Is Google now ahead in the AI race?







Every funding round, every investor, and what the labs promised in return, in one document. Have a look ↓
Summary
Yes, Google is now ahead in the AI race overall, although ChatGPT remains the most popular standalone assistant and Anthropic still leads in several demanding professional tasks.
Google's biggest achievement is that it no longer has to win every model benchmark to win commercially. It builds frontier AI models, designs its own chips, operates cloud infrastructure, and distributes AI through products already used by billions of people. Winning commercially without topping every benchmark is why we check who leads on benchmarks, and who only claims it, in our frontier AI labs report.
Gemini 4 Argon has pushed Google back to the top of important general AI rankings, but the technical race remains split. OpenAI leads on FrontierSWE v2, while Anthropic performs better on several terminal-based and machine-learning engineering evaluations.
Gemini's consumer growth is substantial, but the usage gap remains significant. Google's assistant passed 1 billion monthly active users, while ChatGPT exceeded 1 billion weekly active users. ChatGPT also attracted roughly 56.4% of leading AI assistant website visits in September 2026, against Gemini's 25.0%.
The biggest surprise is that AI has strengthened Google's existing businesses rather than destroyed them. Search revenue grew 17% in the second quarter of 2026, even as Google expanded AI Overviews and AI Mode. The feared collapse of search advertising has simply not materialized.
Google Cloud is becoming the clearest commercial proof of Google's AI momentum. Quarterly revenue jumped 82% to $24.8 billion, far faster than AWS or Azure, while contracted backlog climbed to $514 billion. Google is still smaller than those competitors in cloud revenue, but the growth gap is striking.
Google's custom TPUs create another advantage that is easy to underestimate. The company can optimize its own models and infrastructure together, then sell computing capacity and hardware systems to customers that may use competing AI models. In other words, Google can profit even when somebody chooses Claude instead of Gemini.
Enterprise adoption is broad but harder to interpret. Nearly 90% of Fortune 100 companies reportedly use Gemini Enterprise, yet that does not establish Google as their primary AI provider. Large organizations can use Gemini, Claude, and OpenAI simultaneously for different jobs.
The biggest threat to Google's position may come from its own spending. Alphabet raised its 2026 capital expenditure forecast to $195 billion–$205 billion, while free cash flow fell sharply. Rapid cloud growth supports the investment, but the eventual returns are still an open question.
Google's lead is therefore more structural than technical. OpenAI could release a better model tomorrow, and Anthropic could remain the preferred coding platform, without erasing Google's advantages in Search, cloud infrastructure, custom hardware, distribution, and cash generation. Google has not permanently won the AI race, but it currently has the strongest overall position to profit from it.
Written by the New Market Pitch teamWe track new markets so founders and investors can move faster. About us
70+pages
This is just a preview page. Don’t worry if a number looks old: the report gets refreshed, and you’ll always get the latest version.
Why is Google suddenly looking like the AI leader again?
Google is now one of the strongest candidates to lead the AI race, thanks to the combination of its new Gemini models, fast-growing cloud business, and enormous reach across Search, Android, and other products.
The shift has been dramatic. Google was initially seen as one of the companies most threatened by ChatGPT. OpenAI had captured the consumer market, Anthropic was becoming a favorite among developers, and Microsoft was using its OpenAI partnership to strengthen Azure.
Google has since caught up in model performance and accelerated commercially. Gemini 4 Argon reached the top of LMArena's Text leaderboard, the Gemini app passed 1 billion monthly active users, and Google Cloud revenue jumped 82% year over year to $24.8 billion in the second quarter of 2026. Gemini climbing to the top of a leaderboard shows how much changed in a year, which we go through in our report on frontier AI labs.
The surprise is how much of Google's existing business has benefited from AI. Search advertising is still growing, businesses are spending more on Google Cloud, and Google's custom chips are attracting interest beyond its own infrastructure.
A company once expected to lose customers to AI competitors is now selling models, computing power, and AI products across much of the industry.
What would it take for Google to actually win the AI race?
Google would deserve to be called the overall AI leader if its models remained among the best while its products, infrastructure, and commercial results gave it a stronger position than OpenAI, Anthropic, and Microsoft.
