According to the 2025 Talent Report released by SignalFire, the demand for new graduates from tech companies has dropped significantly, while AI labs are competing fiercely for top talent. Anthropic is a leader in AI with an 80% employee retention rate, and the culture and strategy behind its success are worth delving into. At the same time, the geographical distribution of scientific and technological talents is also being reshuffled, and the status of traditional science and technology centers is challenged by emerging cities. This article will provide an in-depth analysis of these trends, exploring the future direction of the talent landscape in the tech industry and how businesses and job seekers are responding to this change.
Have you ever wondered that computer science majors might no longer be the golden ticket to the tech industry?
In the past, technology companies opened their arms to new graduates, but now this door is gradually closing.
Over the past few years, we have witnessed a shocking phenomenon: while the tech industry as a whole is still growing, employment opportunities for new graduates have shrunk significantly.
According to SignalFire’s latest 2025 Talent Report, hiring for new graduates in the tech industry has dropped by 50% compared to pre-pandemic levels. This is not a simple adjustment of the economic cycle, but a fundamental shift that the entire industry is experiencing.
I delved into the report and found that the reasons behind this are more complex than they seem.
SignalFire tracks over 650 million professionals and 80 million organizations through its Beacon AI platform, providing us with unprecedented industry insights.
After 10 years of interaction design, why did I transfer to product manager?
After the real job transfer, I found that many jobs were still beyond my imagination. The work of a product manager is indeed more complicated. Theoretically, the work of a product manager includes all aspects of the product, from market research, user research, data analysis…
View details >
What they reveal is not just a decline in recruitment numbers, but a restructuring of the entire tech talent ecosystem.
From the war for talent in AI labs to the reshuffle of geographic centers to the emergence of emerging roles, this report paints a complete picture of the talent landscape in the tech industry that is undergoing a seismic shift.
What strikes me the most is that even graduates from top computer science programs are not immune to this change.
In the past, a computer science degree from a prestigious university was almost a passport to major companies such as Google, Apple, Microsoft, etc., but now the recruitment rate of new graduates from these companies has dropped from double digits to single digits.
At the same time, we’re seeing some surprising trends: Anthropic leads AI labs with 80% employee retention, Texas’ tech hub status is declining, and emerging tech cities like Miami and San Diego are rapidly emerging.
These changes are not only affecting job seekers’ choices, but are also reshaping the future of the entire tech industry.
The “empirical paradox” encountered by new graduates
As I analyze this data, what worries me the most is the stark reality faced by new graduates.
Data shows that only 7% of new hires at big tech companies are now fresh graduates, down 25% from 2023 and more than 50% from pre-pandemic 2019. Startups are no better, with new graduates accounting for only 6% of hiring, down more than 30% from pre-pandemic levels. This decline is not a temporary adjustment, but a structural shift.
What is even more worrying is the phenomenon of “empirical paradox” reflected behind this trend.
I find that today’s tech employers are no longer looking for potential, but for proof. This puts new graduates in a typical paradox: you need experience to get a job, but you need a job to get experience. In an environment with leaner teams and tighter budgets, few companies are willing to invest in training new talent.
What’s even more brutal is that many companies post entry-level positions only to end up filling them with senior individual contributors, which is known as the “empirical paradox” phenomenon.
The latest data from the Federal Reserve Bank of New York shows that the unemployment rate for new college graduates has risen by 30% since bottoming out in September 2022, while the unemployment rate for all workers has risen by only about 18%.
This gap is not only reflected in the numbers, but also in the shift in employers’ attitudes.
The report shows that 55% of employers believe Gen Z has difficulty working in teamwork, and 37% of managers say they would rather use AI than hire Gen Z employees.
This cognitive bias further exacerbates the employment dilemma of new graduates.
I especially noticed that even graduates from top computer science programs are not immune to this dilemma.
In the past, a computer science degree from Stanford, MIT, or Carnegie Mellon was almost a guarantee of entry into Silicon Valley’s big companies, but now even these elite graduates are struggling to break through the barriers of the tech industry.
Since 2022, the percentage of new graduates who have found jobs at the “Big Seven” (Alphabet, Amazon, Apple, Meta, Microsoft, NVIDIA, and Tesla) has fallen by more than half. It’s not just a matter of hiring slowdown, but a fundamental shift in talent expectations.
