AI ‘vibe coding' startups burst onto scene with sky-high valuations
[NEW YORK] Two years after the launch of ChatGPT, return on investment in generative AI has been elusive, but one area stands out: software development.
So-called code generation or 'code-gen' startups are commanding sky-high valuations as corporate boardrooms look to use AI to aid, and sometimes to replace, expensive human software engineers.
Cursor, a code generation startup based in San Francisco that can suggest and complete lines of code and write whole sections of code autonomously, raised US$900 million at a US$10 billion valuation in May from a who's who list of tech investors, including Thrive Capital, Andreessen Horowitz and Accel.
Windsurf, a Mountain View-based startup behind the popular AI coding tool Codeium, attracted the attention of ChatGPT maker OpenAI, which is now in talks to acquire the company for US$3 billion, sources familiar with the matter told Reuters.
Its tool is known for translating plain English commands into code, sometimes called 'vibe coding,' which allows people with no knowledge of computer languages to write software. OpenAI and Windsurf declined to comment on the acquisition.
'AI has automated all the repetitive, tedious work,' said Scott Wu, CEO of code gen startup Cognition. 'The software engineer's role has already changed dramatically. It's not about memorising esoteric syntax anymore.'
BT in your inbox
Start and end each day with the latest news stories and analyses delivered straight to your inbox.
Sign Up
Sign Up
Founders of code-gen startups and their investors believe they are in a land grab situation, with a shrinking window to gain a critical mass of users and establish their AI coding tool as the industry standard.
But because most are built on AI foundation models developed elsewhere, such as OpenAI, Anthropic, or DeepSeek, their costs per query are also growing, and none are yet profitable.
They're also at risk of being disrupted by Google, Microsoft and OpenAI, which all announced new code-gen products in May, and Anthropic is also working on one as well, two sources familiar with the matter told Reuters.
The rapid growth of these startups is coming despite competing on big tech's home turf. Microsoft's GitHub Copilot, launched in 2021 and considered code-gen's dominant player, grew to over US$500 million in revenue last year, according to a source familiar with the matter.
Microsoft declined to comment on GitHub Copilot's revenue. On Microsoft's earnings call in April, the company said the product has over 15 million users.
Learn to code?
As AI revolutionises the industry, many jobs – particularly entry-level coding positions that are more basic and involve repetition – may be eliminated. Signalfire, a VC firm that tracks tech hiring, found that new hires with less than a year of experience fell 24 per cent in 2024, a drop it attributes to tasks once assigned to entry-level software engineers are now being fulfilled in part with AI.
Google's CEO also said in April that 'well over 30 per cent' of Google's code is now AI-generated, and Amazon CEO Andy Jassy said last year the company had saved 'the equivalent of 4,500 developer-years' by using AI. Google and Amazon declined to comment.
In May, Microsoft CEO Satya Nadella said at a conference that approximately 20 to 30 per cent of their code is now AI-generated. The same month, the company announced layoffs of 6,000 workers globally, with over 40 per cent of those being software developers in Microsoft's home state, Washington.
'We're focused on creating AI that empowers developers to be more productive, creative, and save time,' a Microsoft spokesperson said. 'This means some roles will change with the revolution of AI, but human intelligence remains at the centre of the software development life cycle.'
Mounting losses
Some 'vibe-coding' platforms already boast substantial annualised revenues.
Cursor, with just 60 employees, went from zero to US$100 million in recurring revenue by January 2025, less than two years since its launch. Windsurf, founded in 2021, launched its code generation product in November 2024 and is already bringing in US$50 million in annualised revenue, according to a source familiar with the company.
But both startups operate with negative gross margins, meaning they spend more than they make, according to four investor sources familiar with their operations.
'The prices people are paying for coding assistants are going to get more expensive,' Quinn Slack, CEO at coding startup Sourcegraph, told Reuters.
To make the higher cost an easier pill to swallow for customers, Sourcegraph is now offering a drop-down menu to let users choose which models they want to work with, from open source models such as DeepSeek to the most advanced reasoning models from Anthropic and OpenAI so they can opt for cheaper models for basic questions.
Both Cursor and Windsurf are led by recent MIT graduates in their twenties, and exemplify the gold rush era of the AI startup scene. 'I haven't seen people working this hard since the first Internet boom,' said Martin Casado, a general partner at Andreessen Horowitz, an investor in Anysphere, the company behind Cursor.
What's less clear is whether the dozen or so code-gen companies will be able to hang on to their customers as big tech moves in.
'In many cases, it's less about who's got the best technology – it's about who is going to make the best use of that technology, and who's going to be able to sell their products better than others,' said Scott Raney, managing director at Redpoint Ventures, whose firm invested in Sourcegraph and Poolside, a software development startup that's building its own AI foundation model.
Custom AI models
Most of the AI coding startups currently rely on the Claude AI model from Anthropic, which crossed US$3 billion in annualised revenue in May in part due to fees paid by code-gen companies.
