Deirdre Bosa
@dee_bosa • 76,954 subscribers
Pronounced Dear-dra. Anchor & reporter @CNBC for 15 yrs. Now building something new at the frontier of media. Still a Leafs fan 🇨🇦
Shorts
Videos
The new Copilot is Microsoft’s bet that the AI race is moving from models to products. It doesn’t need to own the best model if Copilot can choose among them and keep the customer inside Microsoft. The potential secret weapon is Autopilot (what I’ve been calling Muse for Business). It knows your workflows, sits across the apps and data you already use, and keeps working even when you step away. It’s a preview of the battle to come in enterprise AI. OpenAI and Anthropic are racing from models into products. Can they build their own version of this? And can they match Microsoft on the boring-but-critical stuff: permissions, identity, auditability and control, as agents become more autonomous? My full conversation with Satya Nadella. We also talk open vs. closed models, US-China Summit, regulation and the infrastructure buildout. 00:00 Microsoft’s new Copilot 02:02 Why not just give us Autopilot? 05:18 Who pays for always-on AI? 08:06 Copilot picks the model 10:27 Chinese models and OpenAI’s lead 12:04 Competing with OpenAI and Anthropic 13:17 US–China AI talks 16:11 Does AI need new rules? 18:05 The data center backlash 20:40 Is AI being overbuilt? 23:44 Keeping humans in control
Deirdre Bosa344,922 Aufrufe • vor 27 Tagen
San Francisco’s once premier shopping center (Westfield mall) the Sunday before Christmas
Deirdre Bosa3,753,788 Aufrufe • vor 10 Monaten
Claude in Excel is breaking down gated analysis. Built this with just a few prompts and Factset data: base/bull/bear cases on $GOOGL search revenue vs AI infra costs would normally have to wait for this kind of analysis in a Wall St note Earnings season is going to be 🔥
Deirdre Bosa370,194 Aufrufe • vor 9 Monaten
“It is in America’s interest to actually support open source" - Aravind Srinivas when I asked him yesterday about Washington’s fear of Chinese open AI models His view: Nvidia/Nemotron can help close the gap, but the US should support open weights because affordability is what lets businesses use AI.
Deirdre Bosa92,537 Aufrufe • vor 3 Monaten
Narrowing the “cone of uncertainty" Microsoft’s custom AI chips are a hedge against uncertain demand. When forecasts change every week and data centers take 2–3 years to build..... getting more AI out of the same power and budget gives Microsoft room to adjust. "In a world where power is constrained, land is constrained, and materials are constrained... everything you can do to achieve more with less ultimately helps in the build out phase that we're in right now" - MSFT cloud chief Scott Guthrie
Deirdre Bosa10,701 Aufrufe • vor 10 Tagen
"The world needs little tech to win" Why Perplexity CEO Computer says no to big tech buyouts
Deirdre Bosa297,415 Aufrufe • vor 1 Jahr
Tomorrow: Satya Nadella On YouTube, X and Yahoo Finance. Daily show launches end of October.
Deirdre Bosa24,622 Aufrufe • vor 28 Tagen
We went from 3, maybe 4 major AI compute sellers to potentially 6 Amazon, Microsoft, Google, maybe Oracle. Now add SpaceX and Meta... just as customers are getting more efficient w tokens. "You could have a pretty decent-sized speed bump in here" - Dan Niles
Deirdre Bosa77,189 Aufrufe • vor 3 Monaten
Benchmark’s Peter Fenton says 90%+ of tokens could come from open weight models in the next 18-24 months, pressuring frontier model margins. Full convo at 12p PT/3p ET on the livestream, plus Aravind Srinivas on Perplexity’s new orchestrator model and Jeffrey Morgan on why enterprises are moving toward AI models they can download and control
Deirdre Bosa64,460 Aufrufe • vor 3 Monaten
.Lin Qiao is great. Here she is on the question every AI company has to answer: what is the ROI? if what a customer pays doesnt cover the compute they consume, growth can just deepen your losses. So you can find product market fit.... and still “scale into bankruptcy”
Deirdre Bosa42,939 Aufrufe • vor 2 Monaten
"One of the biggest misconceptions" Cerebras CFO Bob Komin pushes back on the small-models narrative. "We serve all models, and there is no limit to the size of the models that we can serve. Today, we're serving trillion parameter models. We're serving trillion parameter models that are internal for OpenAI today. We are currently running OpenAI 5.4 and 5.5 with them."
Deirdre Bosa84,608 Aufrufe • vor 5 Monaten
"SaaS was Software as a Service. I believe it's going to be service as software." General Catalyst's Madhu Namburi on the AI roll-up thesis. Services is a $20 trillion market.... multiple times the size of software. not the PE playbook of debt and cost-cutting. Venture is buying legacy companies and using AI to drive growth
Deirdre Bosa86,713 Aufrufe • vor 5 Monaten
The post-frontier era: cost, control and compute. Learned a ton in this livestream with Aravind Srinivas, Peter Fenton and Jeffrey Morgan. Big takeaway: the model race is becoming a systems race. Routing, open weight models, local compute, enterprise data, and whether the big labs’ pricing power holds once buyers have more options. Worth a full listen. Link below.
Deirdre Bosa52,989 Aufrufe • vor 3 Monaten
Is compute scarcity masking the real economics of AI? interesting point from Peter Fenton: open-weight models are still complex to run well. For now, the big labs have the bundle: model + compute + reliability + access... and thus pricing power. But if compute supply loosens and open model tooling gets easier, routine AI work can more easily move to cheaper/open models.
Deirdre Bosa50,558 Aufrufe • vor 3 Monaten
One of the biggest questions in AI right now: how much autonomy and moral standing do we give increasingly capable systems? Anthropic is more open to models developing independent judgment, even potentially deserving rights or welfare consideration. Microsoft is taking the opposite view: AI stays subordinate to humans. "Anything that doesn’t serve humans or is not in human control is not worth doing"
Deirdre Bosa11,272 Aufrufe • vor 20 Tagen