I think there are multiple factors behind this change.
AI does take on some routine tasks that were previously done by junior employees, but the bigger driver may be the end of the “free money frenzy” driven by low interest rates in 2020-2022 and the resulting overhiring and inflation. Now, with tighter budgets and shorter development cycles, companies are hiring more streamlined and later.
Carta’s data shows that Series A tech startups are 20% smaller than they were in 2020. This shift is not just a reduction in hiring, but a reset of hiring philosophy. As AI tools take on more and more routine entry-level tasks, companies are starting to prioritize positions that can provide high-leverage technical output.
Anthropic’s Talent Defense: The Secret Behind 80% Retention
In the battle for AI talent, one company has impressed me: Anthropic.
While the entire tech industry is struggling with brain drain, Anthropic stands out with an 80% employee retention rate.
This figure stands out in an industry known for its high turnover rates.
Know that 80% of employees who have been employed for at least two years remain with the company at the end of the second year, which is almost miraculous in the tech industry.
To better understand the significance of this number, I compared data from other AI labs. DeepMind is a close second, with a retention rate of 78%, which is also pretty good.
But what surprised me was that OpenAI’s retention rate was only 67%, which while comparable to large FAANG companies like Meta (64%), it appeared relatively low in AI labs. Considering OpenAI’s popularity and influence in the public mind, this result is truly unexpected.
I delved into the reasons for Anthropic’s success and found that their strength lies not only in talent retention but also in their precise strategies to attract talent.
Data shows that engineers are 8 times more likely to jump ship from OpenAI to Anthropic than in the reverse flow. The ratio of flows from DeepMind to Anthropic is nearly 11:1.
While this is partly due to Anthropic’s appeal as a popular startup and the natural turnover of DeepMind’s larger, more senior team, this difference in turnover is still alarming.
What intrigued me more was Anthropic’s cultural strength.
Unlike big tech companies that rely on high salaries and brand appeal, Anthropic’s strength lies in its unique culture that embraces unconventional thinkers and gives employees real autonomy to make an impact.
Employee feedback shows that Anthropic offers flexible work options, with no title politics or mandatory management tracks, and only clear career paths. Employees say Anthropic embraces intellectual discussion and researcher autonomy, making it a magnet for AI talent bound by bureaucracy elsewhere.
I also noticed an interesting phenomenon: Claude is quickly becoming a darling in informal measures for developers, and this product affinity can influence career decisions.
Engineers tend to gravitate towards companies that choose products they admire and use, and this perceived product resonance may give Anthropic an advantage in hiring. When your product is widely recognized and used by the developer community, it becomes easier to attract these developers to your team.
Anthropic’s success isn’t limited to poaching people from competitors.
Big tech companies have also become major talent hubs, with Google, Meta, Microsoft, Amazon, and Stripe being major talent pools in AI labs, and Anthropic has been particularly successful in tapping into senior researchers and engineers from these companies.
This talent mobility model reflects an important trend across the industry: the best AI talent is gathering towards companies that offer more autonomy, a clearer vision, and a better working environment.
Reshuffle the geographical map of science and technology
When analyzing the geographical distribution of tech talent, I found some unexpected trends.
While San Francisco and New York remain undisputed tech hubs, a new power landscape is taking shape, with some traditional tech hotspots losing their luster and others rapidly emerging.
This geographical restructuring is not just a simple turnover but reflects a rethinking of work-life balance, cost-effectiveness, and growth opportunities in the tech industry.
What struck me the most was the decline in Texas’ status as a tech hub.
Austin, once a leader in post-pandemic tech growth, saw a 6% drop in venture-backed startup employees in 2024.
Houston saw an even greater decline, reaching 10.9%. There are multiple reasons behind this change: lagging infrastructure, cultural mismatch, fluctuating housing prices, and a renewed emphasis on hybrid return-to-office policies that are driving startup employees to live closer to traditional tech hubs.
I am particularly concerned about the impact of this change on the Texas tech ecosystem.
Once known as the “Silicon Mountain”, Austin attracts a large number of technology companies and talent, enjoying a lower cost of living, a friendly business environment and a unique cultural atmosphere. But now it seems that these advantages are being struck by the new reality. When companies began to require employees to return to the office, even with a hybrid model, the importance of location was re-highlighted.
