OpenAI\u2019s rapid-fire rollout is aimed at fueling faster growth and indirectly addressing last month\u2019s peak-performance skeptics. The underlying belief remains: more data, greater compute power, and larger models will significantly boost capabilities. For the past two years, the industry has followed this philosophy\u2014investing in GPUs, building massive data centers (even at the risk of legal battles) to gather ever more data, and continuously scaling up the model size.<\/p>\n\n\n\n
\n\n\n\n
II. Has Scaling Up Hit a Wall?<\/h3>\n\n\n\n
By November, mounting voices began to question whether OpenAI\u2019s path of endless expansion had finally reached its limit:<\/p>\n\n\n\n
- \n
- Marc Andreessen<\/strong>, co-founder of Silicon Valley venture firm a16z and investor in multiple big-model companies, remarked, \u201cEven if we keep adding GPUs at the same pace, there\u2019s simply no corresponding leap in intelligence.\u201d<\/li>\n\n\n\n
- Ilya Sutskever<\/strong>, co-founder and former chief scientist at OpenAI, noted, \u201cThe 2010s were all about scaling up; now we\u2019re back to needing breakthroughs and new discoveries.\u201d<\/li>\n<\/ul>\n\n\n\n
Media reports indicated that companies like Google, OpenAI, and Anthropic have struggled to achieve the dramatic improvements seen in earlier generations of models. Although high-ranking executives dismiss the \u201cwall\u201d theory\u2014and evidence shows that efforts to build larger compute centers are not slowing down\u2014they\u2019re now channeling more resources into big-model applications. From OpenAI, Anthropic, and Google to Microsoft and venture capital firms, the focus is shifting toward \u201cAgents\u201d: systems that enable big models to understand human instructions and orchestrate databases and tools to tackle complex tasks.<\/p>\n\n\n\n
\n\n\n\nIII. Data, Compute, and the Quest for Better AI<\/h3>\n\n\n\n
The industry\u2019s mantra has long been \u201cmore data, more compute, and larger models\u201d\u2014a principle known as Scaling Laws<\/strong>. According to this idea, better performance comes simply from throwing more computational power and data at the problem. For the past couple of years, this approach has driven rapid progress\u2014but now, some believe we may be approaching its limits.<\/p>\n\n\n\n
New Signals & Directions<\/h4>\n\n\n\n
- \n
- Synthetic Data:<\/strong>
In a bid to overcome the shortage of high-quality, fresh data, many companies are experimenting with synthetic data. For example, OpenAI is using the synthetic data generated by its September-released o1 to train Orion, though this data isn\u2019t a perfect substitute. Orion\u2019s performance hasn\u2019t met expectations so far.<\/li>\n\n\n\n- Higher Precision Data:<\/strong>
Researchers from Harvard, Stanford, and MIT published a paper on November 7 highlighting that lowering data precision (for example, using 32-bit, 16-bit, or even 8-bit representations instead of 64-bit) can negatively impact model quality. The traditional Scaling Laws did not account for these differences in precision.<\/li>\n\n\n\n- From Pre-Training to Post-Training:<\/strong>
Some researchers are focusing on post-training strategies. By letting a model \u201cthink\u201d longer\u2014asking it a question dozens or even hundreds of times and then picking the best answer\u2014performance can be improved. This approach mirrors OpenAI\u2019s path with o1 and is also being explored by Google and Meta. Meanwhile, several Chinese companies (from Alibaba to local startups) are releasing models in the o1 direction, some even naming them similarly to signal they\u2019re catching up to the cutting edge.<\/li>\n<\/ol>\n\n\n\nBeyond language models, Google\u2019s quantum AI and DeepMind have introduced AlphaQubit\u2014a tool that, as reported on November 20 in Nature<\/em>, can reduce quantum errors by 6% to 30%. Such advances are critical because quantum computing faces numerous challenges from heat, vibrations, electromagnetic interference, and even cosmic rays. Today\u2019s quantum machines typically operate with error rates between 1% and 10%, while many applications require error rates well below 0.000000001%.<\/p>\n\n\n\n