The difficulty is that these companies are winning in different places. OpenAI has the most widely used standalone AI assistant. Anthropic is particularly strong in professional coding. Google operates its own computing infrastructure and distributes AI through products already used by billions of people.
Popularity and economic success also need to be separated. A company can attract hundreds of millions of users while spending heavily to serve them. Another can earn substantial revenue supplying the infrastructure those users depend on.
For Google, the strongest evidence would be a sustained combination of frontier model performance, growing AI adoption, and profitable commercial growth. A temporary first-place benchmark ranking is useful, but the bigger question is whether Google can keep turning AI demand into business across several markets.
Is Gemini 4 now better than ChatGPT and Claude?
Google's Gemini 4 Argon is currently among the strongest AI models available, and it beats leading OpenAI and Anthropic systems on several major tests. But Claude remains better at some coding tasks, and the overall technical ranking is still close.
Google's published benchmark comparisons gave Gemini 4 Argon the highest result in 13 of 19 tests, with one tie and five losses. Its strongest areas included complex knowledge work, long-context understanding, video analysis, and several software engineering evaluations.
Independent results also support its position. On LMArena's Text leaderboard, Gemini 4 Argon reached approximately 1,532 points in the October 20 snapshot, ahead of Meta's Muse Wonder and Claude Opus 5.5.
However, LMArena measures human preferences between model responses. A model that produces answers people prefer does not necessarily write better production code or operate software tools more reliably.
On FrontierSWE v2, OpenAI's GPT-6 Astra scored 65.5%, compared with 55.0% for Gemini 4 Argon. On Terminal-bench 4.0, Claude Opus 5.5 reached 66.4%, against Google's 57.4%.
Google has clearly returned to the technical frontier. The remaining differences matter most to customers who need a model for a specific job rather than general conversation.
70+pages
Just a preview. If a number here looks old, no worries: your report comes with the latest data.
Which AI company actually wins the toughest benchmarks?
Google currently performs especially well in general responses and several knowledge-intensive tasks, while OpenAI and Anthropic still lead important software engineering evaluations.
The most revealing evidence comes from comparing the same models across different tests. Google's Gemini 4 Argon scored 77.9% on DeepSWE v1.1, ahead of Claude Opus 5.5 at 74.2% and GPT-6 Astra at 74.1%.
But Google lost that advantage on FrontierSWE v2. OpenAI scored 65.5%, Anthropic reached 62.3%, and Gemini finished at 55.0%. A model that wins one test and loses the next is why we check who leads on benchmarks and who only claims it in our frontier AI labs report.
Anthropic also performed better on Terminal-bench 4.0 and PostTrainBench, which test command-line work and machine-learning engineering. Those tasks are closely related to the autonomous coding agents companies increasingly want to deploy.
The pattern is pretty clear: Google's lead is strongest in broad model capabilities, while specialist tasks continue to produce different winners. Calling any one of these models the best at everything would be a stretch.
The figures below combine Google's published evaluations with publicly reported leaderboard results. Scores are comparable within each test, but not between different tests.
| Benchmark | Gemini 4 Argon | Leading competitor | Who is ahead? |
|---|---|---|---|
| LMArena Text | ~1,532 | Meta Muse Wonder: ~1,509 | |
| DeepSWE v1.1 | 77.9% | Claude Opus 5.5: 74.2% | |
| FrontierSWE v2 | 55.0% | GPT-6 Astra: 65.5% | OpenAI |
| Vibe Code Bench | 91.9% | Claude Opus 5.5: 90.3% | |
| Terminal-bench 4.0 | 57.4% | Claude Opus 5.5: 66.4% | Anthropic |
| PostTrainBench | 45.3% | Claude Opus 5.5: 49.3% | Anthropic |
| CWE-bench v1 | 68.0% | GPT-6 Astra: 68.0% | Tie |
| LMArena WebDev | ~1,678 | Claude Opus 5.5: ~1,806 | Anthropic |
Has Google Gemini caught ChatGPT in popularity?
Gemini is growing quickly, but ChatGPT remains the clear leader in standalone AI assistant usage.