Employees living in Austin find that if their company is headquartered in San Francisco or Seattle, commuting costs and time become new challenges.
In stark contrast to the cooling in Texas, cities like Miami and San Diego are rising rapidly.
These two cities are attracting tech talent, not by huge budgets, but by sunshine, lifestyle and lower cost of living.
Miami’s combination of tax incentives and quality of life has driven a 12% increase in AI jobs.
Big tech jobs in San Diego grew by 7%, even as startups in the region lost 3.5% of their workforce in 2024, indicating an upward mobility of talent.
Interestingly, startups in San Diego County raised $5.7 billion in venture capital in 2024, marking one of the best-performing years in the region’s history.
PS: As someone who has been in SD for a few years, the environment and safety are indeed very good, especially in several cities in the north.
I noticed that despite various geographical changes, San Francisco and New York’s dominance remained solid.
More than 65% of AI engineers are still concentrated in these two metro areas.
Despite rising housing prices, shrinking salary advantages, and the flexibility of remote work, San Francisco and New York continue to attract more tech talent.
This phenomenon made me think: even in the era of remote work, the geographic concentration effect is still very strong.
The concentration of innovation clusters, network effects, and career development opportunities has kept these traditional tech hubs strongly appealing.
Behind this geographical restructuring also reflects a deeper trend: a shift from “presence” to “proximity”.
The media narrative about returning to the office is exaggerated, and companies are rethinking what really matters.
The new model is that proximity trumps existence.
For many tech companies, it’s not about clocking in five days a week, but about being close enough to have hybrid work arrangements and core schedules.
The result: In-state hiring surges as employers seek a new balance between flexibility and face-to-face time. This trend is likely to continue to shape the geographic distribution of tech talent, creating new regional tech hubs while consolidating existing strong regions.
Predicted hit rates for 2024 vs. bold predictions for 2025
Looking back at SignalFire’s predictions last year, I found them to be quite accurate in their judgments on some key trends, which made me pay more attention to their predictions for 2025.
Last year, they accurately predicted the continued evolution of the score-based work model, the continued growth in demand for cybersecurity talent, and the evolution of remote work rather than disappearing. The accuracy of these predictions gives me more confidence in their judgment of future trends.
I am particularly interested in their predicted implementation of the work of the fractional system.
C-suite positions such as CMOs, CFOs, and CTOs are increasingly working as consultants, and this trend is indeed validated in 2024.
It remains to be seen whether this pattern will continue in stronger markets, but for now it seems to be working for both companies and executives.
I think this reflects the new need for flexibility and the diversification of work styles by senior talent.
In the field of cybersecurity, their predictions have also been confirmed.
With the rise of AI-driven threats, the demand for cybersecurity talent does continue to grow.
Pay levels have risen, positions are harder to fill, and hiring urgency is higher than ever.
This trend reflects a growing awareness of security challenges and a reassessment of the value of professional security talent.
For remote work, their predictions are just as accurate.
The return to the office debate continues, but the reality is more nuanced. Companies are increasingly adopting hybrid models, while talent continues to demand flexibility.
We won’t know the long-term balance point until the next economic cycle resets the balance between supply and demand.
This trend of evolution rather than revolution reflects the gradual adaptation and compromise of enterprises and employees in the new work model.
For 2025, SignalFire has made three predictions that I think are very insightful.
The first is the rise of generalist engineers. Experts win the previous decade, generalists may win the next. As tools like Copilot, Replit, and Cursor mature, engineers no longer need deep machine learning expertise to build AI applications. Companies will prioritize flexible, collaborative generalist engineers who can move quickly and collaborate effectively with powerful tools without the need for a PhD.
The second prediction is that 2025 will be the year of equity advisors. As lean startups are cautious about hiring for both entry-level and C-level positions, founders will leverage experienced experts as equity advisors.
Carta’s data shows that these positions are more affordable than before, providing startups with a low-cost way to gain experience and mentorship without increasing their burn rates.
I think this trend reflects the importance of cost control in startups and the recognition of the value of experience.
The third prediction is the emergence of new jobs, not just those that disappear.
