DeepSeek R1: Was It Fear or Empowerment That Drove Its Innovation?

The answer holds the secret to product team effectiveness.

Fear can spark innovation—or suffocate it. Which was it for the DeepSeek R1 team? Innovation emerged, of course, but did fear power it?

I’ve seen product teams the world over who don’t succeed as a result of fear. My profession is software product management. Too often, I see corporations attempting to build products using fear to motivate. They drive teams to better, faster, and cheaper output.

Deadlines. Budget pressure. Stakeholder-driven scope rigidity. It’s a pressure cooker in most product organizations.

Sound familiar? I bet it does.

The team behind DeepSeek had pressure, but it was of a different kind.

My typical stance on fear is to avoid it as a motivator.

The common result of the fear cycle isn’t success. It’s crippled products, demotivated teams, and unsatisfied customers. And I’m not alone in this assessment. Studies point to 80% of all product features being rarely or never used.¹ And Gallup puts employee engagement at an all-time low of 30% in 2024.²

I see this dire situation as a result of fear-driven development, not empowered teams. The typical fear-driven pursuit of output puts most teams in a box. And any innovative spark can’t start a fire in this type of suffocating environment.

But DeepSeek was cast as the underdog in the AI fight. And being in a disadvantaged position like this creates anxiety. Why did this fear not paralyze them?

DeepSeek R1 came out of nowhere, challenging my thoughts about fear and motivation.

When DeepSeek R1 landed on the scene a few weeks ago, it made me reflect on its breakthrough innovation. And how they did it.

• Were they motivated by fear? It seemed as if this was somewhat true. US GPU import sanctions and limits constrained them. This could have crippled China’s AI ambitions. Instead, it lit an unquenchable fire inside them.
• Was empowerment at play? My gut told me a China team had to have limited empowerment. It’s China, after all (the common stereotype of oppression creeped into my mind). Yet, their innovative spirit with R1 doesn’t line up with that. They had to feel empowered to achieve what they did. But was this different?
• How did they catch the big players like OpenAI flat-footed? With irony, the US used constraints to demotivate AI innovation. Corporations often use constraints to motivate behavior. But the shackles acted as a motivator instead for the DeepSeek team. Everyone wrote China off because of the constraints placed on its compute capability. Wrong move.

These are all intriguing questions for product teams everywhere. It challenges my typical stance on fear and constraints. It might be doing the same for you.

Upon reflection, I’ve found a key nuance that made fear an asset in the case of DeepSeek R1 rather than a liability. And all product teams can learn from what drove this innovative play to disrupt the AI field.

Let’s dig in.

How did the breakthrough DeepSeek innovation emerge?

I’ve long believed I know the secret to great products. It’s an empowered team, anchored by an inspiring purpose and a direct connection to its end-users. They outperform the fear-driven team.

A clear problem to solve. The will to do it. The skills to achieve it. Safety to experiment. This is how you adapt nimbly to meet what appears as an insurmountable challenge. It is the petri dish of empowerment.

Yet, empowered teams are needles in a worldwide haystack of disengaged product teams. And China is not the poster child for empowered workers. It’s an unlikely candidate to produce an innovation that shakes the market. One that outpaces even the Silicon Valley elite. Yet, against all odds, this is what they did.

Upon further reflection, the DeepSeek team had all the right ingredients.

• A tough problem to solve. Create a large language model with high reasoning capabilities, using underpowered compute servers.
• The will to solve it. They were the underdog in this AI race through compute restrictions and sanctions. They had the grit to overcome this constraint and prove their merit in the AI space.
• The skills to achieve it. They had a diverse team of experts from a broad set of specialties. AI and machine learning. Reinforcement learning. Natural language processing. Data engineering. High-performance computing. Quantitative finance. The right multi-disciplinary skillsets were all in play.
• Safety to experiment. By making their solution open source, they invited others to see their work and contribute to it. If everything is open, nobody would perceive them as a threat. This is a safe and transparent play for China. Plus, it shows off their AI chops.

So, this team beat the common perception of disempowered China workers. They had all the right ingredients. And it showed in their breakthrough R1 model.

So, does fear or freedom drive innovation?

Based on what happened with DeepSeek R1, I’ve concluded it’s a bit of both.

Constraints can lock you in and make it difficult to pivot. They can also reduce safety.

After the DeepSeek R1 entrance, I no longer think all constraints are equal. If manufactured to motivate like a whip, a constraint acts as a destroyer of innovation. But if constraints are external, they act as an igniter of motivation and passion. Just like what happened to the China team working on DeepSeek R1.

The will to solve a problem, the skill to solve it, and the safety to solve it depends on the nature of the problem. Nobody believes in a problem imposed on them by a corporate drive to whip and prod behavior. Yet, when that problem is external to the organization, it lights a fire inside a team. It rallies them. It unites them.