Google announced that Gemini had passed 1 billion monthly active users in August 2026, up from 950 million reported the previous month. OpenAI reported more than 1 billion weekly active ChatGPT users around the same period.
That difference in reporting periods is important. Monthly active users only need to use a product once during the month, while weekly active users return within a much shorter period. We cannot calculate an exact engagement gap without directly comparable figures, but ChatGPT's reach remains impressive.
Similarweb's website traffic estimates provide a more consistent comparison. In September 2026, ChatGPT accounted for approximately 56.4% of visits to leading AI assistant websites, compared with 25.0% for Gemini.
Eleven months earlier, ChatGPT had 71.8% and Gemini just 13.6%. Google's share nearly doubled while ChatGPT lost more than 15 percentage points.
Yet Gemini's progress has recently slowed. Its share peaked near 27.8% in May before falling to 25.0% in September. ChatGPT recovered from 52.7% to 56.4% over that same period.
The consumer race is much closer than it was a year ago, although the latest traffic data suggests ChatGPT is defending its lead rather than continuing to lose ground every month.
Are people actually using Gemini, or just seeing it inside Google products?
Gemini has attracted genuine user demand, although Google's existing products make its audience figures harder to interpret than those of a standalone assistant.
Google reported that the Gemini app's daily active users had tripled over the preceding year. The company also disclosed that 63% of users interacted with Gemini through voice and that the system was generating more than 150 million images daily.
Those are signs of people using Gemini for actual tasks, beyond simply encountering an AI-generated answer inside Google Search.
At the same time, Google's reach across other products is extraordinary. AI Mode in Search passed 1 billion monthly users, while Ask YouTube reached more than 140 million users on video watch pages during June 2026.
We cannot add those audiences together, since many people use several Google services. Nor should we assume that someone trying Ask YouTube has the same level of attachment as a paying ChatGPT or Claude subscriber.
Still, the combination is difficult for rivals to match. Google can reach people through products they already use and gradually introduce more advanced AI features without asking them to adopt an entirely new service.
70+pages
This is a preview page. Spotted an outdated number? That’s normal here. The report you get is always the freshest version.
Could Google win the AI race simply because it owns Search and Android?
Google's control of Search, Android, Chrome, Gmail, and YouTube gives Gemini a huge distribution advantage that OpenAI and Anthropic cannot easily match.
Google can place AI features directly inside products with billions of existing users. ChatGPT and Claude depend more heavily on people choosing their applications, subscribing to their services, or accessing them through partners. Owning Search and Android is a big reason we ask who the likely future winners are in our report covering frontier AI labs.
This advantage becomes particularly important when users have no strong preference between models. Someone already working in Gmail or Google Docs may find Gemini convenient enough without opening a separate assistant.
However, Google faces limits on how aggressively it can exploit that position. In September 2025, a US federal court imposed remedies restricting certain exclusive distribution arrangements involving Search, Chrome, Assistant, and Gemini.
European regulators have also scrutinized Google's ability to use its established platforms to favor its own services.
These restrictions could make distribution more competitive, particularly if Google were prevented from using exclusive agreements or preferential placement to steer users toward Gemini.
For now, though, Google retains an extraordinary advantage: it can introduce AI into everyday consumer behavior at a scale few standalone applications can approach. And plenty of users will take the assistant already sitting in front of them.
Is ChatGPT actually hurting Google's search business?
ChatGPT has not caused the collapse of Google Search that many expected, and Google's latest financial results show its advertising business continuing to grow.
Alphabet reported 17% year-over-year growth in Google Search and other revenue during the second quarter of 2026. Its broader Google Services business increased 15% to $94.5 billion.
Those figures are significant because Google had already faced several years of concern about AI assistants replacing traditional searches.
Instead, Google has introduced AI Overviews and AI Mode while maintaining growth in its core business. Management has also said these features are generating additional search activity.
The remaining concern is profitability. Generating detailed AI answers requires more computing power than displaying conventional search results. Google has reported reducing the cost of AI Mode responses through hardware and software improvements, but it has not published enough information to compare the profit margins of AI searches against traditional searches.