\n\n\n\nIV. Competitive Shifts in the AI Arena<\/h3>\n\n\n\n
A. OpenAI\u2019s Uneven Performance & Shifting Market Shares<\/h4>\n\n\n\n
Recent competitive pressures have emerged:<\/p>\n\n\n\n
- \n
- In November, a nonprofit group called METR released an evaluation showing that Anthropic\u2019s Claude Sonnet 3.5 outperformed OpenAI\u2019s o1-preview in five out of seven AI research challenges.<\/li>\n\n\n\n
- According to venture capital data from Menlo Ventures, OpenAI\u2019s share in the enterprise AI market dropped from 50% to 34%, while Anthropic\u2019s share doubled from 12% to 24%.<\/li>\n<\/ul>\n\n\n\n
Despite these challenges, investors remain bullish on OpenAI. Just after o1\u2019s launch, SoftBank announced a $5 billion investment at a $157 billion valuation and has since been aggressively buying shares from OpenAI employees.<\/p>\n\n\n\n
Other companies have also seen significant funding:<\/p>\n\n\n\n
- \n
- xAI<\/strong> announced a $5 billion raise, doubling its earlier valuation to over $50 billion\u2014having raised over $11 billion this year in total.<\/li>\n\n\n\n
- Amazon<\/strong> increased its investment in Anthropic by an additional $4 billion (totaling $8 billion).<\/li>\n\n\n\n
- Writer<\/strong>, founded in 2020, raised $200 million at a $1.9 billion valuation to develop an \u201cauto-evolving\u201d large-model system incorporating a \u201cmemory pool\u201d during training.<\/li>\n<\/ul>\n\n\n\n
B. Video Generation: Sora\u2019s Tepid Reception<\/h4>\n\n\n\n
On the video generation front, OpenAI\u2019s Sora\u2014once anticipated as groundbreaking\u2014has lost much of its initial \u201cwow\u201d factor. On November 26, a group of artists who had early access to Sora gathered on Hugging Face to share its API publicly. Their comments were scathing: although OpenAI offered them free debugging of Sora, critics argued that the company was more focused on public relations than genuine creative expression.<\/p>\n\n\n\n
Meanwhile, competitors continue to move fast. Runway has already launched a video expansion feature, and Tencent\u2019s open-source video model, HunyuanVideo, is clearly positioned as a direct competitor to Sora.<\/p>\n\n\n\n
\n\n\n\nV. Multimodal & Embodied AI: New Funding Waves<\/h3>\n\n\n\n
Several multimodal and embodied AI startups have attracted significant investment in November:<\/p>\n\n\n\n
- \n
- Moonvalley<\/strong> (founded in 2023) raised $70 million in seed funding to develop a \u201ctransparent\u201d video generation model that allows creators to request removal or even compensation if their work is used without consent.<\/li>\n\n\n\n
- Black Forest Labs<\/strong>, an image-generation startup valued at $1 billion, secured $200 million. They are behind the text-to-image model Flux\u2014a tool notably popular on Telegram.<\/li>\n\n\n\n
- Physical Intelligence<\/strong>, another 2023 startup, raised $400 million at a $2.4 billion valuation to develop brains for robots by integrating general AI with physical devices\u2014the first model is called \u03c00.<\/li>\n\n\n\n
- Additionally, Yinhe General<\/strong> (\u94f6\u6cb3\u901a\u7528) and Xinghaitu<\/strong> (\u661f\u6d77\u56fe), both founded in 2023, raised hundreds of millions in RMB to build advanced robot models focused not on humanoid shapes but on adaptable, general intelligence.<\/li>\n<\/ul>\n\n\n\n
\n\n\n\nVI. The GPU Arms Race: Centralizing Compute Power<\/h3>\n\n\n\n
Big tech companies are in a fierce race to concentrate the most GPUs under one roof. At this year\u2019s Goldman Sachs private conference, bankers noted that:<\/p>\n\n\n\n
- \n
- While mergers and acquisitions (particularly tech deals worth over $500 million) are increasingly led by private equity, big companies are spending significantly more on data centers\u2014with capital expenditures more than doubling.<\/li>\n\n\n\n
- Only Amazon, Microsoft, Meta, and Google alone plan on spending over $200 billion this year on infrastructure.<\/li>\n<\/ul>\n\n\n\n