But some startups are attempting to build their own models. In May, Windsurf announced its first in-house AI models that are optimised for software engineering in a bid to control the user experience. Cursor has also hired a team of researchers to pre-train its own large frontier-level models, which could enable the company to not have to pay foundation model companies so much money, according to two sources familiar with the matter.
Startups looking to train their own AI coding models face an uphill battle as it could easily cost millions to buy or rent the computing capacity needed to train a large language model.
Replit earlier dropped plans to train its own model. Poolside, which has raised more than US$600 million to make a coding-specific model, has announced a partnership with Amazon Web Services and is testing with customers, but hasn't made any product generally available yet.
Another code gen startup Magic Dev, which raised nearly US$500 million since 2023, told investors a frontier-level coding model was coming in summer 2024 but hasn't yet launched a product.
Poolside declined to comment. Magic Dev did not respond to a request for comment. REUTERS
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Business Times
22 minutes ago
- Business Times
China's power prices nosedive, providing relief to tariff-hit factories
[BEIJING] A crash in Chinese coal prices, and a flood of clean energy, are cutting power rates in major industrial hubs, easing pressure on some of the factories caught in the crossfire of trade hostilities with the US. In Jiangsu province, the average price for June contracts settled by centralised bidding plunged 24 per cent year on year to 313 yuan (S$56) per megawatt-hour (MWh), the lowest level regulators allow. In Guangdong, the rate dropped 8.3 per cent to the government's floor of 373 yuan per MWh. The two coastal powerhouses are heavily dependent on exports, with a combined economy bigger than that of France. The main driver is a collapse in the price of coal – down 30 per cent over the past year – which accounts for more than half of the electricity generated in China. A shift to market-based pricing, particularly for renewables, is also helping to weaken rates. 'Generation costs are cracking,' said David Fishman, a Shanghai-based principal at consultancy The Lantau Group. 'Coal generators, when they are bidding into the system, they are able to entertain lower prices because all of the generation costs are lower.' Just as economic conditions worsen due to the trade war, Chinese industry is finding relief in the government's drive for energy security. That effort has been most successful in power markets. The country is sitting on a glut of coal that's built up over years and leads the world in its roll-out of wind and solar. Lower power costs are not universal though, because of the way the market is structured. The monthly contract declines apply to only about a quarter of electricity sales, Fishman estimated, with some 70 per cent of transactions decided on a yearly basis. Spot or short-term trades account for the rest. BT in your inbox Start and end each day with the latest news stories and analyses delivered straight to your inbox. Sign Up Sign Up Still, a lot of annual deals are at least partially tied to monthly or short-term pricing, which will have expanded the number of beneficiaries, Fishman said. 'I would be surprised if there were a lot of power buyers out there just seething, looking at the prices dropping, thinking how come I didn't see that?' he said. Policy shifts Recent policy changes have also pressured electricity prices. The sharp drop in Jiangsu's monthly rate was largely due to the start-up of its spot market in June, said Nannan Kou, head of China research at BloombergNEF. Power generators were keen to lock in buyers ahead of the extra competition from spot traders, and lower coal prices allowed them to offer discounts while still eking out a profit, he said. Most regions in China are expected to have at least trial operations for spot markets, which rely on near instantaneous assessments of supply and demand, by the end of the year. Clean energy is a key pillar of China's embrace of market forces. Starting Jun 1, all renewables generators have to sell their power on the open market, although projects connected to the grid before that date will retain price protections that limit their downside. That creates the incentive to sell electricity without worrying too much about falling prices, said Lantau's Fishman, which could put more pressure on rates and deliver even lower costs to customers. BLOOMBERG
Straits Times
43 minutes ago
- Straits Times
OpenAI finds more Chinese groups using ChatGPT for malicious purposes
FILE PHOTO: A response in Chinese by ChatGPT, an AI chatbot developed by OpenAI, is seen on its website in this illustration picture taken February 9, 2023. REUTERS/Florence Lo/Illustration/File Photo SAN FRANCISCO - OpenAI is seeing an increasing number of Chinese groups using its artificial intelligence technology for covert operations, which the ChatGPT maker described in a report released Thursday. While the scope and tactics employed by these groups have expanded, the operations detected were generally small in scale and targeted limited audiences, the San Francisco-based startup said. Since ChatGPT burst onto the scene in late 2022, there have been concerns about the potential consequences of generative AI technology, which can quickly and easily produce human-like text, imagery and audio. OpenAI regularly releases reports on malicious activity it detects on its platform, such as creating and debugging malware, or generating fake content for websites and social media platforms. In one example, OpenAI banned ChatGPT accounts that generated social media posts on political and geopolitical topics relevant to China, including criticism of a Taiwan-centric video game, false accusations against a Pakistani activist, and content related to the closure of USAID. Some content also criticized U.S. President Donald Trump's sweeping tariffs, generating X posts, such as "Tariffs make imported goods outrageously expensive, yet the government splurges on overseas aid. Who's supposed to keep eating?". In another example, China-linked threat actors used AI to support various phases of their cyber operations, including open-source research, script modification, troubleshooting system configurations, and development of tools for password brute forcing and social media automation. A third example OpenAI found was a China-origin influence operation that generated polarized social media content supporting both sides of divisive topics within U.S. political discourse, including text and AI-generated profile images. China's foreign ministry did not immediately respond to a Reuters request for comment on OpenAI's findings. OpenAI has cemented its position as one of the world's most valuable private companies after announcing a $40 billion funding round valuing the company at $300 billion. REUTERS Join ST's Telegram channel and get the latest breaking news delivered to you.
