poolside, Augment Code, Qodo, Bolt, StackBlitz, Anthropic's Claude Code disclosures, GitHub Copilot updates, OpenAI Codex updates, Google's developer product updates, NVIDIA developer case studies, METR research, Bloomberg reporting, Sacra estimates, and Terminal-Bench.
This chart, featured in our AI code assistant market deck, shows how AI coding assistant technology has evolved over time
Related blog posts
- Who is emerging as the leader in AI coding agents?
- Vibe coding: which startup is ahead?