Anthropic CEO Dario Amodei recently predicted that by 2026, computing clusters costing over $10 billion will emerge\u2014with some companies dreaming of clusters worth $100 billion. On November we saw a policy proposal from OpenAI calling for a \u201cNorth America AI Compact\u201d to build a data center that might cost as much as $100 billion.<\/p>\n\n\n\n
Elon Musk made headlines on July by transforming an appliance factory into a cluster of 100,000 H100 GPUs in just 122 days\u2014a pace that Nvidia CEO Jensen Huang described as nearly unparalleled in the industry. Reports in November also detailed how Sam Altman had a heated discussion with a Microsoft infrastructure executive after seeing Musk\u2019s announcement on X (formerly Twitter), worried that xAI might soon deploy an even larger and faster cluster. Rival companies have even resorted to flying helicopters to capture aerial footage of Musk\u2019s data center construction.<\/p>\n\n\n\n
Nvidia and Others Push the Envelope<\/h4>\n\n\n\n
Nvidia is standing at the center of the datacenter arms race. In a bold move, they accelerated their GPU upgrade cycle from every two years to annually. Despite initial delays with their Blackwell series\u2014which were postponed from a planned mid-year release until November due to overheating issues in custom server racks\u2014Nvidia\u2019s expansion plans continue. In November, the chip startup Enfabrica raised $115 million in Series C funding to develop network architecture chips to better interconnect GPUs\u2014positioning themselves as potential competitors in the networking space. On the same day at Nvidia\u2019s Japan summit, Jensen Huang playfully teased SoftBank CEO Masayoshi Son, highlighting the high stakes in this hardware race. Meanwhile, competitors like Graphcore (backed by SoftBank) are aggressively expanding their teams, increasing headcount by 20% in just four months.<\/p>\n\n\n\n
Nvidia isn\u2019t stopping there\u2014they\u2019ve announced plans to release a chip tailored for robotics, codenamed Jetson Thor, in the first half of next year.<\/p>\n\n\n\n
![\"\"](\"https:\/\/blog.nexussup.com\/wp-content\/uploads\/2025\/04\/640-2-1-1024x537.png\")
<\/figure>\n\n\n\n
\n\n\n\nVII. Applications on the Rise: From Coding to Agents<\/h3>\n\n\n\n
Big models have already become an integral part of daily work and life. Recent statistics underline this adoption:<\/p>\n\n\n\n
- \n
- Menlo Ventures reported in November that enterprise spending on generative AI surged 500% this year, reaching $13.8 billion.<\/li>\n\n\n\n
- OpenAI revealed that ChatGPT\u2019s weekly active users hit 250 million in November\u2014only rivaled by apps like TikTok. By early December, that number had grown to 300 million.<\/li>\n\n\n\n
- Originality AI found that since 2018, 54% of long-form posts on LinkedIn might be AI-generated.<\/li>\n<\/ul>\n\n\n\n
A fascinating report from Slack\u2014surveying 17,000 employees in 15 countries\u2014showed that by August, 36% of respondents used AI at work, a 16-percentage-point jump since January 2023. The five most common scenarios included:<\/p>\n\n\n\n
- \n
- Sending messages as leaders<\/li>\n\n\n\n
- Messaging colleagues<\/li>\n\n\n\n
- Evaluating subordinates\u2019 performance<\/li>\n\n\n\n
- Composing emails to clients<\/li>\n\n\n\n
- Brainstorming ideas<\/li>\n<\/ol>\n\n\n\n
Interestingly, nearly half (48%) of those who do use AI at work keep it hidden from their bosses out of concern that they might be perceived as lazy or deceitful. In contrast, Apple\u2019s November \u201cApple Intelligence\u201d advertisement openly celebrates using AI for tasks like writing impressive emails and handling unexpected meeting challenges\u2014even if it sparked a backlash on social media, leading Apple to close the comments on YouTube.<\/p>\n\n\n\n
Even corporate executives are not immune. A survey by Wharton\u2019s management school and GBK found that nearly 72% of senior decision-makers use generative AI at least once a week\u2014doubling from the previous year. In software development, AI coding assistants are revolutionizing the field: Google CEO Sundar Pichai noted that over 25% of new code is AI-generated, while Microsoft executives revealed that GitHub Copilot contributed to nearly half of the startup scripts in their applications. However, even with these advances, seasoned programmers still report that overreliance on AI can introduce a significant number of bugs.<\/p>\n\n\n\n