A fire raging inside drives curious exploration. You uncover ingenious workarounds to the limitations you face. Just look at what the DeepSeek R1 team did to outperform AI giants like OpenAI and Google:

• Using only 2,000 GPUs → Open AI needed 100,000+
• Training their model for $5M → Not the typical $100M+
• Maximizing token efficiency → By analyzing phrases, not words
• Requiring 75% less memory → Can runs on a standard gaming GPU
• Deploying expert models only when needed → Dramatic runtime cost savings

Some of DeepSeek’s claims may be somewhat skewed to appear within their sanctions. But it’s an impressive feat nonetheless.

The innovation coming out of DeepSeek could only emerge from a place of empowerment. In this case, external fear and constraints fueled it. I now see fear in a different light.

External fear can be good (internal, imposed fear will always be bad).

What can product teams learn from the breakthrough release of DeepSeek R1?

I have derived some truths from the DeepSeek R1 team’s innovation:

• Fake constraints kill innovation. Real constraints fuel it. DeepSeek R1 proved it.
• Fear can ignite or suffocate creativity. The key? It should be an external force, not an internal one.
• Innovation thrives under external pressure. It dies under internal control.

Below are three key lessons for product teams.

1: Unlimited resources aren’t necessary for innovation. Quite the opposite.

Many ambitious corporate product efforts get mired in aspects of scale.

• Bigger teams.
• Larger budgets.
• Higher compute.
• Cutting-edge technology.

But DeepSeek actually proved how you can do more with less. A fraction of the compute capability. 200 people versus 1000s. Subpar GPUs. Less memory. The list goes on.

Scale can hide the simple solution. Resourcefulness and a scarcity mindset can encourage creative thinking. So, start small and focus on the essential solution that gets a large chunk of the value. It may not capture all the possible value, but optimizing to do that may cost much more than it’s worth.

DeepSeek found the magic 20% of the effort to capture 80% of the value.

2: External constraints are the powerful driver, not internal, imposed constraints.

Turns out, external constraints and the resulting anxiety work as compelling motivators.

External factors forced DeepSeek to be the underdog. Imagine if China had forced the team to create DeepSeek with similar restrictions. Would they have innovated like they did? No, they wouldn’t have. I’ve seen the psychology of this more times than I can count. The team would have seen this as an act of command and control, a lack of support. It would have created an uninspiring vision and demotivated the team.

Yet, the imposed external constraints generated a kind of positive fear. DeepSeek had a mission they believed in. And they had a point to prove. The mission was theirs. They owned it. And they delivered.

It’s time for corporations to stop manufacturing fake constraints. External factors should drive team passion for success. This could be a tough customer problem to solve, a savvy competitor to beat, or a market opportunity to seize fast.

Let the thrill of the pursuit set the tone.

Now, we get to the final lesson: empower your teams to get there.

3: Empowerment remains the king of innovative results. It needs the right fuel.

Problem to solve. External constraints known. Point to prove. Check. Check. Check.

These are ingredients to ignite the environment for an empowered team to flourish. It will light their internal fire.

This raging flame will drive their will to solve the problem. It will demand the right skills to solve it to be on the team. It will rally leadership to remove any obstacles (not put them up with fake constraints).

The DeepSeek team had a point to make. And it made it. It had empowerment. Does your team?

---

And that sums up the product team lessons I took away from DeepSeek’s R1.

1. Be crafty and start small with resourcefulness.
2. Use external motivators, not fake internal imposed constraints
3. Ignite a purposeful fire to drive your teams to carve their path to success.

Fear has nuance. It can destroy, but it can also motivate. External constraints can give us a purpose, a stand to take, a platform to make a point, a reputation to prove. Fake constraints imposed on a team only destroy effectiveness. It’s time we stop those. Stamping a mark on the world is a superior motivating force.

• External constraints → Motivation & ingenuity
• Internal constraints → Fear & stagnation

Fake constraints kill teams. Real constraints fuel them. DeepSeek R1 proved it.

What will you do to remove internal fear controls on your product teams? How will you inspire them with external motivators? It’s time we all motivate in better ways. It worked for DeepSeek. And it’ll work for us.

How are you using real constraints to fuel innovation (not fake ones that kill it)? Let’s swap insights—comment below.

---

➡️ Sign up for weekly insights like this. Learn about returning to the fundamentals that make product teams unstoppable. Join a community of like-minded professionals committed to achieving outcomes sooner.

---

References

1. Thomas, S. (2019). 2019 Feature Adoption Report. Pendo.io. Retrieved from https://go.pendo.io/rs/185-LQW-370/images/2019%20Feature%20Adoption%20Report%20Digital.pdf
2. Jim Harter, “U.S. Engagement Hits 11-Year Low,” Gallup, 10 Apr. 2024, https://www.gallup.com/workplace/643286/engagement-hits-11-year-low.aspx.