While headlines warn of job loss due to AI, SignalFire sees another shift: the emergence of new roles. Positions such as AI Governance Lead, AI Ethics and Privacy Specialist, Agent AI Engineer, and Non-Human Security Operations Specialist are expected to emerge. It will take time to scale, but these are some of the roles that new graduates should focus on.
I am particularly interested in this prediction because it offers a more optimistic perspective that technological advancements, while replacing some jobs, will also create new opportunities.
I believe these predictions reflect a deep understanding of technological developments and labor market dynamics.
The rise of generalist engineers reflects the trend of AI tools lowering the barrier to skill, the equity advisory model reflects pragmatic choices in economic environments, and the emergence of new jobs illustrates that innovation always creates new needs and opportunities.
For those planning their career progression, these predictions provide valuable directional guidance.
I think deeply about the future of scientific and technological talents
After analyzing this detailed report, I have some deep thoughts on the future of the talent landscape in the tech industry.
I think we are at a critical turning point that will not only reshape the career trajectory of individuals, but also redefine the direction of the entire industry.
The data from the past year clearly shows that technology alone cannot build the future, and talent is the key. Anthropic’s talent retention and reinvention of its tech center are both proof that the real advantage lies in how it recruits, develops, and retains top talent.
For new graduates, I think the reality is cruel, but also full of opportunities.
The traditional “training round” is gone, fewer entry-level positions are available, and the path ahead will rely on bootcamps, open-source projects, freelancing, and creative projects.
It’s not enough to stay on top of the latest AI tools, it’s more important to learn to fix their flaws.
Debugging chaotic machine-generated code could become a superpower for the next generation of developers. This shift requires new graduates to be more proactive, creative, and resilient.
I am particularly concerned about the impact of this change on the education system.
Traditional computer science education models may require fundamental adjustments. If businesses are no longer willing to invest in training new people, educational institutions need to take on more responsibility to ensure that graduates have ready-to-use skills.
This could mean more hands-on projects, industry collaborations, and skills-based curriculum design. At the same time, I have seen the rise of new learning models, such as online bootcamps, open-source contributions, and project-driven learning methods.
For employers, I think strategies to reduce junior hiring in the short term may pose long-term risks. AI may reduce short-term demand for junior employees, but skipping them altogether can disrupt the long-term talent pipeline.
The future of the industry depends on providing the next generation with skills that match the evolving technological landscape. Companies that can find a balance between leveraging AI to improve efficiency and investing in talent development will gain a competitive advantage in the future.
I am relatively optimistic about the long-term impact of AI on the job market.
While certain jobs will be replaced by automation, history teaches us that technological advancements often create new jobs.
The new role types predicted by SignalFire, such as AI Governance Lead and Agent AI Engineer, are just the beginning. I believe that there will be more new positions in the future that we cannot imagine now.
The key is to stay able to learn and adapt rather than sticking to outdated skills and mindsets.
From a geographical perspective, I think the change in the distribution of tech talent reflects the evolution of a healthier ecosystem.
While Silicon Valley and New York will remain dominant, more diverse tech hubs will offer more options for talent from different backgrounds and preferences.
This decentralization may reduce the cost of living, increase the inclusivity of innovation, and create a diverse tech community.
I’m particularly interested in the successful experiences of companies like Anthropic in talent retention.
In an industry with high talent mobility, being able to maintain an 80% retention rate speaks to the importance of corporate culture and values.
This reminds us that in an era of rapid technological advancement, human care and corporate culture may be more important than salary and benefits.
Companies that can create inclusive, autonomous, and meaningful work environments will have an advantage in the competition for talent.
From the perspective of the development of the entire industry, I think the current adjustment period is healthy and necessary.
Overhiring and the flood of funds during the pandemic did create some unsustainable models.
The current contraction and restructuring, while painful for individuals, is beneficial to the long-term healthy development of the industry.
It forces companies and individuals alike to think more carefully about value creation and skill development, which will lead to higher quality innovation and more sustainable growth.
Looking ahead, I believe the tech industry will continue to be an important engine of economic growth and social progress.
But the industry needs to learn to find a balance between technological progress and talent development, between efficiency improvement and social responsibility, and between global competition and local development.
Individuals and businesses that can find their place in these complex balances will be the winners of this new era. The future of technology depends not only on what technologies we develop, but also on how we cultivate and use talents to guide these technologies to serve humanity.