We also lack a clean measure of how many queries have shifted from conventional search pages to AI-generated answers.
Nevertheless, after several quarters of growing Search revenue alongside expanding AI usage, the immediate threat of a major collapse in Google's advertising business looks considerably weaker than it did when ChatGPT first became popular.
Is Google Cloud growing faster than AWS and Microsoft Azure?
Google Cloud is currently growing much faster than AWS and Azure, although Amazon and Microsoft still operate larger cloud businesses.
Alphabet's second-quarter 2026 results showed Google Cloud revenue increasing 82% to $24.8 billion. That followed 63% growth in the preceding quarter, suggesting Google's cloud business is accelerating rather than merely enjoying a favorable comparison with the previous year.
AWS generated approximately $42.2 billion in quarterly revenue, growing 37%. Microsoft's Azure business grew by roughly 42% to 43%.
Google therefore grew about twice as fast as its largest competitors, although AWS still generated around 70% more quarterly revenue.
Another useful figure is Google's cloud backlog. It reached $514 billion, up from $462 billion just one quarter earlier. That $52 billion increase represents around 11% growth in contracted future commitments in only three months.
Backlog includes contracts that may be recognized over several years, so it should not be treated as current revenue. Still, combined with accelerating sales, it shows that customers are committing more money to Google's infrastructure.
Google Cloud also includes Workspace and non-AI services, and its growth reflects several factors beyond Gemini. But the overall pattern strongly supports the view that Google is becoming one of the main beneficiaries of rising AI infrastructure spending.
| Cloud business | Quarterly revenue | Annual growth | Annualized revenue |
|---|---|---|---|
| Amazon AWS, Q2 2026 | $42.2B | 37% | $168.8B |
| Microsoft Intelligent Cloud, Q2 2026 | $39.3B | 32% | $157.2B |
| Microsoft Azure, Q2 2026 | ~$29.4B | ~42–43% | ~$117.6B |
| Google Cloud, Q2 2026 | $24.8B | 82% | $99.2B |
| Google Cloud, Q1 2026 | ~$20.0B | 63% | ~$80.0B |
| Google Cloud, Q2 2025 | ~$13.6B | - | ~$54.4B |
| Google Cloud backlog, Q1 2026 | $462B | - | Not applicable |
| Google Cloud backlog, Q2 2026 | $514B | +11% vs. Q1 | Not applicable |
70+pages
Preview only. Some numbers may have changed since we made this page, but your report will be up to date.
Are big companies choosing Gemini instead of OpenAI and Claude?
Google is winning major enterprise AI customers, but the available figures do not show a large-scale switch away from OpenAI or Anthropic.
During its July 2026 earnings call, Alphabet said nearly 90% of Fortune 100 companies were using Gemini Enterprise. It also reported that more than 9 million developers were building with Gemini models each month across its development platforms and APIs.
Google has named companies such as PepsiCo, Intel, HSBC, Macy's, Bell Canada, and SIGNAL IDUNA among its enterprise customers. Their applications range from internal research and analytics to customer service and operational workflows.
The bigger question is how deeply those organizations depend on Gemini. A company might use Google's AI tools in Workspace while relying on Anthropic for coding and OpenAI for customer-facing applications.
This explains why enterprise adoption figures can look impressive for several competitors simultaneously.
A more convincing indicator of Google's commercial position is customer spending. Alphabet has reported that existing Google Cloud customers were consuming more than 50% above their committed amounts, while customer acquisition had more than doubled year over year.
That suggests businesses are expanding usage after adoption, rather than merely running isolated experiments.
Google is clearly capturing more enterprise AI spending. But without comparable revenue by model provider or workload, we cannot establish that Gemini has become the preferred AI system across large corporations.
Does Google's own AI chip business give it an advantage over OpenAI?
Google's custom TPU chips give it one of the strongest infrastructure advantages in the AI industry, especially compared with companies that depend heavily on external computing providers.
Google has been designing Tensor Processing Units since long before the launch of ChatGPT. Its seventh-generation Ironwood TPU continues that strategy, with hardware built specifically for large-scale machine-learning workloads. Chips built just for machine learning are a technology bet, and we follow which technologies are gaining traction in our frontier AI labs report.