The excitement in AI coding continues to drive investment. In November, OpenAI integrated its desktop ChatGPT with major IDEs such as VS Code, Xcode, TextEdit, and Terminal. This seamless integration means developers can call upon ChatGPT for code processing without the hassle of copying and pasting code manually.<\/p>\n\n\n\n
Meanwhile, startups focused on AI coding are attracting robust funding. Two noteworthy companies received over $50 million in investments:<\/p>\n\n\n\n
- \n
- Tessl<\/strong> (founded in 2024, with a current valuation of $750 million) aims to create an AI capable of writing software and is planning a product launch early next year.<\/li>\n\n\n\n
- Lightning AI<\/strong> (founded in 2019) also secured $50 million to streamline processes for AI development.<\/li>\n<\/ul>\n\n\n\n
Some domestic investors have disclosed plans to back AI coding tools. For instance, after departing from Noisee in late September, 1998-born Ming Chaoping started his own AI coding venture, which reportedly wrapped up two funding rounds between October and November.<\/p>\n\n\n\n
Agents: The Next Frontier<\/h4>\n\n\n\n
From OpenAI to Apple, various strategies are being deployed to integrate big models into everyday applications. Companies are focusing on enabling \u201cAgents\u201d\u2014intelligent systems that function as orchestrators, capable of understanding user requirements and autonomously coordinating databases and tools to complete complex tasks.<\/p>\n\n\n\n
OpenAI\u2019s GPT-4 technical report, released in March last year, even showcased an early demonstration of a model fabricating its own visual impairment to solicit help in deciphering CAPTCHAs from gig workers. Since then, many companies (especially in China) have announced \u201cAgent\u201d systems\u2014though most are essentially chatbots with a superficial layer of additional functionality. In one rough tally, the number of \u201cagent\u201d assistants in Chinese big-model products numbered in the dozens, if not hundreds.<\/p>\n\n\n\n
The real breakthrough might come from the next generation of Agent products. Anthropic led the way in October, demonstrating how Claude could operate a computer much like a human\u2014browsing for information, planning trips (like finding the best spot to watch the sunrise at the Golden Gate Bridge), and more. By November, announcements piled up: OpenAI\u2019s internal talks hinted at an \u201cOperator\u201d agent set for a January launch, capable of coding or planning trips. Meanwhile, Chinese big-model companies like Zhipu released AutoGLM, which claims to execute tasks across multiple apps on smartphones\u2014sometimes involving over 50 steps. Anthropic even introduced a \u201cModel Context Protocol\u201d intended to guide how agents extract information from business tools, software, or databases, pushing the competition into a new realm.<\/p>\n\n\n\n
Several Agent startups also attracted significant funding:<\/p>\n\n\n\n
- \n
- \/dev\/agents<\/strong> (founded in 2024) raised $56 million at a $500 million valuation to develop an operating system for Agents that work across phones, laptops, and even cars, theorizing that if Agents become as ubiquitous as apps, a dedicated OS (akin to Android or iOS) will be necessary.<\/li>\n\n\n\n
- Rox<\/strong> (founded in 2024) secured $50 million in funding with a modest team of 15, focusing on fully automated AI Agents for sales and customer service.<\/li>\n\n\n\n
- 11x<\/strong> (founded in 2022) raised $50 million at a $320 million valuation to develop Agents that automate end-to-end workflows, freeing users to focus on more important tasks\u2014with annual recurring revenue near $10 million.<\/li>\n\n\n\n
- Cresta<\/strong> (founded in 2017) raised $125 million and is developing AI software to enhance call center communications and automate routine tasks.<\/li>\n\n\n\n