Business Times
an hour ago
- Business Times
Court partially allows Goh Jin Hian's appeal, finds he did not breach duty by not probing IPP's red flags
[SINGAPORE] The Appellate Division of the High Court has partially allowed an appeal by Goh Jin Hian against having to pay damages for breaching his duty of care as a then-director of the insolvent marine fuel supplier, Inter-Pacific Petroleum (IPP). The ruling on Thursday (Jun 5) said that Goh had breached his duty of care as a result of not being aware of IPP's cargo trading business – not because he had failed to open a probe into red flags surrounding the company. The justices presiding were Tay Yong Kwang, Woo Bih Li and Kannan Ramesh. Goh was also found not to have breached his duty to act in the best interests of IPP's creditors regarding drawdowns on bank facilities in relation to fraudulent cargo trades. This follows his being found liable in February 2024 for breaching of his director's duties, statutory duties and the losses suffered by the firm, which came to US$146 million plus interest. The liquidators of IPP had sued Dr Goh, the son of former prime minister Goh Chok Tong, to recover US$156 million in losses, accusing him of 'sleepwalking through his time as a director' and failing to discover and stop the drawdowns in trade financing between June 2019 and July 2019, said to have been funding non-existent or sham transactions. BT in your inbox Start and end each day with the latest news stories and analyses delivered straight to your inbox. Sign Up Sign Up In his grounds of decision released last July, High Court Justice Aedit Abdullah said Dr Goh had not taken 'reasonable steps', such as by making the necessary inquiries, when red flags surrounding the company arose. Goh was also unaware of the existence of IPP's cargo trading business, despite being a director of the company, and therefore did not know this business was a fraudulent scheme perpetrated by IPP, said the justice. Following the appeal, the judgement has been set aside, and Dr Goh no longer has to pay damages to IPP. While the Appellate Division agreed with the previous judgement that Goh had breached his duty of care by being unaware of IPP's cargo trading business, it found that the three red flags raised in the previous judgement were not 'red flags that would have put Dr Goh on a train of inquiry leading to the fraud in the cargo trading business being uncovered'. One such red flag was an audit confirmation request relating to amounts of receivables due to IPP from customer Mercuria Energy Trading, which Goh signed and was sent to Mercuria on Feb 7, 2018. The sum due was US$132 million. While Justice Aedit said Goh should have made inquiries upon receiving the audit confirmation request, the Appellate Division said the fact that this sum was requested by Mercuria was 'not, in and of itself, enough to put him on inquiry'. This was because Mercuria was a big company and that the size of the receivable could have been explained by IPP's sizeable trading volume, amounting to about US$1 billion, with it. Two other issues that IPP's liquidators had called red flags – the suspension of IPP's bunker craft operator licence in June 2019 and three confirmations of indebtedness signed by Dr Goh in July 2019 – were also found not to be red flags by the Court of Appeal. In the case of the suspension, 'even if Dr Goh had made the inquiries... it is unclear if he would have uncovered fraud in the cargo trading business, even if he had learned that IPP was carrying on such business'. The judges were not persuaded that the suspension of the licence was a red flag. As for the confirmation of indebtedness, there was no assertion in the confirmations that the debts were for the cargo trading business, and they were thus not considered red flags. The Appellate Division therefore departed from Justice Aedit's finding that Dr Goh breached the care duty regarding the red flags. It also disagreed with Justice Aedit that Dr Goh did not breach his duty to act in the best interests of the respondent's creditors on the drawdowns for fraudulent cargo trades made on IPP's bank facilities. It found that IPP bears the legal burden of proving that the fraud would have been detected, and that the resulting loss would have been averted had Dr Goh known that IPP was undertaking the cargo trading business, but failed to discharge this burden. Dr Goh was represented by TSMP Law Corporation, led by joint managing partner Thio Shen Yi; IPP's liquidators were represented by LVM Law Chambers, led by managing director Lok Vi Ming. After the appeal, Thio said the decision has practical implications for all directors, as the Court of Appeal has clarified that it 'cannot be part of a director's duty of supervision and oversight to pick up fraud unless there are tell-tale warning signs'. 'Directors owe fiduciary obligations and the duty of care to the company, but the Appeals Court has crucially recognised the practical and commercial limits to their ability to scrutinise for and detect fraud, especially deep-seated fraud,' he added.