The advantage extends beyond chip design. Google operates the data centers, develops software for those chips, builds frontier models, and distributes AI applications through its own services.
That gives its engineering teams opportunities to optimize several parts of the system together.
A particularly important development came in 2026, when Google started delivering TPU systems directly to customer data centers. Previously, Google's custom accelerators were mainly available through its own infrastructure and cloud services.
Alphabet's management said revenue from those direct system agreements would remain relatively small in 2026, with most of the contracted revenue expected in 2027.
This creates a new business opportunity. Google can sell computing infrastructure to companies developing competing AI models, potentially including businesses that compete directly with Gemini.
Anthropic's existing relationship with Google's cloud and TPU infrastructure already illustrates this overlap.
We still cannot quantify Google's full hardware cost advantage because comparable training and inference economics are not public. But its ability to design and supply its own accelerators gives it flexibility that most independent AI laboratories lack.
70+pages
This is just a preview page. Don’t worry if a number looks old: the report gets refreshed, and you’ll always get the latest version.
Can Google make AI cheaper than OpenAI and Anthropic?
Google is competing aggressively on AI pricing, and its combination of custom chips and smaller Gemini models gives it several ways to lower costs.
Gemini 4 Argon launched with introductory API prices of $2 per million input tokens and $10 per million output tokens. Cached input tokens received a 95% discount, reducing their effective price to approximately $0.10 per million tokens.
Caching can be especially useful for applications that repeatedly access the same instructions, documents, or large codebases.
Google has also been improving its Flash models, which prioritize lower latency and cost over maximum reasoning performance. In August 2026, Gemini 3.7 Flash launched with introductory pricing around half that of the original 3.6 Flash.
This is a recurring part of Google's strategy: offer a powerful flagship model for demanding tasks and cheaper alternatives for high-volume applications.
However, the listed price per token is only part of the equation. A cheaper model that makes more mistakes, requires longer responses, or needs repeated attempts can cost more to complete a job.
We would need independent comparisons of total cost per successful task to identify a clear winner.
Google's infrastructure ownership makes its pricing strategy particularly credible, but whether it has the lowest overall costs remains difficult to prove.
Is Google spending too much money on AI?
Google is spending an extraordinary amount on AI infrastructure, and the biggest financial risk now is whether revenue can grow fast enough to justify those investments.
Alphabet originally projected $180 billion to $190 billion in capital expenditures for 2026, roughly six times its $31 billion spending in 2022.
After its second-quarter results, the company increased its annual forecast to $195 billion to $205 billion because of faster infrastructure deployment.
At the midpoint, $200 billion represents approximately $548 million in spending every day over a full year. That's a staggering amount, even by Google's standards.
The pace is just as striking. Alphabet spent $80.6 billion during the first six months of 2026, compared with $39.6 billion during the equivalent period in 2025. Spending more than doubled in a single year.
Most of the investment is directed toward technical infrastructure, although not every dollar is exclusively for generative AI.
Alphabet can support this spending because its existing businesses generate enormous amounts of cash. The company reported $119.8 billion in second-quarter revenue and an operating margin of 34%.
But spending is growing much faster than revenue. S&P Global's analysis of Alphabet's second-quarter results found that free cash flow had fallen approximately 74% year over year as capital expenditure surged.
Google also expects higher depreciation, energy costs, and data center operating expenses.
We do not yet know whether the current infrastructure expansion will produce attractive returns over its full economic life. Strong cloud demand supports the investment, but rising spending is now the clearest financial weakness in Google's AI leadership case.
70+pages
Just a preview. If a number here looks old, no worries: your report comes with the latest data.
Is Anthropic becoming a bigger threat to Google than OpenAI?
Anthropic is currently the more dangerous competitor for some high-value professional AI tasks, while OpenAI remains Google's bigger rival for mass-market assistant usage.
Claude's performance in software development is particularly important. Coding agents are among the most practical applications of advanced AI because their work can be evaluated through completed tasks, tests, and changes to production software.