- Pyramid Analytics<\/strong> (founded in 2008) raised $50 million to automate data preparation and analytics, reducing human intervention while boosting accuracy.<\/li>\n<\/ul>\n\n\n\n
Y Combinator partners have observed a flood of Agent applications in Silicon Valley, targeting diverse fields such as recruitment, onboarding, digital marketing, customer support, quality assurance, and debt management. They even speculate that vertical AI Agents may become a new SaaS segment, with over 300 unicorns emerging in the space.<\/p>\n\n\n\n
\n\n\n\nVIII. Beyond Big Models: Expansion into Mobility and Pharma<\/h3>\n\n\n\n
Unmanned Taxis and AI-Driven Drug Discovery<\/h4>\n\n\n\n
Big models aren\u2019t only reshaping digital workflows\u2014they\u2019re making an impact in the physical world as well:<\/p>\n\n\n\n
- \n
- Autonomous Taxis:<\/strong> Waymo has expanded its driverless taxi service throughout Los Angeles. Previously, 300,000 people lined up for rides; in San Francisco, in August, Waymo completed an average of 8,800 daily trips\u2014surpassing the city\u2019s average taxi trips of 6,307.<\/li>\n\n\n\n
- AI Pharmaceutical Startups:<\/strong> In the realm of AI-driven drug discovery, notable funding rounds include:\n
- \n
- Cradle<\/strong> (founded in 2021) raised $73 million to leverage AI for speeding up the discovery of drug-like molecules tailored to specific needs such as high-temperature tolerance.<\/li>\n\n\n\n
- Enveda<\/strong> (founded in 2019) secured $130 million to employ AI in finding therapeutic compounds\u2014currently exploring 10 different molecules aimed at treating conditions like eczema and inflammatory bowel disease, with a Phase I trial underway for an oral drug targeting atopic dermatitis or eczema.<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n
A Few Final Developer Hacks<\/h4>\n\n\n\n
As a side note, many developers have adopted clever techniques to maximize the potential of big models. Instead of relying on a single, expensive model to solve a problem outright, they use multiple models in tandem. A typical approach is to feed background information into an open-source model (such as Llama or Mistral) to retrieve and summarize key details, and then pass the refined summary to a more powerful model for processing\u2014saving costs without sacrificing quality. One popular prompt that appears to boost model performance is:
“if you don\u2019t give me the correct answer, I will be fired.”<\/strong><\/p>\n - Enveda<\/strong> (founded in 2019) secured $130 million to employ AI in finding therapeutic compounds\u2014currently exploring 10 different molecules aimed at treating conditions like eczema and inflammatory bowel disease, with a Phase I trial underway for an oral drug targeting atopic dermatitis or eczema.<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n
- AI Pharmaceutical Startups:<\/strong> In the realm of AI-driven drug discovery, notable funding rounds include:\n
- Rox<\/strong> (founded in 2024) secured $50 million in funding with a modest team of 15, focusing on fully automated AI Agents for sales and customer service.<\/li>\n\n\n\n
- Lightning AI<\/strong> (founded in 2019) also secured $50 million to streamline processes for AI development.<\/li>\n<\/ul>\n\n\n\n
- Tessl<\/strong> (founded in 2024, with a current valuation of $750 million) aims to create an AI capable of writing software and is planning a product launch early next year.<\/li>\n\n\n\n
- Black Forest Labs<\/strong>, an image-generation startup valued at $1 billion, secured $200 million. They are behind the text-to-image model Flux\u2014a tool notably popular on Telegram.<\/li>\n\n\n\n
- Amazon<\/strong> increased its investment in Anthropic by an additional $4 billion (totaling $8 billion).<\/li>\n\n\n\n
- xAI<\/strong> announced a $5 billion raise, doubling its earlier valuation to over $50 billion\u2014having raised over $11 billion this year in total.<\/li>\n\n\n\n
- Higher Precision Data:<\/strong>
- Ilya Sutskever<\/strong>, co-founder and former chief scientist at OpenAI, noted, \u201cThe 2010s were all about scaling up; now we\u2019re back to needing breakthroughs and new discoveries.\u201d<\/li>\n<\/ul>\n\n\n\n