
Thinker, strategist, enterprise architect, keynote speaker, analyst, book author, futurist on IT, CIO, cloud, AI. @SDABocconi Heads research at @Frontier_Now
Washington, D.C.
Joined July 2006
- Tweets75.4K
- Following7.2K
- Followers55.8K
- Likes69.9K
Pinned Tweet
New Research: The CIO in 2030: Running the Enterprise Intelligence Economy
thisisfrontier.ai/research/c…
Fourteen operating imperatives for a strategic world where people, agents, software, compute, data, and trust are dynamically allocated to outcomes. #CIOChat
Dion Hinchcliffe retweeted
The engineering and process question is whether we can make intelligence increasingly autonomous without making authority increasingly autonomous.
1/ New Post: AI Is Starting to Build Its Successor. Who Gets to Say Wait?
thisisfrontier.ai/blog/ai-is…
The pace of AI is increasing quickly. This likely won’t be curtailed. So humans must *always* have oversight.
@diegocaleiro (cited) @sarbjeetjohal @Scobleizer @vijayasankarv
An enormous leap in not just mathematics but in what AI-led research can do.
Moon shot research is now possible using the latest advanced models.
If this holds, we’ll soon see breakthroughs in science, technology, and medicine that were unimaginable even earlier this year.
Oh my god: Anthropic mathematician Levent Alpöge (@__alpoge__) says he has proven the Riemann Hypothesis, open for 167 years.
"I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis."
He says the method also covers the Generalized Riemann Hypothesis. The proof is out as Lean code (a proof language where a computer checks every step), so experts should be able to check it quickly, Scientific American reports.
Same Alpöge who was behind this year's Jacobian conjecture result and the Navier-Stokes work with Tristan Buckmaster.
I assume we will get an official statement very soon from him!!
The new Amazon Web Services (@AWS) Transform unifies migration assessment and execution
thisisfrontier.ai/news/aws-t…
My take: CIOs have often had to govern migration economics and migration execution as related but artificially separate work products. No longer.
Cloud repatriation is on the rise — but tricky questions remain
cio.com/article/4245967/clou…
Very interestingly for CIOs, it’s AI workloads that most seem headed back on-prem.
1/ New Post: AI Is Starting to Build Its Successor. Who Gets to Say Wait?
thisisfrontier.ai/blog/ai-is…
The pace of AI is increasing quickly. This likely won’t be curtailed. So humans must *always* have oversight.
@diegocaleiro (cited) @sarbjeetjohal @Scobleizer @vijayasankarv
2/ Our new companion research report explores these issues:
The fierce pace of AI development puts a demand on leaders to make it as safe as possible.
How CIOs and CISOs can verify what enters the enterprise, contain what agents can do, and keep critical work running while seizing the strategic advantages that AI confers:
thisisfrontier.ai/research/a…
1
72
What is beyond AGI? This is the exponential mark that business leaders now have to somehow hit.
Because understanding what is possible in 3-5 years is vital in order to be ready to wield the possibilities for your stakeholders and customers.
That’s the question that stayed with me from @GoogleDeepMind researcher Tim Genewein’s new essay on what comes after AGI.
The numbers behind the argument:
• The world’s effective AI computing capacity is increasing at least 10× a year.
• AI is getting about 13× cheaper each year.
The price of world-class expertise of any kind will fall to near zero, when you can train a top specialist once and copy it a million times, experience and all.
What will come next:
• A task that costs $1 billion in AI compute today will cost about $1,000 in ten years if hardware and software efficiency keep improving at roughly 4× a year.
• The world will have the capacity to run over 1,000× as many of today’s agents by 2030.
• Over a billion times as many agents within a decade.
• Four routes to superintelligence will reinforce each other.
- Vastly scale up models and data.
- Invent much better ways for AI to learn.
- Organize billions of agents into breakthrough teams.
- Let AI improve its own successors exponentially.