Claude Opus 5.5 led LMArena's WebDev leaderboard, and Google's own Gemini 4 Argon comparisons showed Anthropic ahead on several terminal-based and machine-learning engineering tests. Claude topping a web-development leaderboard shows how close the race is, and we measure how far the challengers are behind in our report on who's ahead in frontier AI labs.
That technical advantage connects to a broader trend: businesses are moving from using AI mainly for questions and writing toward systems that can operate software and complete more complicated tasks.
Anthropic has focused heavily on these professional workflows, including Claude Code and related development tools.
OpenAI presents a different challenge. ChatGPT still attracts much more direct assistant traffic than Gemini and has more than 1 billion weekly active users.
Google therefore faces two different battles. With OpenAI, the central question is whether Gemini can become people's preferred everyday assistant. With Anthropic, it is whether Google's models can match the reliability and usefulness of Claude in demanding professional workflows.
Anthropic may be the tougher competitor in coding, but OpenAI remains the company Google must catch to become the most popular standalone AI assistant.
Could Google make more money from AI even if ChatGPT stays number one?
Google could become the biggest commercial winner from AI without Gemini ever overtaking ChatGPT in direct assistant usage.
The reason is that Google can earn money from several parts of the same market. Consumers use AI features inside Search and other Google products. Businesses pay for Google Cloud and Workspace. Developers pay for Gemini APIs, while other AI companies purchase computing capacity.
Google can even benefit financially when a customer chooses a competing model running on its infrastructure.
Alphabet's figures show how different parts of this business are growing together. Search and other advertising revenue increased 17% in the second quarter of 2026, Google Cloud revenue rose 82%, and paid model API usage reached approximately 22 billion tokens per minute.
That API activity was up from 16 billion tokens per minute one quarter earlier, an increase of about 37.5%.
The three measures represent different activities, so we cannot combine them into a single AI revenue figure. But they show how Google's exposure extends across consumer services, developer tools, and enterprise computing.
This is Google's biggest commercial advantage over a standalone AI laboratory.
OpenAI can remain the most popular assistant, and Claude can remain the preferred model for certain coding tasks, while Google continues earning money from the infrastructure and services supporting the wider market.
Can Google's AI lead last, or will OpenAI catch up again?
Google's current technical lead could change quickly, but its underlying advantages in infrastructure, distribution, and research are much harder for competitors to reproduce.
Frontier AI rankings have changed repeatedly as OpenAI, Anthropic, Google, and Meta release new models. Meta's Muse Wonder, for example, moved close to Google's leading Text leaderboard result shortly after Gemini 4 Argon appeared.
That kind of competition makes any claim of permanent model superiority premature.
Google's longer-term advantages have a different history. It has developed TPUs for years, operated massive global infrastructure, and invested heavily in machine-learning research long before generative AI became a mainstream consumer product.
The company's scientific work also strengthens its position. Google DeepMind helped develop AlphaFold, whose protein structure prediction research contributed to the 2024 Nobel Prize in Chemistry awarded to Demis Hassabis and John Jumper.
More recently, Google's work on genomic models, weather forecasting, and AI-assisted scientific research has demonstrated capabilities beyond general-purpose chatbots.
These scientific achievements do not automatically translate into commercial revenue, but they show the depth of Google's research organization.
The more serious long-term question is whether its infrastructure investment will keep producing enough profitable demand. If Google Cloud continues growing rapidly while its AI costs become more manageable, its advantage could strengthen even if OpenAI or Anthropic releases the best model next.
70+pages
This is a preview page. Spotted an outdated number? That’s normal here. The report you get is always the freshest version.
Is Google now ahead in the AI race?
Yes, Google currently has the strongest overall position in the AI race, although OpenAI still leads consumer assistant usage and Anthropic remains ahead in some important professional tasks.
Our judgment rests on the combination of several advantages that competitors have struggled to match simultaneously.
Google is back at the frontier of model performance with Gemini 4 Argon. Its Gemini app has reached more than 1 billion monthly users, while Search, Android, and YouTube give it distribution on an enormous scale. Google Cloud has accelerated to 82% annual growth, and its custom TPU business is beginning to sell infrastructure beyond Google's own data centers.
Alphabet also has a profitable advertising business that helps finance its AI investments, although planned annual capital expenditure of up to $205 billion creates serious pressure to deliver future returns.
