IBM produces first working chips modeled on the human brain
IBM has been shipping computers for more than 65 years, and it is finally on the verge of creating a true electronic brain. — Big Blue is announcing today that it, along with four universities and the Defense Advanced Research Projects Agency …
Context & Ripple Effects
This announcement extends an arc IBM has been building for years: researchers had already packed a supercomputer onto a single chip back in 2007, and 2011 has been a showcase year for Big Blue's ambition to move past conventional computing — Watson out-argued human champions on Jeopardy! in February, and just weeks ago the company detailed phase-change memory breakthroughs it says will reach servers by 2016.
The new chips, developed with DARPA funding and four universities, mark the point where 'cognitive computing' moves from slide decks to working silicon. For IBM, riding record quarterly results ($26.7B in Q2 revenue, up 15 percent), the brain-modeled chip is both a research milestone and a statement that the next computing platform should be built around the brain's architecture rather than faster versions of today's.
First-order effects
- DARPA gets its first demonstrable return on the program: working neuromorphic prototypes validate continued federal investment in cognitive computing rather than paper studies.
- IBM now holds an early-mover position in brain-inspired silicon, complementing its software-side Watson effort and giving its research division a flagship to recruit university talent around.
Second-order effects
- The memory subsystem becomes a competitive frontier: IBM's own phase-change memory work targets exactly the dense, low-power storage that brain-style architectures need, so progress on one strengthens the other.
- Other chipmakers and defense contractors face pressure to answer DARPA's cognitive-computing push with their own programs or risk ceding the government-funded research pipeline to IBM's consortium.
Third-order effects
- If the pattern holds, mainstream computing splits into two tracks — conventional processors for precision work and neuromorphic coprocessors for perception-heavy tasks — reshaping what 'a computer' means at the architecture level.
- Sustained DARPA involvement signals that national-security demand, not consumer electronics, may be the first customer base for brain-inspired hardware, setting the funding model for the field.
The trend: Computing is beginning its first serious architectural departure from the von Neumann model since the transistor era, with government agencies like DARPA bankrolling brain-inspired silicon at IBM and its university partners.