Genewein’s yardstick for superintelligence is an entire expert community. Imagine compressing decades of a scientific field’s progress into months (or days, or hours), across fields as different as medicine, materials and clean energy. This is what is coming:
institute.deepmind.com/essay…
1
249
Speaking of AI governance, organizations today must rapidly get their arms entirely around their AI + agentic development and operations, with policy and guardrails.
Our new AI governance guide breaks down the leading enterprise options:
thisisfrontier.ai/research/y…
We’ll keep it updated, since the space evolve extremely fast!
More signs that recursive self-improvement (RSI), the Holy Grail of rapid AI development, is nearly here.
Many are cautious about this, worried it will outstrip our ability to manage and control.
I’ll more to say about this soon. In any case, a very important development.
RSI breakthrough from Google
Google DeepMind Made Recursive AI Self Improvement More Powerful
The system uses Self Improvement Memory (SIM) to remember previous optimizations, why changes were made, and how well they performed. It then uses this knowledge to continuously improve AI agent systems across different tasks, datasets, and AI models.
In one experiment, performance increased from 68.41 to 84.75, compared with just 73.42 without memory.
Across ten independent coding experiments, SIM also delivered statistically significant improvements.
With the release of the head-turning Gemini 4 Neon model, it’s worth looking back at Google’s @SundarPichai summarizing their ten years of being “an AI-first” company:
- Now processing quadrillions of tokens monthly
- 5 products with over 3 billion users
- Gemini usage at ~1 billion MAU
- Nearly 10 million developers
A key conclusion: The sleeper has awakened.
Now that I finished surveying general AI governance tools (for now), I’ve started examining agentic governance tools. I believe traceability is going to be critical to that process.
Example: @BlueFlag_Sec’s new capabilities are interesting because a developer’s name can appear on a code approval even when an AI agent used their credentials. The developer never even saw the change.
BlueFlag can flag a pull request an agent approved, trace it through the build, and require a human review before the code actually merges. This approval gating will become vital IMO. In 22% of the organizations BlueFlag studied, an autonomous agent had already merged code without any human in the loop.
Our latest analysis:
thisisfrontier.ai/news/bluef…
8
6
384
Google just selected @BlueOrigin to launch its solar-powered TPU satellites for Project Suncatcher. Slated for Q1 2029 on a New Glenn launch vehicle.
Each satellite launched in this mission (plural was used, so it’s more than one) can carry about one data center rack of compute.
In contrast, @SpaceX expects to already have a significant amount of commercial AI compute in orbit by then.
1
3
582
FAA ground stop offers IT leaders stark modernization lessons
cio.com/article/4250479/faa-…
Namely, that updating a critical IT or telecom service without a rapid fallback or backup system in place is a very risky business.
4
5
296
Dion Hinchcliffe retweeted
Replying to @dhinchcliffe
#CIOChat Q4: Human exception team. By the way I would invert the titles on the image.
Dion Hinchcliffe retweeted
Replying to @CPetersen_CS
Yes, I get carried away myself. CIO topics are my favorite.
We’ll reconvene #CIOChat on December 3rd, 2026!
Dion Hinchcliffe retweeted
Replying to @waynesadin
This is a good point, Wayne. Thank you for observing.
So, we’ll push to the week after!
#CIOChat
Dion Hinchcliffe retweeted
Replying to @dhinchcliffe
#CIOChat A3: Not if the service desk is within the department, and owned by one of its staff. The idea of a unified help desk is the past revisited.
Dion Hinchcliffe retweeted
Replying to @dhinchcliffe @joannefriedman
It's an engineering problem, which means time & money WILL fix it
#CIOchat
#CIOChat Great chat today! Join us at 2pm ET next Thursday. Topic: The Agent Kill Switch
Hope you enjoy our movie tie-ins to intro next week’s topic.
Please suggest topics and invite your IT colleagues. Every Thursday at 2pm ET.
🤖 Made with AI