OpenAI remains stronger in direct consumer adoption, with ChatGPT accounting for more than half of leading AI assistant website visits. Anthropic continues to outperform Google on certain coding and agentic benchmarks, while Amazon and Microsoft still run larger cloud businesses.
These differences prevent us from declaring Google the winner of every part of AI.
But when we look across the industry, Google has a broader set of advantages than any single rival. It develops frontier models, builds specialized hardware, operates infrastructure, sells enterprise services, and reaches billions of users through established products.
Perhaps the most revealing change is what happened to Google's core business. Generative AI was initially expected to undermine Search and threaten the company's dominance. So far, Google has continued growing Search revenue while turning AI into an additional driver of cloud growth and product adoption.
Google has not won the AI race permanently. For now, though, it looks like the company best positioned to profit from AI across the largest number of markets, even if ChatGPT remains more popular and Claude remains better for particular jobs.
70+pages
Preview only. Some numbers may have changed since we made this page, but your report will be up to date.
Our methodology
This analysis assesses whether Google now leads the AI race by comparing its position with OpenAI, Anthropic, Microsoft, Amazon, and other major AI developers. We examine frontier model performance, consumer adoption, product distribution, enterprise demand, cloud infrastructure, financial performance, investment capacity, and long-term competitive advantages.
We used recent company disclosures, official financial reports, independent usage measurements, technical benchmarks, and institutional records. We prioritized primary sources and direct measurements, looking at how the competitive position of each company has changed rather than relying on individual product announcements or short-lived leaderboard rankings.
For model performance, we compared Gemini 4 Argon with leading OpenAI, Anthropic, and Meta models using Google's published benchmark evaluations and independent leaderboards. We treated each benchmark as a measurement of a particular capability rather than combining unrelated scores into an overall ranking. The analysis distinguishes human-preference tests from software engineering, terminal operation, and other agentic evaluations.
For consumer adoption, we examined active-user disclosures from Google and OpenAI alongside website traffic estimates. We kept monthly and weekly active-user figures separate because they measure different levels of engagement. We also distinguished direct Gemini app usage from AI features integrated into Search, YouTube, and Google's other products to avoid treating overlapping audiences as unique users.
For commercial performance, we compared Alphabet's second-quarter 2026 results with Amazon's AWS figures and Microsoft's cloud disclosures. We considered revenue growth, business scale, cloud backlog, enterprise adoption, API activity, and investment requirements. Azure revenue estimates were kept separate from Microsoft's broader Intelligent Cloud segment, while contracted backlog was treated as future revenue commitments rather than current sales.
We also examined Google's custom TPU infrastructure, official Gemini API pricing, enterprise customer adoption, and the effect of AI on its Search business. Investment risk was assessed using Alphabet's capital expenditure guidance, operating performance, and reported free cash flow trends. We distinguished announced spending commitments and management expectations from realized financial outcomes.
Our final assessment considers the strength and durability of Google's advantages across these different areas. We did not assign an arbitrary numerical score or assume that the best general-purpose model must belong to the overall industry leader. Instead, we weighed technical performance against adoption, infrastructure ownership, commercial growth, profitability, and competitive position.
Key sources include Google DeepMind's Gemini 4 Argon model card, Arena's Text leaderboard, Arena's WebDev leaderboard, Terminal-Bench's official evaluations, and Anthropic's Claude Opus documentation for model comparisons.
For adoption and financial performance, we used Google's Gemini usage announcement, OpenAI's ChatGPT adoption disclosures, Alphabet's second-quarter 2026 financial results, Alphabet's earnings commentary, Amazon's second-quarter 2026 results, and Microsoft's fiscal 2026 fourth-quarter results.
Additional sources include Google I/O 2026 announcements, Alphabet's June 2026 investor presentation, Google's Ironwood TPU announcement, official Gemini API pricing, Google's August 2026 AI updates, the US Department of Justice's search remedies announcement, and the Nobel Prize's 2024 Chemistry announcement.
Written by the New Market Pitch teamWe track new markets so founders and investors can move faster. About us









